The Web That Never Was - Dylan Beattie

The Web That Never Was - Dylan Beattie — NDC Conferences

NDC Conferences

AI-assisted summary of The Web That Never Was - Dylan Beattie — original source. Check the original for full context.

Source published:

Key ideas

  • Technological dominance in software is rarely determined by Darwinian survival of the fittest; everyday tools and standards often succeed because they were merely 'good enough' to overcome analysis paralysis and historical accidents. (5:16)
  • Modern computing and the web hinge on narrow historical forks, such as IBM's fateful 1981 operating system deal and the hasty 10-day creation of JavaScript under corporate deadlines. (54:09)
  • Dylan Beattie constructs an alternate history—featuring the GRID protocol, HyperLisp, and Gary Kildall's dominance—to illustrate how arbitrary choices shape our digital landscape. (18:40)

Detailed notes

## Background Published by NDC Conferences on YouTube on November 13, 2018, this keynote presentation features software developer Dylan Beattie. According to the page description, the talk investigates an alternate history of the World Wide Web, exploring what the software industry might look like without Microsoft, Windows, Google, or JavaScript if past compromises and corporate meetings had unfolded differently.

## TL;DR Dylan Beattie weaves an elaborate counterfactual history of computing where Gary Kildall’s Digital Research dominates the PC market over Microsoft, Tim Berners-Lee establishes 'the GRID' using SQL verbs and tilde-based markup instead of the World Wide Web, and HyperLisp replaces JavaScript. In this timeline, Marc Andreessen and John Carmack team up to build networking for Nintendo, PerfectDigital launches GEM 95, and Steve Jobs unveils an ultra-thin anodized NeXT laptop powered by a search engine called Goggle. Beattie then dissects the real-world near-misses behind these scenarios—from Brendan Eich's original intent to embed Scheme into Netscape to Tim Berners-Lee's regret over the double-slash in URLs. The presentation concludes that modern technology is defined not by Darwinian perfection, but by pragmatic, 'good enough' compromises that helped engineers move forward.

## Atmosphere & style The presentation opens with an engaging acoustic parody of Baz Luhrmann's 'Wear Sunscreen' before settling into a warm, comedic blend of speculative fiction and deep computer history. Beattie commands the stage with dry wit, theatrical pacing, and palpable affection for obsolete hardware and retro software trivia. Watching the talk feels like sitting through a clever, geeky stand-up routine that successfully morphs into a thoughtful meditation on engineering pragmatism and the arbitrariness of technological fate.

## Bulletin - [05:16](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=316s) Technological dominance in software is rarely determined by Darwinian survival of the fittest; everyday tools and standards often succeed because they were merely 'good enough' to overcome analysis paralysis and historical accidents. - [54:09](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3249s) Modern computing and the web hinge on narrow historical forks, such as IBM's fateful 1981 operating system deal and the hasty 10-day creation of JavaScript under corporate deadlines. - [18:40](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1120s) Dylan Beattie constructs an alternate history—featuring the GRID protocol, HyperLisp, and Gary Kildall's dominance—to illustrate how arbitrary choices shape our digital landscape.

## Key takeaways - [05:16](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=316s) Technological adoption is driven by pragmatic compromises, technical debt, and arbitrary timing rather than objective technical superiority. - [05:16](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=316s) Douglas Adams' three rules of technology illustrate that what exists at birth feels natural, what appears between ages 15 and 35 enables careers, and anything after age 35 violates the natural order. - [08:44](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=524s) IBM's inability to move quickly due to corporate bureaucracy led to an isolated skunkworks project that built the original PC using off-the-shelf parts. - [12:01](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=721s) Before Paul Mockapetris independently designed DNS, all internet host mapping was manually managed in a centralized HOSTS.TXT file at the Stanford Network Information Center. - [15:26](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=926s) Competing hypermedia platforms like the University of Southampton's Microcosm stored links in external databases rather than embedding them directly in documents. - [18:40](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1120s) In an alternate timeline, Tim Berners-Lee designs 'the GRID' using SQL-style HTTP verbs to preserve neat, fixed-width server logs. - [24:04](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1444s) The alternate timeline eliminates angle brackets and URL double slashes in favor of tildes (~) and exclamation marks (!) based on ASCII character frequencies. - [27:36](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1656s) Digital Research's Gary Kildall could have captured the PC operating system monopoly by recognizing the value of licensing CP/M-86 to third-party clone makers. - [31:43](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1903s) A redirected career path where Marc Andreessen and Jim Clark join Nintendo to develop console networking leaves browser innovation open to competitors. - [35:46](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=2146s) Instead of JavaScript, an alternate web standard embeds HyperLisp, leveraging prefix syntax, first-class lambdas, and continuation-passing style for asynchronous client-server interactions. - [57:04](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3424s) Real-world tech history shows that Brendan Eich was originally hired to put Scheme into Netscape before rushed commercial deadlines mandated curly-brace JavaScript. - [54:09](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3249s) Tim Berners-Lee has openly acknowledged that the double slash in URL protocols was completely redundant and could easily have been omitted. - [57:04](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3424s) Major software giants exist due to near-misses, including Excite turning down an $800,000 acquisition offer for Google in 1999 and Apple almost buying BeOS instead of NeXT. - [01:01:05](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3665s) Progress in the software industry comes from taking pragmatic, 'good enough' steps forward rather than stalling under analysis paralysis.

## Detailed walkthrough ### [00:06](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=6s) Advice for the Class of '17: The "Wear Sunscreen" Parody

Dylan Beattie opens his keynote with a musical parody of Baz Luhrmann's 1997 spoken-word hit "Everybody's Free (To Wear Sunscreen)", offering humorous, hard-won software engineering wisdom to the "class of '17." Subverting the original song's advice to wear sunscreen, Dylan insists that using flat screens is the single most vital tip for the future: “The benefits of flat screens have been proved in Hollywood whereas the rest of my advice has no basis more reliable than my own meandering experience.” He tells developers to revel in the fleeting optimism of greenfield projects, warning that they will never truly appreciate that confidence until things inevitably break, only to look back after shipping and realize how surprisingly productive they were. He takes aim at common architectural anxieties, pointing out that worrying about changing database vendors is largely pointless: “Don't worry about changing database providers or worry, but know that every company who used an ORM in case they needed to switch databases never actually did it.” Instead, the genuine dangers are unpredictable external dependencies—like an engineer abruptly resigning at 4:00 p.m. on a quiet Tuesday.

Interspersing his monologue with rhythmic imperatives—*optimize*, *rebase*, *refactor*, *re-read*—Dylan touches on modern software culture, ego, and tooling. He warns against obsessing over frameworks, emphasizing that end users only care about the final experience, not the tools used to construct it. He urges developers not to feel insecure if they do not understand functional languages like F#, noting that he has worked with stellar junior developers and veteran system architects who do not know it either. Above all, he highlights mental health and sustainability: taking breaks and guarding against burnout, because the human brain is the most powerful computer any developer will ever possess. Developers must accept inevitable truths: code will have bugs, deadlines will be missed, and anyone might one day end up in management, where they will look back through rose-tinted glasses and imagine that their past code was flawless and developers diligently tuned their database indexes.

- **On code maintenance:** Do not reformat someone else's codebase, and do not tolerate people arbitrarily reformatting yours. - **On hardware:** Hold onto old hard drives, but throw away obsolete network cards. - **On community and media:** Remember positive retweets, ignore online flame bait, and never read YouTube comments because they will only induce rage. - **On engineering practices:** Cache your package dependencies against sudden disappearance; inspect log files as the primary early-warning system for failures; and never deploy to production without automated integration tests or QA. - **On programming languages:** Write code in x86 Assembler once until before it makes you smug, and write Visual Basic once before it makes you stupid. - **On open source:** Sharing code is noble, but it requires far more effort than simply slapping a `README` onto a stray project folder and uploading it to GitHub in the hope that strangers will resolve your bugs.

---

[05:16](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=316s) Douglas Adams and How We Experience Technology

Transitioning to the keynote address, Dylan greets the Sydney audience, asks if they are enjoying the conference and attending PubConf, and introduces a framework for understanding human perceptions of innovation borrowed from author Douglas Adams. Adams proposed a set of rules describing how people process technological change throughout their lives. Dylan summarizes Adams' rule: “Anything that is in the world when you're born is normal, it's just a natural part of the way the world works; anything invented between when you're 15 and 35 is new and revolutionary and you can probably get a career in it; anything invented after you're 35 is against the natural order of things.” Dylan maps his own biography onto this curve: born in 1978—just too late to see the original *Star Wars* in theaters—he entered a world where color television, mainframe computers, and the ARPANET were already established realities.

The World Wide Web arrived right around when Dylan turned 15, allowing him to forge a career as a webmaster learning HTML. Conversely, items that arrived after he crossed the age of 35—such as the Apple Watch, fidget spinners, and the dance move known as "dabbing"—instinctively feel unnatural and alien to him. Dylan uses this to emphasize that human reactions to technology are deeply subjective and tied to the exact moment an individual first encounters an invention. Quoting Adams again, he reminds the audience that technology is simply "stuff that doesn't work yet"; once it works reliably, it fades into the background as ordinary "stuff," much like chairs, which no longer fail and are therefore never considered technology.

Dylan cautions against the comforting myth that technological progress follows an orderly Darwinian path:

```mermaid flowchart LR accTitle: The Myth of Technological Evolution accDescr: A flow showing prototypes evolving into production systems through supposed natural selection versus reality. subgraph Myth["The Darwinian Myth"] A["Prototypes & Ideas"] --> B["Survival of the Fittest"] B --> C["Best Solution Wins"] end subgraph Reality["The Historical Reality"] D["Gaffer Tape & Tech Debt"] --> E["Dumb Luck & Market Flukes"] E --> F["Entrenched Standards"] end ```

Every piece of ubiquitous equipment today originated as fragile prototypes, hacks, and technical debt held together with gaffer tape, code, and string. He cautions that it is dangerous to assume the best technical solution always triumphs: “It's easy to think that there is some sort of Darwinian survival of the fittest that we have all these prototypes and all these ideas and all these solutions to problems like smartphones and digital communication and the best solution is going to win, but that is absolutely not true.” Instead, history turns on mundane accidents, compromises, and arbitrary choices.

---

[08:44](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=524s) The IBM PC and the Roots of the Personal Computer Era

To demonstrate how arbitrary decisions shaped modern computing, Dylan takes the audience back to the late 1970s. In the beginning, IBM reigned supreme over the computing landscape with the mainframe. Dylan references the apocryphal remark attributed to Thomas J. Watson—the IBM chairman after whom the Watson AI platform is named—speculating that the entire world market might ultimately have room for only about five computers. Over the post-war decades, computing transitioned from vacuum tubes (valves) to discrete transistors and integrated circuits, and software moved from direct machine code to assembly and high-level interpreted or compiled languages. Gradually, computers escaped universities and research labs to inch toward the home.

By the late 1970s, IBM commanded an absolute corporate monopoly over information technology. Dylan points out that while modern giants like Alphabet, Apple, and Amazon possess immense sway, competitors still successfully sell alternative laptops and phones; in contrast, IBM virtually owned the entire sector. Recognizing that personal computers were no longer a fringe hobbyist market and that Watson’s five-computer limit was completely wrong, IBM resolved to create a computer that an ordinary consumer could fit into the back of a car and carry home. However, IBM’s colossal bureaucracy was notoriously slow:

> “IBM in the '70s was a juggernaut: you couldn't steer it. It was this massive enterprise that was incapable of doing anything quickly because it would take six weeks just to approve the font for the purchase orders.”

Terrified of being left behind by nimbler rivals, IBM formed an isolated skunkworks team to construct the IBM Personal Computer (the PC). They made the radical choice to abandon their traditional vertical manufacturing and instead assemble the machine entirely from commercial off-the-shelf components sourced worldwide: motherboards, Intel processors, and serial buses from manufacturing hubs in Taiwan, China, and Singapore. The only critical missing ingredient was an operating system, sending IBM shopping for the premier operating system company of 1981. Meanwhile, parallel to IBM's consumer hardware push, the military's Advanced Research Projects Agency (ARPA) was pioneering resilient computer networks designed to withstand Cold War attacks, setting the stage for the protocols that would eventually underpin the global network.

[12:01](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=721s) From ARPANET Maps to TCP/IP and the Scaling Limits of the HOSTS File

Dylan Beattie shows an ARPANET map from 1973, pointing out that before it became the internet, every single device on the network and what software it was running was drawn by hand by a draftsperson with pen and paper. Around 1981, the specification for IPv4 was published, and on January 1st, 1983—a date known as “Flag Day”—a strict rule went into effect: “that was the day they said from now on everything connected to our network must run TCP/IP, that's it.” In 1982, SMTP (Simple Mail Transfer Protocol) was ratified, finally providing both a system for computers worldwide to communicate and a way for human operators to exchange email.

As the network expanded, it ran into one of the fundamental hurdles of computer science: naming things. Before DNS was created, adding a new server required plugging it in and literally calling Elizabeth "Jake" Feinler at the Network Information Center (NIC) at Stanford: “Jake, I got a new computer, can you add it to the hosts file please?” Organizations kept track of connected systems through that single `HOSTS.TXT` file, which was copied via overnight FTP during cheaper phone rates, ferried back on tape, or picked up in person by someone visiting Stanford for a meeting. The operational limitations were severe:

> “If you wanted to add a machine to the hosts file on a public holiday you couldn't, because the people who had permission to edit the file were all off celebrating.”

Machines could not be added at night, over the weekend, at Thanksgiving, or at Christmas. While manageable when the network was small, the approach could not survive the growth of the internet.

[14:14](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=854s) Paul Mockapetris and the Invention of Scalable DNS

To replace the manual hosts file, the University of Southern California tasked Paul Mockapetris with evaluating five competing engineering proposals to find a compromise that kept everyone happy. Beattie quips about USC's storied past—noting it also gave the world Unix and LSD—before praising Mockapetris's actual decision:

> “He looked at all five of these solutions, went 'these are all crap,' and threw them away, and he designed and invented by himself DNS.”

Mockapetris created a hierarchical database of hostnames and network addresses engineered to scale naturally with network complexity. The design ensured that as more hosts joined, the proliferation of DNS servers distributed the load, cached responses locally, and reduced root-level lookups. Because of this self-scaling architecture, DNS has survived virtually unchanged from its original specification, continuing to power hundreds of millions of domain names today.

```mermaid flowchart TD accTitle: Evolution of Internet Host Naming accDescr: Diagram contrasting the centralized HOSTS.TXT system at Stanford with the decentralized, hierarchical DNS architecture.

subgraph Centralized["Pre-DNS Architecture"] NIC["Stanford NIC<br/>Central HOSTS.TXT File<br/>(Jake Feinler)"] ClientA["Host A"] -->|"Phone call / Manual FTP"| NIC ClientB["Host B"] -->|"Tape copy / Nightly transfer"| NIC end

subgraph Distributed["Mockapetris DNS Architecture"] Root["Root Nameservers"] TLD["TLD Nameservers (.com,.org)"] Auth["Authoritative Nameservers"] Local["Local Caching Resolvers"]

Root --> TLD TLD --> Auth Auth -.->|"Cached records"| Local end ```

[15:26](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=926s) Tim Berners-Lee, "Vague But Exciting," and Alternative Hypermedia

By the late 1980s, the necessary foundation was laid: commodity hardware was cheap, and TCP/IP, DNS, and SMTP were working standards. At CERN on the French-Swiss border, Tim Berners-Lee—who still proudly uses the official job title “Web Developer”—observed that his mathematician and physicist colleagues lacked a digital medium to publish and cross-reference academic research. He drafted a proposal titled *Information Management* and submitted it to his supervisor, who returned it with the famous note: “vague but exciting.”

Beattie emphasizes that the web's eventual dominance was not inevitable. While studying at the University of Southampton in 1997, two of his professors recalled presenting their own competing hypermedia platform, Microcosm, at an international conference just before Berners-Lee took the stage to pitch the World Wide Web. Microcosm ran on Windows 3.1 and operated on a fundamentally different philosophy: it rejected embedded hyperlinks so users would not have to rewrite existing files. Instead, it ran an independent link database alongside untouched text, RTF, or WordPerfect documents, presenting hypermedia links in a companion window curated by information analysts and librarians. While the web overtook it for public browsing, that core link-indexing technology survived into a commercial incarnation called Multicosm PLC, serving large-scale legal discovery for subpoenaed corporate servers.

[18:40](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1120s) Alternate Timeline: The Grid and SQL-Style HTTP Verbs

Transitioning into an alternate-history narrative with an homage to *The Twilight Zone*, Beattie reimagines early 1990s CERN. In this timeline, Tim Berners-Lee is experimenting with an information system he originally called "the Mesh." While contemplating the name "World Wide Web," the network goes down, forcing him to describe the concept over the telephone:

> “He picks the phone up and he goes, 'Yeah, it's this WWW,' and they say, 'Hang on a second, that acronym is actually longer than the thing it's short for; if this catches on it might sound a little bit daft.'”

Conceding the point over afternoon tea, Berners-Lee solicits alternatives from colleagues. Suggestions include: - The "International Information Interchange" (the `iii`), quickly mocked because “it sounds like pirates.” - The "Multinational Metadata Matrix," suggested by an engineer who had just finished reading William Gibson's *Neuromancer*. - The winning compromise: the "Global Repository of Interlinked Documents"—**the GRID**.

Needing a transfer protocol, Berners-Lee coins Hypertext Transfer Protocol (HTTP) but borrows its operational verbs directly from SQL rather than the familiar methods of our timeline. To keep fixed-width server log files neatly aligned, every verb is designed with exactly six letters: - `SELECT`: fetch a document from a server. - `CREATE`: publish a new document onto the Grid. - `UPDATE`: modify an existing document with corrections. - `DELETE`: remove a document when legal demands require it. - `SUBMIT`: send instructions or parameters to a remote system to execute tasks.

[21:57](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1317s) Designing the Grid: From SURF to the Bang Syntax

In this alternate timeline, Tim Berners-Lee works on standardizing how documents are addressed across a distributed network. He develops the Standard Unique Resource Finder (SURF), musing that the phrase might actually catch on when people talk about "surfing the grid." His initial concept includes the entire URL structure we know today: `http://user:pass@grid.domain.com:port/path?query#fragment`. However, a colleague immediately challenges him on including authentication details right in the visible address bar: “Why are you putting username and password credentials in plain text on something that's going to be part of the user interface?” Tim realizes the obvious hazard and immediately agrees: “Good point. We shouldn't support that; that would be stupid.”

With credentials removed, Tim continues streamlining the address format. Queries and fragments are made optional, relegated purely to search and indexing tasks, and default documents are introduced so users can simply address `grid.domain.com`. Looking critically at the remaining elements, he questions the UNIX file system inheritance of the double slash: “What is the slash-slash thing? That is... you know, he got it from UNIX file systems, but this isn't a file system; this is a completely new thing. Let's get rid of it.” When someone raises a technical concern that a numeric host name could collide with port numbers, he replaces the colon-slash notation entirely with an exclamation point delimiter: `http!grid.domain.com`. He then introduces an elegant shortcut: if a host shares its name with the protocol in use, you can drop the protocol segment altogether. The final, fully qualified address for a default server document becomes simply `grid!domain.com`. When a team member asks, “Tim, don't you think grid!cern.ch sounds a bit silly?”, Tim retorts with devastating common sense: “Not as silly as http://www.cern.ch.”

```mermaid flowchart TD accTitle: Alternate Address Evolution accDescr: The step-by-step simplification of Tim Berners-Lee's grid resource addresses.

subgraph InitialDraft ["1. Verbose Draft"] A["http://user:pass@grid.domain.com:port/path?query#fragment"] end

subgraph Cleanups ["2. Trimming & Delimiters"] B["Drop plain-text credentials"] C["Make queries and fragments optional"] D["Remove UNIX slash-slash (//)"] E["Switch to bang (!) to avoid numeric host/port collisions: http!grid.domain.com"] end

subgraph FinalAddress ["3. The Default Convention"] F["grid!domain.com"] end

A --> B --> C --> D --> E --> F ```

[24:04](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1444s) HTML Without Angle Brackets: The Tilde Revolution

Next, Tim tackles the presentation layer for hypermedia documents. Turning to Standard Generalized Markup Language (SGML)—used since the early 1980s in industrial typesetting—he drafts HTML version 0.1. However, CERN is a high-energy physics laboratory filled with mathematicians and physicists whose scientific texts are packed with mathematical inequality symbols. They push back fiercely: “Please don't use angle brackets. Like, literally every document we've got will break the second we try and publish it on your grid system, because we use those angle brackets for something else.”

Taking their complaint seriously, Tim runs a frequency analysis across all plain text files across CERN's FTP servers and network shares to discover the least frequently used character in basic ASCII. The winner is ASCII character 126 (hex `0x7E`): the tilde (`~`). Tim immediately adopts the tilde as his language's fundamental markup signifier. He sets an ingenious baseline rule: a double line break represents a paragraph break. Because existing plain-text files already use double carriage returns and avoid tildes, vast archives of scientific writing become instantly valid Grid documents with zero manual editing. By February 1992, he releases HTML 0.2, introducing structured metadata blocks, a clean `:key value` syntax, headings, paragraphs, image embedding, and a tag called `link`. When an associate asks why he doesn't call it an "anchor," Tim rejects the idea outright: “What is anchor? Why would you call it that? That makes no sense.”

- **The Tilde (`~`)**: Selected via ASCII frequency analysis (code 126 / `0x7E`) as CERN's least-used character to prevent collisions with mathematical formulas. - **Double Line Break**: Preserved as native paragraph breaks so legacy text files immediately render as valid documents. - **HTML 0.2 Features**: Explicit document metadata for machine indexing, visible body text for users, clean attribute syntax, native image inclusion, and the self-explanatory `link` tag instead of the cryptic `<a>`.

[26:34](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1594s) Grid Fever and NCSA Mosaic

With the specification published online in February 1992, widespread enthusiasm takes hold across the tech landscape. Dylan describes the onset of global "grid fever," where commercial demand explodes overnight. Individuals take weekend crash courses, print fresh business cards claiming to be "grid masters," and launch lucrative consultancies on the premise that clients have no idea what they are doing either—an environment Dylan jokes is precisely how he landed his own first job in technology.

The technology hits an inflection point with the release of NCSA Mosaic by the National Center for Supercomputing Applications at the University of Illinois. Mosaic serves as an open-source, graphical multi-protocol client: - Displays full inline graphics and hypermedia documents over the new Grid protocol. - Integrates legacy protocols seamlessly, allowing users to browse Usenet via NNTP, navigate Gopher menus, and transfer files over FTP. - Unifies disparate internet tools into a single visual window, making the wider computing world sit up and take notice.

[27:36](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1656s) Gary Kildall's Masterstroke: The Triumph of Digital Research

Dylan circles back to the pivotal turning point in personal computing: IBM's search in the early 1980s for an operating system for their upcoming PC. Instead of turning to Microsoft, IBM approaches the premier microcomputer system architects of the era: Gary Kildall and Digital Research. Kildall grasps a fundamental reality that completely eludes the executives at Big Blue: “What Gary Kildall has realized and IBM haven't is that if your flagship computer is made out of off-the-shelf parts that you bought from China and Singapore and Taiwan, you don't actually own anything.” Kildall understands that when hardware is commoditized, the operating system holds all the true platform value.

Kildall strikes a brilliant non-exclusive licensing bargain: he sells IBM a license to distribute CP/M-86 at a 50% discount in exchange for retaining the unrestricted right to sell the operating system to other vendors. IBM launches the PC with CP/M-86, while Digital Research continues rapid development. When IBM releases the 16-bit PC/AT in 1984, Digital Research is ready with MP/M-86 (Multi-Programming Monitor for Microcomputers). It delivers true protected-mode multitasking on the Intel 80286 processor, cleanly sidestepping the infamous 640 KB conventional memory wall. As low-cost beige PC clone manufacturers flood the global market, they all license Digital Research's software. By the time IBM realizes they gave away the farm, Digital Research is exploding into the largest technology titan on Earth.

[29:47](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1787s) Steve Jobs, NeXT, and the Rise of PerfectDigital

Meanwhile, in California, Apple faces fierce corporate turmoil. Following a hostile board takeover in 1987, shareholders oust Steve Jobs from the company he co-founded. Dylan recounts how Jobs, determined to prove them wrong, sets up shop right across the street and founds NeXT. He turns his back on common beige plastic aesthetics and introduces the NeXTstation: a striking eight-inch matte-black magnesium cube costing $10,000—vastly outpricing commodity $800 IBM clones. The machine is aimed squarely at high-end cinema production, scientific research, and advanced software engineering. Tim Berners-Lee procures one of these NeXT computers at CERN, using its advanced development environment to build and run the world’s very first grid browser (appropriately titled `GridBrowser.app`).

At the same time, Digital Research executes an industry-shaking consolidation by merging with the world's second-largest software outfit, the WordPerfect Corporation. Dylan notes that while the US Department of Justice spends 18 months fretting over the monopoly—crying, “Something must be done! We can't have a company that powerful! Won't somebody think of the children?”—regulators fail to stop the deal. The merger finalizes under the banner of **PerfectDigital**. Armed with the world’s most dominant operating system, an industry-standard word processor, and an unrivaled office suite, PerfectDigital sits poised to dictate the future of computing.

[31:43](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1903s) The Nintendo Opportunity and the Death of Mozilla

In this alternate timeline, Perfect Digital needs a browser to let users navigate "the Grid" from their operating system platform. To solve the problem, they seek out two prominent figures in the computing scene. The first is Jim Clark, founder of Silicon Graphics, who spent the 1980s engineering high-performance 3D graphics workstations. The second is Marc Andreessen, fresh out of the National Center for Supercomputing Applications (NCSA), where he had been one of the co-creators of the pioneering NCSA Mosaic browser. Both have recently left their respective posts and are exploring new ventures. Their initial impulse is to build an outright "Mosaic killer" under the project code name "Mozilla" (“how about that for a name”).

Before that effort can take off, the phone rings. Nintendo Corporation reaches out with an enticing proposition: they are developing their next-generation console, the Nintendo 64, and want players to be able to plug into a wall phone jack to play titles like *Mario Kart* and *Donkey Kong* against friends worldwide. Clark and Andreessen reflect on the offer and decide it is too compelling to pass up; as Dylan Beattie notes, “the Mozilla project never happens.” They relocate to Seattle to join Nintendo, abandoning their browser plans entirely.

Upon arriving at Nintendo's Seattle office, Clark and Andreessen encounter another formidable duo: John Carmack and John Romero. A couple of years earlier, powered by pizza, caffeine, and an intense marathon weekend, Carmack and Romero had ported *Super Mario Bros. 3* to run on MP/M-86. They delivered a disk to Nintendo demonstrating smooth side-scrolling performance squeezed out of a humble 286 CPU—a technical feat widely deemed impossible at the time. Nintendo was so impressed that they hired both programmers on the spot. With Nintendo bankrolling the dream team, Clark (the workstation hardware pioneer), Andreessen (the network browser engineer), Carmack (“probably the finest graphics programmer of any generation”), and Romero (“who's quite good at level design and having good hair”) disappear into secret console development, leaving the browser space wide open.

```mermaid flowchart TD accTitle: The Divergent Browser Founders accDescr: Flowchart showing how Jim Clark and Marc Andreessen join Nintendo instead of creating Netscape/Mozilla.

subgraph BrowserTrack["Abandoned Browser Plans"] A["Jim Clark (Silicon Graphics)"] & B["Marc Andreessen (NCSA Mosaic)"] C["Proposed Project: 'Mozilla' (Mosaic Killer)"] A --> C B --> C end

subgraph NintendoConsortium["Nintendo Seattle Dream Team"] N["Nintendo 64 Network Initiative"] JC["John Carmack"] & JR["John Romero"] Mario["Ported Super Mario Bros. 3 to MP/M-86"] --> JC & JR N --> A & B JC & JR --- N end ```

[34:02](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=2042s) Cello, Grid Perfect, and the SAFE Protocol

With Andreessen and Clark off the market, Perfect Digital must look elsewhere. They are left with their GEM/3 desktop interface on a 286 visual operating system and still desperately need a grid browser. Scanning the next candidates on their list, they locate Thomas Bruce, a librarian running the legal library at Cornell Law School, who has authored an MP/M and GEM browser named Cello. Because WordPerfect is already the entrenched word-processing standard across American legal practices, Perfect Digital enjoys deep existing ties with the legal community.

A deal is struck rapidly to acquire Bruce’s Cello source code. Perfect Digital optimizes its underlying libraries, layers on new capabilities, and fits the software directly into its branded lineup alongside WordPerfect, LetterPerfect, and DataPerfect: "Grid Perfect." They bake in a crucial addition: SAFE, or Secure Application Frame Encryption. SAFE runs alongside the GRID (Global Repository of Internet Documents) protocol, marking the first time a grid client can communicate with a grid server without third parties sniffing or eavesdropping on transmissions.

This cryptographic foundation instantly ignites the protocol's built-in `SUBMIT` verb. The network expands far beyond mere document retrieval, updates, and deletions. Because Perfect Digital embeds core WordPerfect editing tools directly into the browser, every user who consumes the grid can also write to it; publishing capability is accessible from day one via integrated WYSIWYG authoring tools.

- **Grid Perfect Lineage**: Acquired from Thomas Bruce's Cello browser developed at Cornell Law School. - **Protocol Suite**: GRID combined with SAFE (Secure Application Frame Encryption) for secure end-to-end transmissions. - **Read/Write Integration**: Full WYSIWYG WordPerfect engines embedded inside the client turned every viewer into an immediate publisher.

[35:46](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=2146s) The Rejection of Java and the Triumph of HyperLisp

Static documents quickly prove restrictive for complex interactive software, prompting demand for an embedded programming language across clients and servers. In the mid-1990s business software landscape, Beattie notes the natural choice is Lisp—a half-century-old language with “your father's parentheses elegant weapons for a more civilized age.” At the same time, Paul Graham is steering his startup Viaweb, having realized that Lisp's native support for continuation-passing style is ideal for stateless grid protocols. Continuation passing enables server processes to snapshot execution state mid-stride and resume it later without persistent memory allocations or shared-state overhead.

Sun Microsystems attempts to derail this by pitching their own creation, Java—the cross-platform, object-oriented language they are pushing aggressively everywhere. Sun visits Perfect Digital and encourages them to adopt Java. When Perfect Digital mentions their existing language, HyperLisp, Sun asks if they might rename it to "JavaLisp" or "JavasScript" to ride the trend.

To resolve the issue, Perfect Digital invites Paul Graham in for an afternoon discussion. Graham lays out his programming philosophy, details Lisp's historical strengths, breaks down symbol types, and explains the concept he later popularized as the "Blub paradox"—the idea that programming languages are fundamentally unequal in power. By late afternoon, Sun's Java proposal is dead. Graham seals the case with two decisive arguments: 1. **Homoiconicity and State Mobility**: In HyperLisp, every piece of program code is an expression, and every expression is a data structure. Because the language is entirely homoiconic and mutable, developers never have to write serialization or deserialization layers. A running program can freeze its execution, serialize into a native data structure, travel across the network wire to a client, rehydrate instantly, and resume work where it stopped. 2. **The Economic Argument for the Bean Counters**: Graham appeals directly to the accountants by explaining that this architecture offloads compute to the user: “look this basically means everyone who starts using your application is effectively donating their computer to you so you can run code on that so you don't need to invest in so much infrastructure.” Much like DNS, application scalability naturally tracks user adoption.

[39:29](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=2369s) HyperLisp in Action: Asynchronous distributed scripting

HyperLisp becomes the universal client-server standard, cleanly embedded into HTML (specifically version 4.01, because version 4.0 had typographical errors in the specification). Code is embedded directly within web pages using Lisp's characteristic prefix notation. In the simplest use case, developers bind an event directly to an element:

```lisp (bind (document:load) (lambda (element event) (alert "Hello, world!"))) ```

Rather than treating the DOM as an awkward document object, HyperLisp uses native symbol types that map browser primitives directly into the runtime without requiring verbose namespacing.

```mermaid flowchart LR accTitle: HyperLisp Client-Server Execution Loop accDescr: Flowchart of a HyperLisp client firing an asynchronous grid-submit callback to a server and updating the DOM span.

subgraph ClientBrowser["Grid Perfect Browser"] DOM["HTML 4.01 Element"] Callback["(lambda (response)...)"] Span["DOM Span Update"] DOM -->|"submit event"| Callback end

subgraph ServerInfra["Grid Server"] Endpoint["Grid Processing"] end

Callback -->|"grid-submit (asynchronous)"| Endpoint Endpoint -->|"hydrated data response"| Callback Callback --> Span ```

In a more sophisticated example, a form's submit handler triggers an asynchronous callback across the network using a built-in `grid-submit` method. When the server returns data, the callback immediately assigns the response into a target `span` element:

```lisp (bind (form:submit) (lambda (element event) (grid-submit "/api/data" (lambda (response) (set-value! result-span response))))) ```

As Beattie points out with clear amusement, this delivers full asynchronous distributed computing out of the box—sparing this universe the fate of another timeline, where “in a parallel universe they take 10 years to invent and they'd call it Ajax and it would be horrible.” Arithmetic and function calls rely on prefix syntax, leaving hyphens completely free for identifier names. Because hyphenated variable names just work without clashing with subtraction operators, the community avoids decades of divisive debates over `camelCase` versus `snake_case`. Rich, first-class lambdas provide natural asynchronous communication right out of the box, establishing an elegant, unified foundation for distributed computing.

[04:11:37](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=15097s) GEM 95 and the Beatles Reunion

Perfect Digital launches GEM 95, heralded as their most powerful desktop operating system ever. To celebrate, they pull off the ultimate marketing coup: getting the Beatles back together for a massive launch party. The "Grid" browser is baked directly into the operating system from the ground up, and to encourage content creation, they ship a free version of WordPerfect. This version has no printer support, meaning its sole purpose is to let users author documents and publish them directly to the Grid.

The Beatles reunion is a sensation, and Perfect Digital capitalizes on it by featuring the browser's "back" button heavily in all of their marketing literature. The strategy is an overnight success, and the public falls in love with the seamless integration of the Grid. In this timeline, the future of computing looks incredibly bright—unless, of course, you happen to be working for Apple.

- **GEM 95 Launch:** Integrated Grid browser, bundled with a printer-free version of WordPerfect for Grid authoring. - **The Beatles Reunion:** The band reunites for the launch party, and the browser's "back" button becomes a central marketing icon.

[04:12:18](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=15138s) Apple's Struggle and the Rise of BeOS

At Apple, things have gone from bad to worse. Having previously ousted Steve Jobs, the board is now deeply regretting the decision as the company flounders. During a tense board meeting, CEO Gil Amelio tentatively suggests a radical move: buying NeXT to bring Steve Jobs back. The board flatly rejects the idea. As Dylan Beattie notes, “They laugh him out of the room this is the dumbest idea and there is no way Steve Jobs is ever going back to Apple after the way they kicked him across the street.”

However, Jobs is not the only visionary engineer to have left Apple. A startup called Be Incorporated, founded by former Apple executives, has developed BeOS. This brand-new, ground-up, high-performance operating system features dedicated multi-threading designed specifically for streaming media, real-time communication, and audio processing. While the software is brilliant, Be Incorporated is struggling to manufacture and sell the physical computer hardware to run it. Recognizing this bottleneck, they pivot their strategy to port BeOS directly onto Apple's hardware.

```mermaid flowchart TD accTitle: Alternative History OS and Browser Mergers accDescr: A flowchart showing how BeOS, Apple, and Telenor's Opera browser merged to create the Blink browser on yellow Banana Boxes. subgraph BeInc ["Be Incorporated"] A["BeOS (Multi-threaded OS)"] end subgraph Apple ["Apple Computer"] B["Struggling Hardware Maker"] end subgraph Telenor ["Telenor (Norway)"] C["Opera Browser (Presto Engine)"] end A -->|"Ported & Acquired"| B B -->|"Banana Boxes"| D["Yellow Apple Computers"] C -->|"Acquired & Renamed"| E["BeLink (Blink) Browser"] D & E -->|"Combined Platform"| F["First-class competitor to Perfect Digital"] ```

[04:13:48](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=15228s) Banana Boxes and the Birth of Blink

Be Incorporated pitches their high-performance, protected-mode operating system to Apple. Desperate for a savior, Apple enthusiastically acquires Be Incorporated. They completely overhaul their product design, releasing a new line of computers that the press instantly dubs "banana boxes" because they are bright yellow. Up until this point, computers have only ever been gray or beige; the striking yellow design makes them an instant home design sensation.

The only thing the new BeOS-powered Apple computers lack is a native Grid browser. Apple tries to license Cello, but Perfect Digital has already locked it down. They turn to the NCSA, but Mosaic is open-source, limping along, and insufficient. Out of the blue, Apple contacts Telenor, the Norwegian telephone company. As Beattie tells it:

> “Telenor say you know this is funny we have a grid browser we've been working on here it's a thing we call opera and we're not quite sure what to do with it.”

Apple and Be Incorporated strike a deal to acquire Opera, including its Presto rendering engine and network stack. They rename the browser "BeLink" to match the BeOS branding, though autocorrect instantly and permanently dubs it "Blink." Supporting all the standard Grid protocols—Create, Update, Delete, Select, and Submit—and featuring a built-in WYSIWYG editor, Blink becomes a formidable, first-class competitor to Perfect Digital.

[04:15:38](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=15338s) The Search Problem and the Birth of Goggle

With these competing platforms, the Grid experiences an unprecedented explosion. Internet service providers offer dial-up access bundled with five megabytes of free Grid space. Because heavy JavaScript frameworks were never invented in this timeline, five megabytes is an enormous amount of storage, allowing users to publish vast amounts of content. However, because anyone can publish anything, the Grid quickly becomes cluttered with low-quality pages. The massive, unsolved problem of this era becomes discovery: how do users find high-quality content?

At Stanford, two PhD students, Sergey Page and Larry Brin, are working out of a garage on a search algorithm they call "BackRub." The algorithm determines a document's credibility by analyzing how many other pages link back to it. Sergey suggests renaming the search engine after a "googol"—the number 10 followed by 100 zeros—to represent the vast scale of the Grid. However, when they send someone to register the domain, autocorrect strikes again, and they end up registering `goggle.com`.

Goggle features a beautifully minimalist homepage in stark contrast to the ad-heavy portals of the era. The technology works flawlessly, but the founders face a major dilemma: they have absolutely no idea how to monetize a free, minimalist search engine. They strongly dislike the idea of running ads, leaving them searching for a viable business model.

- **Storage Efficiency:** 5MB of free space is incredibly generous because the timeline lacks bloated JavaScript frameworks. - **The BackRub Algorithm:** PageRank's equivalent, relying on backlinks to establish document authority. - **The Autocorrect Typo:** The search engine is registered as `goggle.com` instead of `google.com`.

[04:17:45](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=15465s) The Black Envelopes and the May 4th Event

The monetization problem is solved when Larry Page receives a call from Steve Jobs, who is secretly working on a major project at NeXT and wants to acquire Goggle's PageRank technology. While the exact financial details of the deal remain shrouded in mystery, the consequences shake the tech world.

In April 1999, while the rest of the world is gripped by Y2K millennium bug hysteria, Silicon Valley tech journalists arrive at their desks to find hand-delivered, mysterious black envelopes. Inside is a black printed card that reads: "Get ready for the next generation" on the front, and "Menlo Park amphitheater, 4th of May 1999" on the back.

Rumors run wild. Many assume it is a promotional stunt for a new *Star Wars* movie because of the "May the Fourth" date. However, when journalists contact Lucasfilm, the studio dismisses the idea entirely. Beattie jokes that Lucasfilm replied, “Don't be ridiculous why would anyone make a sequel to Star Wars just we are not gonna mess with perfection.”

[04:19:13](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=15553s) Steve Jobs Unveils the NextBook

On May 4, 1999, the Menlo Park amphitheater is packed. Steve Jobs takes the stage and delivers a characteristically captivating presentation about the Grid, digital publishing, and the social impact of technology. He then reaches behind the podium and pulls out a laptop called the "NextBook."

The crowd has never seen anything like it. It is less than an inch thick, machined from a single piece of lustrous, anodized black aluminum. It completely dispenses with legacy ports: there is no floppy disk drive, no CD-ROM drive, and no serial port. It features only two USB ports and a headphone jack. When the audience wonders how they are supposed to load software or connect devices, Jobs reveals his masterstroke.

On cue, two dozen Apple employees planted throughout the audience reach under their seats, pull out NextBooks, open the lids, and instantly begin playing a live, wireless video stream of Jobs' presentation. The audience is completely blown away; nobody has ever seen Wi-Fi or streaming video before, let alone on a laptop this portable. Journalists abandon their professional composure, clambering over seats just to get a glimpse of the machine.

- **NextBook Specifications:** 333 MHz ARM RISC processor, 64 MB of RAM, and a 5 GB flash drive. - **Operating System:** Runs the next-generation version of the NeXTSTEP operating system. - **Connectivity:** Features built-in wireless internet, shipping with a compact base station called the NextEnd.

[51:07](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3067s) The NeXT Book, the Millennium Hangover, and Nintendo’s Reality

Continuing the speculative alternate history of personal computing, Steve Jobs’ NeXT introduces the "NeXT Book." Designed for effortless connectivity, users simply plug it into a standard phone jack. It comes equipped with an internal CD-ROM drive and blankets the home with wireless networking, eliminating the need to sit tethered to a wall socket. NeXTSTEP ships with a brand-new "grid browser" called Nexus, whose technical heritage traces back directly to the original browser application created by Tim Berners-Lee on NeXTSTEP hardware a decade earlier. Nexus boasts unprecedented responsiveness, running the fastest HyperLisp interpreter ever built directly within the core operating system architecture, fully integrated with Goggle search technology to pull up queries across local folders, backups, emails, or remote grid servers at once.

The transition past midnight into the year 2000 arrives with little catastrophe: the missiles do not fire, the software world survives the Millennium bug, and everyone wakes up on New Year’s Day with an immense hangover to face an unrecognizable landscape of rival systems. The ecosystem is split: NeXT Books run Nexus; "banana boxes" (BeBoxes) surf using Blink (acquired from Norwegian telco Telenor); Perfect Digital markets Gen 95 with an engine descended from Thomas Bruce’s Cello browser; while Sun Microsystems clings to relevance by continuing to market NCSA Mosaic with Java support that nobody cares about.

Then, in December 2002, Nintendo breaks years of quiet following their acquisition of John Carmack, John Romero, Jim Clark, and Marc Andreessen. After an early hit with *Super Mario Quake World*, Nintendo ships a platform simply called "Reality." It realizes the immersive cyberspace envisioned in Neal Stephenson’s *Snow Crash* and William Gibson’s *Neuromancer*: Carmack and Romero handle the visualization, graphic algorithms, and interface; Andreessen writes the networking stack; and the Silicon Graphics pedigree supplies the high-performance 3D rendering pipeline. Virtual reality takes over global communications.

```mermaid flowchart TD accTitle: Alternate Computing Ecosystem circa 2002 accDescr: Mapping hardware platforms and software stacks in Dylan Beattie's alternate history timeline.

subgraph Corporate ["The Alternate Landscape"] direction TB NXT["NeXT Corporation"] -->|ships| NB["NeXT Book (NeXTSTEP)"] NB -->|runs| NEX["Nexus Browser + HyperLisp"]

BE["Be Inc."] -->|ships| BB["BeBox Hardware"] BB -->|runs| BLK["Blink Browser (ex-Telenor)"]

NIN["Nintendo"] -->|funds Carmack, Romero, Clark & Andreessen| REAL["'Reality' 3D Metaverse (Dec 2002)"] end ```

- **NeXT Book specs**: Built-in CD-ROM, native base-station wireless routing, embedded phone jack interface. - **Nexus search**: Instant, unified querying spanning local disks, archival backups, email archives, and remote network nodes via Goggle algorithms. - **Industry reactions**: Sun Microsystems heavily subsidizes Mosaic/Java promotions, facing widespread indifference from engineers who favor HyperLisp.

---

[54:09](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3249s) The Historical Forks: Kildall, Berners-Lee, and Language Syntax

Dylan Beattie shifts from the counterfactual narrative to inspect the real-world historical pivot points that made our actual timeline just as improbable. The dominance of Microsoft began on a single afternoon in 1981 when IBM went shopping for a microcomputer operating system. They initially visited Digital Research founder Gary Kildall to license CP/M for the original IBM PC. Kildall missed the meeting because he went flying in his private plane on a clear day—though alternate accounts suggest he disliked IBM's terms and simply walked away. Because Kildall was unavailable, IBM turned to Bill Gates and Microsoft, then a modest outfit of 10 to 15 people who were writing debuggers and language tools in Washington. Had Kildall signed the agreement, IBM’s PC revolution would have cemented Digital Research as the software hegemon instead of Microsoft.

```mermaid flowchart LR accTitle: The 1981 IBM OS Decision Fork accDescr: Flowchart contrasting the decision to visit Gary Kildall versus Bill Gates for the IBM PC operating system.

IBM["IBM PC Project (1981)"] --> Choice{"First Meeting"} Choice -->|"Gary Kildall (Digital Research)"| CP["Kildall out flying / rejects terms"] Choice -->|"Bill Gates (Microsoft)"| DOS["Signs OS deal; births modern Microsoft empire"] CP -.->|"What could have been"| CPMS["CP/M becomes ubiquitous personal computing standard"] ```

Even fundamental components like URLs were contingent on quick personal choices. In an archival interview snippet, Tim Berners-Lee reflects on whether he would change any architectural decisions in the World Wide Web:

> “It turns out really if you think about you don't need this knife slash I could have designed it to not have the session boy now that people know people all over the people on the radio are calling it backslash backslash means that everybody's people haven't they use that finger so much think of the amount of you know of print now that would have been sage if we wouldn't have sales if I just remove the double slash but there you go yeah and you can't it seems like a good idea at the time could have been different” — Tim Berners-Lee

Markup syntaxes faced similar forks. HTML adopted its angle-bracket convention because Berners-Lee drew upon SGML, but markup could have readily mimicked LaTeX (popular across Sun and academic workstations since the early 1970s) or Microsoft’s Rich Text Format (RTF), which was established by the late 1980s.

- **Gary Kildall**: Pioneer behind CP/M, whose missed meeting with IBM redirected operating system history to MS-DOS. - **URL syntax regret**: The `//` was added to signify an authoritative path starting from the top domain root, but proved redundant since addresses always required the `http:` scheme header. - **Markup alternatives**: SGML's tag paradigm won out over backslash-heavy LaTeX commands and curly-brace RTF formatting blocks.

---

[57:04](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3424s) Near-Misses: Viaweb, Netscape, Nintendo, and Excite

The alternate world’s use of HyperLisp was rooted in real engineering trends of the 1990s. Paul Graham and Robert Morris built Viaweb entirely on Common Lisp, establishing the first viable e-commerce platform and selling it to Yahoo to fund the creation of Y Combinator. Embedded Lisp variants already commanded enterprise software: AutoCAD shipped with AutoLISP, and Emacs ran Emacs Lisp. When Netscape recruited Brendan Eich, their opening mandate was to embed Scheme (a dialect of Lisp) into the browser. JavaScript only emerged because management mandated a rush delivery under fierce deadlines, demanding a curly-brace syntax that superficially resembled Sun’s C-like Java to avoid alienating mainstream programmers.

Parallel close calls defined the early hardware and business landscape: 1. **The id Software Nintendo Pitch**: John Carmack and John Romero successfully engineered smooth side-scrolling on PC hardware and ported *Super Mario Bros. 3* to the IBM PC overnight, shipping a floppy disk to Nintendo. Instead of offering jobs, Nintendo threatened catastrophic legal action because physical game cartridges represented their sole, impenetrable protection against PC piracy. 2. **Netscape’s Origins**: Marc Andreessen and Jim Clark held high-level discussions with Nintendo to design a proprietary online gaming infrastructure before pivoting to found Mosaic Communications (later Netscape). 3. **Cello and BeOS**: Thomas Bruce truly wrote Cello as a Windows-based browser while serving as legal librarian at Cornell Law School; BeOS was built by ex-Apple executives, and Apple CEO Gil Amelio came within inches of purchasing BeOS before pivoting back to NeXT, bringing Steve Jobs back into Apple. 4. **The Excite-Google Buyout**: In 1999, Larry Page and Sergey Brin offered to sell Google’s search algorithms to web portal Excite for $800,000 because they lacked an ad business model. Excite’s executive team declined, deciding Google was not worth the money.

Dylan Beattie synthesizes these near-misses into an overarching critique of modern tech development, cautioning software creators against paralyzing perfectionism:

> “So many of these systems that we use and we rely on every day are not there because somebody worked really really hard to find the best solution or to make the best decision or to find the best possible team or circumstances to work with they're there because people found something that was just good enough and they thought about what can we do with it what's gonna be next cuz you know we have this thing in our industry analysis paralysis where you get overwhelmed with choice you don't know what to do next” — Dylan Beattie

[01:01:05](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3665s) The Contingent Web and Making Progress

Dylan Beattie brings his counterfactual journey through web history to a close by reflecting on the sheer accidental nature of modern computing foundations. He recalls how easily the industry could have pivoted down completely different paths—whether into an alternate universe where key pioneers went off to build an online gaming network instead of the web browser, or our own reality, where global digital infrastructure ended up powered by a programming language hastily cobbled together in just ten days under severe corporate marketing pressure from Sun Microsystems. Looking back at these pivotal turning points, Beattie argues that historical technology choices cannot simply be categorized as correct or flawed moves.

Instead, the reality of engineering history is pragmatic and messy: what matters at any given moment is merely finding a viable path forward. As Beattie observes, “Even if you don't get those decisions right there isn't really a right and wrong all there is is have we got enough to go on and and make progress and as you see in this talk we could have ended up making a very very different kind of progress.” While musing that the future might still take unexpected detours, he welcomes the audience back to our present reality, encouraging everyone to appreciate the everyday web browsers we take for granted—Google Chrome, Mozilla Firefox, Apple Safari, and Microsoft Edge—before concluding his exploration of the web that never was.

## Notable quotes > [00:06](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=6s) Don't worry about changing database providers or worry, but know that every company who used an ORM in case they needed to switch databases never actually did it. — Dylan Beattie

> [27:36](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1656s) What Gary Kildall has realized and IBM haven't is that if your flagship computer is made out of off-the-shelf parts that you bought from China and Singapore and Taiwan, you don't actually own anything. — Dylan Beattie

## Mentioned Alphabet, Amazon, Apple, Apple Safari, ARM RISC, ARPA, ARPANET, Banana Boxes, Baz Luhrmann, Be Incorporated, BeLink, BeOS, Bill Gates, Blink, Brendan Eich, Cello, CERN, Cornell Law School, CP/M, CP/M-86, Digital Research, DNS, Douglas Adams, Dylan Beattie, Elizabeth 'Jake' Feinler, Excite, Gary Kildall, GEM 95, GEM/3, Gil Amelio, Global Repository of Interlinked Documents (GRID), Goggle, Google Chrome, Grid Perfect, GridBrowser.app, HOSTS.TXT, HTML, HTML 4.01, HyperLisp, IBM, IPv4, James Clark, Java, JavaScript, Jim Clark, John Carmack, John Romero, Larry Page, LaTeX, Marc Andreessen, Microcosm, Microsoft, Microsoft Edge, Mosaic, Mozilla Firefox, MP/M-86, Multicosm PLC, National Center for Supercomputing Applications (NCSA), NCSA Mosaic, NeXT, NextBook, NextEnd, NeXTstation, NeXTSTEP, Nexus, Nintendo, Nintendo 64, Opera, Paul Graham, Paul Mockapetris, PC/AT, PerfectDigital, Presto, Reality, Robert Morris, SAFE (Secure Application Frame Encryption), Scheme, Sergey Brin, SGML, Silicon Graphics, SMTP, Standard Unique Resource Finder (SURF), Stanford Network Information Center (NIC), Steve Jobs, Sun Microsystems, Super Mario Bros. 3, TCP/IP, Telenor, The Beatles, Thomas Bruce, Thomas J. Watson, Tim Berners-Lee, University of Southampton, University of Southern California (USC), Viaweb, WordPerfect, WordPerfect Corporation, Y Combinator, Yahoo

## Video description The story of the web is a story about freedom. It's a story about information, about breaking down barriers, about creating new ways for people to communicate, to collaborate, and to share their ideas. It’s also a story that has as much do with marketing, money and meetings as it does with research and innovation. It’s a story of mediocre ideas that succeeded where brilliant ideas failed, a story of compromises, rushed deadlines and last-minute decisions. And it could so easily have been very, very different. What if IBM had hired Digital Research instead of Microsoft to develop the operating system for their first PC, way back in 1980? What if Marc Andreessen and Jim Clark had gone to work for Nintendo in 1993 and never founded Netscape? What if one of the team at CERN had said “Tim, won’t it sound a bit silly if everyone spends the next fifty years saying double-you-double-you-double-you all the time”?

In this talk, Dylan Beattie will explore alternative history of the world wide web - a web with no Microsoft, no Windows; no Firefox, no Google and no JavaScript. A software industry from another timeline, a world of platforms, protocols and programming languages that are unmistakably alien - and yet strangely familiar.

So strap in, hold tight, and join us as take you on a journey through... the web that never was.

NDC Conferences https://ndcsydney.com https://ndcconferences.com

[Source](https://www.youtube.com/watch?v=8JOD1AQGqEg)

中文简报

未曾存在的万维网——Dylan Beattie Dylan Beattie 构想了一段精妙的计算机架空历史:Gary Kildall 创立的 Digital Research 击败微软主导了 PC 市场;Tim Berners-Lee 建立了采用 SQL 动词和波浪号标记语言的“GRID”,而非万维网;HyperLisp 则取代了 JavaScript。在这一时间线中,Marc Andreessen 与 John Carmack 联手为任天堂开发网络系统,PerfectDigital 推出了 GEM 95,Steve Jobs 则展示了一款由名为 Goggle 的搜索引擎驱动的超薄阳极氧化 NeXT 笔记本电脑。随后,Beattie 剖析了这些设想背后真实世界中曾擦肩而过的历史瞬间——从 Brendan Eich 最初打算将 Scheme 嵌入 Netscape,到 Tim Berners-Lee 对 URL 中双斜杠设计的遗憾。演讲最后总结道,现代技术的定义并非源于达尔文式的完美演化,而是源于帮助工程师不断前行的务实且“足够好”的妥协。 软件领域的技术主导地位极少由达尔文式的适者生存所决定;日常工具和标准的成功,往往只是因为它们“足够好”,足以打破分析瘫痪并借由历史偶然胜出。 (5:16) 现代计算和网络的发展维系于微小的历史分岔点,例如 IBM 在 1981 年命运攸关的操作系统交易,以及在商业截止日期压力下仓促用 10 天时间创建的 JavaScript。 (54:09) Dylan Beattie 构建了一个包含 GRID 协议、HyperLisp 以及 Gary Kildall 占据主导地位的架空历史,以此阐释偶然的选择是如何塑造我们如今的数字格局的。 (18:40)

Reading summary

Software architect Dylan Beattie argues that the modern technology stack is not the product of Darwinian optimization, but the result of arbitrary compromises, mundane accidents, and 'good enough' solutions. Through an alternate-history thought experiment, he illustrates how slight real-world divergences could have completely reshaped personal computing, operating systems, and web architecture. The Darwinian Myth and Accidental Standards (5:16) Technologists often assume the industry evolves through the survival of the fittest, where the superior technical architecture inevitably wins out. Beattie disputes this narrative, asserting that pervasive standards regularly originate as fragile prototypes, technical debt, and compromises held together by circumstance rather than deliberate foresight. Relying on Douglas Adams' aphorism that technology is merely 'stuff that doesn't work yet,' he emphasizes that once systems achieve baseline reliability, users cease questioning their structural absurdities. Historical Near-Misses That Built Today's Web (54:09) Modern computing history pivoted on fragile personal decisions and corporate coincidences. IBM turned to Microsoft for an operating system only after Digital Research's Gary Kildall missed a pivotal 1981 licensing meeting—or walked away over unfavorable terms—allowing MS-DOS to displace the technically superior CP/M. Similarly, Brendan Eich was originally tasked with embedding Scheme into Netscape's browser, but corporate deadlines and pressure to mirror Sun Microsystems' Java produced the rushed, curly-brace syntax of JavaScript in just ten days. The Counterfactual Grid and Distributed HyperLisp (18:40) Beattie constructs an alternate timeline to demonstrate how sensible engineering trade-offs could have produced an entirely different architecture: the 'GRID.' In this reality, Tim Berners-Lee replaces redundant UNIX double-slashes and query strings with an exclamation mark delimiter (`grid!cern.ch`), swaps colliding HTML angle brackets for tildes (`~`), and adopts SQL-style six-letter verbs like `SELECT`, `CREATE`, and `SUBMIT`. Rather than dealing with persistent memory allocation, browser vendors adopt Paul Graham's 'HyperLisp,' exploiting homoiconicity and continuation-passing style to seamlessly move serialized execution states across network wires. Alternative Platform Wars and Lost Convergences (31:43) The speculative world traces other historical paths that almost occurred. Nintendo hires John Carmack, John Romero, Jim Clark, and Marc Andreessen to build an online gaming console network, preventing the creation of both Mozilla and *Doom*, and ultimately releasing an immersive 3D cyberspace platform dubbed 'Reality.' Meanwhile, Apple acquires BeOS instead of NeXT, releasing yellow 'banana box' hardware powered by Telenor's Presto-based 'Blink' browser, while Larry Page and Sergey Brin sell their BackRub search technology directly to Steve Jobs' standalone NeXT Corporation. Overcoming Analysis Paralysis in System Design (57:04) Beattie grounds the speculative narrative back in contemporary software development, cautioning engineers against the analysis paralysis that stems from chasing optimal architectures. Real-world standards like DNS endured because Paul Mockapetris scrapped competing compromise proposals and built a simple, self-scaling hierarchical system, while Tim Berners-Lee openly admitted the URL's double slash was an unnecessary design quirk. Software builders must recognize that success depends not on flawless choices, but on finding pragmatic, workable paths forward. Modern computing infrastructure remains deeply contingent on historical flukes rather than preordained technical superiority. Beattie concludes that software developers must abandon perfectionist paralysis, accept that foundational technologies are imperfect workarounds, and focus simply on making progress with whatever solutions prove good enough.

精炼阅读摘要

Dylan Beattie 指出,现代计算体系并非依循“最优技术胜出”的达尔文式演进逻辑,而是由偶发事件、商业妥协以及“刚好够用”的技术债务堆砌而成。通过构建一套高度合理的平行历史,他揭示了万维网与当代软件生态在形成过程中的高度偶然性与路径依赖。 技术演化的达尔文神话与早期基石的妥协 (5:16) 软件工程界普遍存在一种虚假认知,误以为当下的技术栈是通过理性竞争筛选出的最佳成果。Dylan Beattie 引用作家 Douglas Adams 对技术感知的论述指出,人们往往将出生前已存在的事物视为自然规律,而将后来出现的技术神圣化或妖魔化。实际上,现代数字基础设施的起点充满了极具偶然性的妥协。例如,在域名系统(DNS)诞生之前,全美联网计算机依靠人工致电斯坦福网络信息中心的 Elizabeth Feinler,在单一的 HOSTS.TXT 文件中手动注册;逢节假日或夜间便无法增减节点。当南加州大学要求 Paul Mockapetris 在五份互不相让的妥协提案中做选择时,Mockapetris 索性摒弃了全部提案,独自设计出沿用至今的分层自治 DNS 架构。这表明基础协议的确立往往不是共识博弈的最优解,而是个人果断推行实用架构的结果。 平行宇宙中的网格系统与语法改良 (18:40) 在演讲虚构的替代时间线中,Tim Berners-Lee 在欧洲核子研究组织(CERN)构建的不是“万维网”,而是“网格”(the GRID)。该体系基于极简实用主义对现实世界的缺陷进行了修正:HTTP 动词被设计为与 SQL 对齐的六字符统一动词(SELECT、CREATE、UPDATE、DELETE、SUBMIT),不仅规范了日志格式,还避免了端点语义混乱;URL 摒弃了源自 UNIX 文件系统的冗余双斜线与明文认证信息,改用感叹号分隔符。由于高能物理学家大量使用数学不等号,标记语言放弃了 SGML 的尖括号,转而采用 ASCII 中出现频率最低的波浪号(~)作为标记符,并将双换行符直接作为段落划分,使海量旧科研文本在免修改的前提下即时兼容。这一思想实验表明,许多现行网页语法标准并非出于底层计算原理的必然,仅仅是特定历史环境下的随手采纳。 平台格局的分流:商业重组与编程范式逆转 (31:43) 虚构历史进一步推演了操作系统与脚本语言的走向:Gary Kildall 凭借对硬件外包本质的敏锐洞察,让 Digital Research 在与 IBM 的谈判中保留了操作系统的分销权,随后与 WordPerfect 合并为巨头 PerfectDigital;Marc Andreessen 与 Jim Clark 接受任天堂的邀约去研发 Nintendo 64 联网对战平台,使原版 Netscape/Mozilla 胎死腹中。与此同时,Perfect Digital 采纳了 Paul Graham 的建议,将原生支持延续传递风格与同代码数据性的 HyperLisp 确立为网络编程通用标准。借助 Lisp 代码即数据的特性,系统无需复杂的序列化即可在客户端与服务端之间冻结和迁移运行状态,算力被天然分担给终端设备。这一设想规避了现实中长达十年后才迟缓诞生的 Ajax 机制,也消除了关于变量命名法与复杂面向对象层级的旷日持久的争论。 现实历史的转折点:偶然性主导的技术世界 (54:09) 回到现实历史,现代计算架构的分水岭同样由微小偶发事件决定。1981 年,IBM 因未能如期与驾驶私人飞机的 Gary Kildall 敲定协议,转而将操作系统订单交给当时仅有十几名员工的微软,自此重写了个人电脑产业版图。在万维网设计上,Tim Berners-Lee 后来公开坦承,URL 中的双斜线完全是多余设计,平白浪费了全球海量的纸张印刷与键入时间。而在脚本语言方面,Netscape 最初招募 Brendan Eich 本意是将简洁的 Scheme(Lisp 方言)内置入浏览器,却迫于与 Sun 公司 Java 营销结盟的商业死线,仅用十天仓促赶制出具备 C/Java 风格花括号的 JavaScript。此外,Carmack 与 Romero 曾成功将《超级马力欧兄弟 3》移植至 PC 并呈递任天堂,却招致版权威胁;Excite 曾有机会以 80 万美元收购最初的 Google 算法,却因管理层短视而拒绝。历史事实印证了计算世界的演化充满断裂与巧合。 Dylan Beattie 总结道,开发者无须陷入追求完美架构与绝对正确选型的“分析瘫痪”。现代社会所依赖的核心技术底座,几乎都是在特定现实约束下由“勉强够用”的方案逐步演变而来;工程实践的本质并非寻找唯一真理,而是在既定环境下抓住足以维持前进势头的务实路径。

Detailed notes — zh

## Background 本内容由技术大会频道 NDC Conferences 于 2018 年 11 月 13 日在 YouTube 上发布。演讲由 Dylan Beattie 主讲,旨在通过构建一个没有微软、没有 JavaScript、没有 Google 的平行技术宇宙,探讨万维网历史上那些因偶然巧合、商业妥协和仓促决策而塑造的演进轨迹。 ## TL;DR Dylan Beattie 在本次演讲中推演了一条万维网的平行历史线:如果 IBM 当年选择了 Digital Research 而非微软,Tim Berners-Lee 抛弃了双斜杠与尖括号而设计出基于波浪号与 6 字母协议动词的“GRID”,网景创始人转投任天堂研发联机游戏,浏览器中原生嵌入的是基于延续传递的 HyperLisp 而非仓促出炉的 JavaScript,现代数字世界将会是完全不同的景象。随后,他将这些脑洞一一对照真实历史,揭示出万维网的发展从来不是达尔文式的“最优解胜出”,而是由无数仓促的商业博弈与“刚刚好够用”的妥协拼接而成的产物。最后,他勉励开发者不必陷入分析瘫痪与技术崇拜,因为整个现代软件生态正是建立在一系列偶然且充满缺陷的基石之上。 ## Atmosphere & style 整场演讲充满英式幽默、戏谑讽刺与极具感染力的极客浪漫主义。开场借用防晒霜演讲的调侃迅速拉近距离,随后以《迷离境界》般的悬疑戏剧腔将观众拽入逻辑严密却荒诞不羁的架空技术史,引发全场阵阵会心大笑。随着真相层层剥开,情绪由荒诞的推演升华至对软件工程本质的深刻敬畏与反思,给听众带来兼具思想实验快感与现实治愈力的独特观验。 ## Bulletin - [07:28](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=448s) 万维网当今的技术形态并非工程上的最优化选择,而是由一系列历史偶然、商业博弈、仓促妥协与“刚刚好够用”(Good Enough)的方案共同塑造的产物。 - [54:09](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3249s) 蒂姆·伯纳斯-李曾坦言 URL 中协议后的双斜杠(//)纯属多余,如果当年省去这两个字符,全世界将节省难以计数的键盘输入时间与纸张油墨。 - [56:40](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3400s) 网景最初招募布兰登·艾克是为了在浏览器中嵌入 Scheme(一种 Lisp 方言),但迫于交付工期和让语法贴近 C 语言的市场营销压力,艾克在短短十天内仓促拼凑出了 JavaScript。 - [58:25](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3505s) 早期商业决策极具戏剧性:Excite 曾因不看好其商业前景而拒绝以 80 万美元收购 Google,而苹果在收购 NeXT 之前也曾险些选择收购 BeOS,若成真乔布斯将不会重返苹果。 ## Key takeaways - [07:28](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=448s) 打破“技术达尔文主义”迷思:现实中的技术胜利者往往并非架构最优者,而是因商业妥协、偶然机遇或先发优势胜出的平庸或凑合方案。 - [05:16](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=316s) 道格拉斯·亚当斯的技术认知三定律揭示了人类的技术偏见:15岁前存在的事物被视为自然常理,15到35岁间诞生的被视为时代机遇,而35岁后出现的往往被视作反常异端。 - [12:01](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=721s) 在 DNS 发明之前,全网主机依赖斯坦福 NIC 人工编辑维护的一份 hosts.txt,这一单点故障催生了 Paul Mockapetris 独立设计的分布式高扩展域名体系。 - [15:26](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=926s) 超媒体系统的另一条路线:南安普敦大学的 Microcosm 曾主张外置链接数据库(Linkbase),反对像 HTML 那样在文档内部硬编码嵌入标签,至今该技术仍在法律电子数据取证中发挥作用。 - [54:09](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3249s) 微型计算机操作系统的转折点:1981 年 IBM 曾优先拜访 Digital Research 的加里·基尔代尔,因未能谈拢才转投微软比尔·盖茨,从而彻底改写了个人电脑产业格局。 - [56:40](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3400s) 标记语言的偶发性:HTML 沿袭 SGML 引入尖括号语法,而当时学术界公式广泛使用小于号和大于号;若借鉴 LaTeX 或 RTF,网络文档语法可能完全不同。 - [56:40](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3400s) 任天堂的封闭策略改写了历史:约翰·卡马克与约翰·罗梅洛曾成功将《超级马力欧兄弟 3》移植至 PC 并寄给任天堂,但任天堂为保护卡带硬件生态发出了侵权警告,扼杀了 PC 版马力欧的可能。 - [58:25](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3505s) 安德森与克拉克的关键抉择:马克·安德森与吉姆·克拉克曾认真考虑为任天堂开发在线游戏网络,若该合作达成,网景公司将不会创立,互联网浏览器普及史也将推迟或重写。 - [58:25](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3505s) 苹果收购抉择的双重命运:苹果在危机时曾严肃考察收购 BeOS,若未选择收购 NeXT,乔布斯便无从回归,现代苹果生态(iMac、iPod、iPhone)可能根本不会出现。 - [58:25](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3505s) 警惕技术人员的“分析瘫痪”陷阱:不必过度执着于寻找绝对完美的架构方案,技术进步的本质在于借助“刚刚好够用”的现有工具持续向前迭代。 ## Detailed walkthrough ### [00:06](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=6s) 程序员版“防晒霜演讲”:写给开发者的箴言 Dylan Beattie 在轻缓的背景音乐中模仿了著名的《防晒霜》(Wear Sunscreen)演讲风格,向现场观众献上了一段针对软件工程师的幽默箴言。他以戏谑的口吻开篇,宣称自己如果只能给未来留下一条建议,那就是“使用扁平显示器(flat screens)”,因为扁平显示器的好处早在好莱坞就已得到证实,而他接下来的其余建议则完全基于自己漫无边际、随波逐流的人生经历。他谈到了全新项目(Greenfield projects)带来的初期盲目自信:在事情搞砸之前,人们往往无法体会到这种乐观的可贵,但当代码最终发布时,回头看去才会意识到当初是多么简单高效。 接着,他列举了一系列戳中开发者痛处的行业观察与技术调侃。不要把精力浪费在无谓的架构担忧上,比如为了“防止以后更换数据库”而引入复杂的 ORM;事实证明,几乎没有哪家公司真的换过底层数据库。真正会拖垮项目的,往往是无法控制的外部依赖,比如“某位工程师在一个百无聊赖的周二下午四点突然提离职”。在日常编码中,不要去随手重构别人的代码格式,也别纵容别人来格式化你的代码;不必对各种框架盲目崇拜,框架时好时坏,用户根本不在乎底层用什么实现,唯有体验才是核心。Dylan Beattie 还特别提醒道:“Do not read the comments on YouTube they will only make you feel angry.”(不要去看 YouTube 的评论区,它们只会让你感到愤怒。) - **依赖与日志管理**:务必在本地缓存包依赖,天知道什么时候某个开源包就会彻底从网络上消失;要勤读日志文件,这是系统异常时最早报警的第一手信源。 - **语言与底层认知**:编程语言会潮起潮落,但底层设计模式历久弥新;不妨写一次 x86 汇编,但要在自己变得傲慢之前停下;写一次 Visual Basic,但要在变蠢之前收手;哪怕不懂 F# 也不必自责,许多杰出的架构师同样不擅长函数式语言。 - **软件工程的必然真相**:代码永远会有 bug,截止日期一定会延期;当有一天开发者晋升为管理者时,总会产生“自己当年写代码没有 bug、从不拖延、还会调优数据库索引”的幻觉。 - **开源与界面的反思**:对开源心存感激,但不要盲目运行别人的代码;开源不是把项目文件夹随手扔上 GitHub 配个 README 就指望别人来帮你擦屁股;另外不要过度折腾 UI,否则产品交付时看起来会像个廉价的卡拉OK包厢。 --- ### [05:16](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=316s) 道格拉斯·亚当斯的技术定律与代际偏见 在结束了这段兼具调侃与共鸣的开场白后,Dylan Beattie 向悉尼现场的观众热情致意,并引入了科幻作家道格拉斯·亚当斯(Douglas Adams)关于人类面对新技术时心理反应的经典三定律。第一条是,任何在你出生时就已经存在于世上的东西,都是平凡而正常的,属于世界运作规律的天然组成部分;第二条是,任何在你 15 岁到 35 岁之间被发明出来的东西,都是革命性的、令人兴奋的新事物,你大概率能围绕它建立起一份职业生涯;而第三条则是,任何在你 35 岁之后才被发明出来的东西,都是违背自然法则、大逆不道的异端。 Dylan Beattie 拿自己作为这套法则的生动范例。他出生于 1978 年,刚好错过了第一轮去电影院看《星球大战》的黄金时机;在他出生时,世界上已经有了阿帕网(ARPANET)、彩色电视和基础的计算机网络,但现代意义上的互联网仍未破晓。当他大约 15 岁时,万维网(World Wide Web)横空出世,他顺理成章地以此为生,从自学 HTML、担任网站管理员(Webmaster)起步,构建了自己的技术生涯。然而当他跨过 35 岁的门槛后,世界上冒出了 Apple Watch、指尖陀螺,以及风靡一时的街舞动作“Dab”,对于这些 35 岁后出现的新事物,他半开玩笑地坦白自己完全无法理解。这个定律揭示了一个深刻的事实:人类对科技的认知极为主观,完全取决于我们在生命周期的哪个时间节点与这些软硬件系统发生碰撞。 ```mermaid flowchart TD accTitle: Douglas Adams's Rules of Technological Reaction accDescr: A vertical flowchart illustrating the three stages of how humans perceive technology across different age spans. subgraph Perception["道格拉斯·亚当斯的人类技术认知法则"] Birth["0 - 15 岁:出生时已存在"] -->|"视为理所当然"| Normal["正常自然<br/>构成世界基本规则"] Youth["15 - 35 岁:青年期接触"] -->|"视为时代机遇"| Career["革命性突破<br/>可在此领域建立终身事业"] Mature["35 岁之后:中年期涌现"] -->|"视为违背常理"| Unnatural["反常异端<br/>违背事物自然秩序"] end ``` --- ### [07:28](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=448s) 打破“技术达尔文主义”的迷思 Dylan Beattie 顺着亚当斯的话题进一步延伸。道格拉斯·亚当斯曾经说过,所谓的“技术”,其实就是那些“目前还经常出故障的东西”;一旦某样东西变得绝对可靠、从不出错,它就退化成了日常用品(比如椅子坏掉的概率极低,因此没人会把椅子当成技术)。今天人们身边这些被视为理所当然的现代奇迹——智能手机、电动汽车与数字通信,在诞生之初全都是用胶带、绳子、临时代码与技术债务生拉硬扯拼凑出来的简陋原型。如果一个人最初见证的是左侧那些千疮百孔的原型,他至今可能仍觉得这些概念充满争议;而如果一个人出生时就置身于右侧高度成熟的产品世界,就会理所当然地认为智能手机与电动车天生就该存在。 人们普遍倾向于相信一种技术领域的“达尔文适者生存论”,即认为所有互相竞争的技术原型与解决方案中,最终胜出的一定是工程上最优越的那一个。Dylan Beattie 明确反驳了这一观点:“it's easy to think that there is some sort of Darwinian survival of the fittest... but that is absolutely not true.”(人们很容易认为存在某种达尔文式的优胜劣汰……但这完全不是真的。)他明确指出,当今的科技格局充满了因愚蠢偶然、商业博弈或仓促妥协而胜出的平庸方案。本次演讲的核心意图,正是去回溯万维网发展史上那些“纯粹出于偶然或荒谬理由”而偏离轨道的历史节点,推演一段如果当年事情发生了一丁点改变就会截然不同的平行技术宇宙。 --- ### [09:08](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=548s) 巨人生病:IBM 制造个人电脑的仓促决断 Dylan Beattie 将故事的时间轴拉回到 20 世纪下半叶的计算黎明期。在那个年代,IBM 创造了大型主机(Mainframe),计算技术以不可阻挡之势飞速迭代,硬件从真空管进化到晶体管,再到集成电路;编程方式从机器打孔卡、汇编语言演变到解释型与编译型高级语言。当时 IBM 创始人 Thomas J. Watson 甚至曾留下一句著名的误判,认为“全世界也许只需要五台计算机”。然而到了 1970 年代末,计算机逐步走出实验室和大学,开始向个人家庭渗透,IBM 几乎统治了整个信息技术世界。尽管今天人们会提到苹果、Alphabet 或亚马逊的行业影响力,但在当年的 IT 领域,没有任何公司的掌控力能与 1970 年代末的 IBM 相提并论。 然而,彼时的 IBM 已经成长为一个组织庞大、动作极其迟缓的官僚巨兽,甚至内部为了审批采购订单的打印字体就能扯皮六个星期。当意识到个人电脑(PC)可能蕴藏着巨大商机、不能坐以待毙被时代抛弃时,IBM 做出了一个违背其传统自主研发风格的激进决策:他们不再自行制造 PC 的所有零部件,而是决定直接从全球各地的供应链上采购现成配件(COTS)。主板随处采购,CPU 交给英特尔,串行总线则采购自中国台湾或新加坡的代工厂。然而,一台没有灵魂的硬件堆砌体需要操作系统的驱动。于是在 1981 年,IBM 的采购代表们走出了办公室,前去拜访当时在个人计算机操作系统领域无可争议的行业领军企业。 与此同时,大洋彼岸的美国国防高等研究计划署(ARPA)也在推进一个疯狂的构想:为了应对冷战时期的核打击风险,他们试图通过网络将多台计算机互联起来,哪怕其中一台遭到摧毁,其余节点仍能保持通信与计算韧性。正当 IBM 紧锣密鼓地准备发布个人 PC 之际,ARPA 也在分布式网络协议领域迈出了关键一步,两者共同拉开了现代计算世界的帷幕。 ### [12:01](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=721s) 互联网的早期黎明:从手工绘制的 ARPANET 地图到 TCP/IP 与 DNS 的诞生 计算机网络的演进并非一蹴而就,而是一系列标准与妥协的结晶。1981年前后,互联网络协议的标准规范发布,也就是后来大家熟知的 IPv4。在这之前,网络的规模小到不可思议——Dylan Beattie 展示了一张 1973 年 ARPANET 的拓扑图,整张地图甚至是由一位制图员用纸和笔纯手工绘制出来的,清晰地记录了当时接入网络的每一台物理设备及其运行的操作系统。到了 1983 年 1 月 1 日,这一天被称为计算机网络史上的“国旗日”(Flag Day),网络运营者们下达了强制命令:“从现在起,所有连接到我们网络的设备都必须运行 TCP/IP”。随后在 1982 年,简单邮件传输协议(SMTP)正式获批通过。至此,全球计算机不仅实现了底层互联,使用者之间也终于拥有了收发电子邮件的标准途径。 然而,随着互联网的快速扩张,计算机科学中最经典的痛点随之而来——正如业界名言所调侃的,计算机科学只有两个真正困难的问题:缓存失效和给东西命名。在域名系统(DNS)出现之前,网络主机管理原始得令人发指: ```mermaid flowchart TD accTitle: 早期基于 hosts 文件的网络节点注册流程 accDescr: 展示在 DNS 出现前,依靠人工编辑 Stanford NIC 单一 hosts 文件的繁琐且脆弱的注册流程。 subgraph User["用户节点"] A["新服务器接入互联网"] --> B["致电联系 Stanford NIC 的 Jake"] end subgraph Stanford["Stanford NIC(单一故障点)"] B --> C["人工打开并编辑唯一的 hosts.txt 文件"] end subgraph Sync["文件分发与同步"] C --> D["等待参会人员手工带回磁带拷贝"] C --> E["半夜利用低价电话线通过 FTP 同步"] D --> F["各组织更新本地主机解析"] E --> F end ``` 如果有人在网络上新架设了一台服务器,唯一的接入方式就是给斯坦福网络信息中心(Stanford NIC)的人打电话:“嘿,Jake,我加了台新电脑,麻烦帮我加进 hosts 文件里。”所有的主机映射全都依赖这一份集中存放的 `hosts.txt` 文件。如果恰逢节假日、感恩节、圣诞节或者周末深夜,负责维护权限的人全都放假狂欢去了,新机器就根本无法接入。这种脆弱的人肉运维机制显然无法跟上网络的爆发式增长。 南加州大学(University of Southern California)此时聘请了 Paul Mockapetris 来解决这个问题。校方交给他五套不同的工程解决方案,希望他能博采众长、各方平衡。但 Mockapetris 并没有陷入这种无休止的妥协,而是展现了一名顶级工程师的果断:“他看了看所有这五套方案,觉得‘全是垃圾’,直接全扔了,凭一己之力设计并发明了 DNS。”这个由他一手打造的层次化分布式主机名与网络地址数据库,展现出了极其惊人的工程弹性。其核心架构设计完美契合了网络规模的扩展:连接的节点越多,部署的 DNS 服务器就越多,负载越容易分散,缓存命中率越高,向根服务器发起的递归查询反而越少。时至今日,这项最初的设计几乎无需结构性改动,依然稳健地承载着全球数以亿计的域名解析。 - 1981年:IPv4 规范发布;此前 1973 年的整个 ARPANET 拓扑尚能手工绘制在一张纸上。 - 1983年1月1日(Flag Day):全网强制统一切换至 TCP/IP 协议套件。 - 1982年:SMTP 邮件协议正式标准化。 - 早期主机名解析完全依赖存放于斯坦福网络信息中心(Stanford NIC)的人工编辑 `hosts.txt` 文件,节假日及周末无法更新。 - Paul Mockapetris 摒弃了现成的五套设计,独立构思并设计了高度可伸缩的分布式 DNS 层次系统。 ### [15:26](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=926s) 欧洲核子研究组织(CERN)与超媒体的另一条路线:Microcosm 系统的哲学 底层协议成熟后,历史的舞台交给了欧洲核子研究组织(CERN)。在法国与瑞士交界处的地下实验室里,时任软件工程师的 Tim Berners-Lee(他如今依然喜欢谦逊地自称“Web 开发者”)面对着一个痛点:实验室里聚集了大量优秀的物理学家和数学家,内部网络虽然先进,但他们发表、查阅学术论文的方式却极度匮乏,缺少一种数字化的学术期刊载体。于是,Tim 撰写了一份名为《信息管理:一份提案》(Information Management: A Proposal)的报告提交给上级。主管在封面上留下一句传世评价:“含糊但令人兴奋”(“vague but exciting”)。 后人往往理所当然地认为,以 HTML、HTTP 和单向超链接构成的万维网形态是历史的必然选择。但 Dylan Beattie 指出,当时在超媒体系统领域存在着截然不同的技术流派。1997 年 Beattie 在南安普敦大学就读时,他的两位教授就曾在第一届国际超媒体大会上与 Tim Berners-Lee 同台竞逐。当时教授展示的系统名为 Microcosm。Microcosm 坚持一种截然不同的设计哲学:反对将超链接直接硬编码到文档内部。他们认为现实世界中已经存在海量的纯文本、富文本(RTF)和 WordPerfect 格式文档,强制用户重写所有文档去嵌入标签是荒谬的。 | 架构特性 | Tim Berners-Lee 的 WWW 方案 | 南安普敦大学的 Microcosm 方案 | |:--- |:--- |:--- | | **超链接存储方式** | 内嵌于文档内部(嵌入式标签,如 HTML `<a>`) | 独立存储在网络链接数据库(Linkbase)中 | | **对原始文档的影响** | 必须修改、重构原始文档格式 | 绝不篡改原始文档,支持任意只读格式 | | **超媒体解析逻辑** | 浏览器渲染时就地解析文档内的标记 | 后台独立索引服务提取上下文与关键词 | | **界面呈现方式** | 链接作为文档样式的一部分直接点击跳转 | 内容窗口旁常驻独立的链接侧边栏窗口 | | **技术维护角色** | 文档编写者自行排版与插入链接 | 图书馆员与信息分析师统筹维护链接关系 | Microcosm 的核心理念是让内容原汁原味地呈现,超文本逻辑则交由外部的专业信息管理者通过索引数据库施加在内容之上。虽然这套构想未能在后来的公开互联网竞争中战胜简单粗暴的万维网,但它并未消亡。Microcosm 后来演化为 Multicosm PLC,这套在海量未加工文件中建立外置链接与索引的技术,至今仍广泛应用于针对企业被传讯服务器的大规模电子数据证据开示(e-discovery)与跨国法律诉讼中。 Dylan 引用其教授当年的现场心理活动回忆道: > “我就坐在那儿,上台向大家展示了我的系统。接着上台的是这个来自 CERN、名叫 Tim 的家伙。我坐在台下心想:‘得,这个领域我们是赢不了了。’” ### [18:40](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1120s) 进入“迷离境界”:平行时空下的“网格”(GRID)与全新协议诞生 伴随着经典电视剧《迷离境界》(The Twilight Zone)的标志性开场白,Dylan Beattie 将演讲引入了一个架空的平行历史世界:如果当年在细节上出现哪怕一点小小的偶发事件,今天的数字生活会变成什么模样? 在 1990 年代初的 CERN 办公室里,Tim Berners-Lee 正在推敲一套名为“Mesh”的系统原型。由于周边有太多同名项目,他急需一个朗朗上口的新名字。在真实世界里他敲定了“World Wide Web”,但在演讲构建的平行时空中,办公室网络偶然故障了。Tim 不得不通过电话向同事口头汇报这项构想。当他在电话里念出“W-W-W”时,对方立刻打断了他,指出一个语言学上的致命荒谬:“等一下,这个缩写念起来竟然比它的全称还要长!如果这东西真的火了,全世界天天把‘double-you-double-you-double-you’挂在嘴边,听起来未免太蠢了。” 被一语点醒的 Tim 在下午四点的茶歇时间向同事们集思广益。大家给出了各种五花八门的备选提案: - **III**(International Information Interchange,国际信息交换体系):立刻被否决,因为念起来像海盗的口号(“Aye-Aye-Aye”)。 - **MMMM**(Multinational Metadata Matrix,跨国元数据矩阵):刚读完科幻小说《神经漫游者》(Neuromancer)的同事提出的赛博朋克风命名。 - **GRID**(Global Repository of Interlinked Documents,全球互联文档知识库):最终胜出的名称,简洁硬朗。 确立了项目名称为 GRID 后,下一步是打造底层通信协议。在命名惯例上,网络协议多被称为某种“Transfer Protocol”(如 FTP、SMTP),处理超文本传输的协议顺理成章地被拟定为 HTTP。在这个平行时空里,Tim 借鉴了结构化查询语言(SQL)的动词规范,并融入了一种具有强迫症美感的极简工程美学——所有核心请求动词被刻意约束为完全相同的字符长度(全为 6 个字母),以便在等宽字体的日志终端中对齐得严丝合缝: - `SELECT`:从远端服务器检索指定文档并回传(对应现实中的 `GET`)。 - `CREATE`:向 GRID 网络首次发布全新文档。 - `UPDATE`:修改已有文档中的笔误或内容(对应现实中的 `PUT` / `PATCH`)。 - `DELETE`:强制下架内容(例如法务部门介入要求删除时)。 - `SUBMIT`:同事建议增加的操作,用于向远端主机发送执行指令并要求处理返回(对应现实中的 `POST`)。 在这个“网格”(GRID)与 6 字母动词协议搭建完毕的世界里,万维网的寻址机制也即将迎来一种截然不同的演化路线。 ### [21:57](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1317s) 设计更合理的定位符:从 URL 到 Grid 标识体系 在 Dylan Beattie 构建的这条平行宇宙时间线里,蒂姆·伯纳斯-李(Tim Berners-Lee)琢磨了一阵子,发明了一种标准资源定位方案——“唯一资源查找器”(Standard Unique Resource Finder,简称 SURF)。大家觉得“surf”这个词很有可能流行起来,比如以后人们可能会把上网查资料形象地称为“在网格上冲浪”(surfing the grid)。 他的第一版草案沿袭了许多早期设想:`http://user:pass@grid.domain.com:port/path?query#fragment`。然而同事立刻提出了致命质疑:为什么要在一个将要暴露在用户界面上的地址里,以纯文本形式包含用户名和密码凭据?蒂姆欣然接受并回应:“好提议,我们不应该支持这个,那样做太蠢了(Good point, we shouldn't support that, that would be stupid)。” ```mermaid flowchart TD accTitle: 平行时间线中 Grid 地址格式演进 accDescr: 展示从原始草案到去除双斜杠,再到使用惊叹号和隐藏默认协议前缀的演化过程。 subgraph Step1 ["第一代:包含凭据与冗余符号"] A["http://user:pass@grid.domain.com:port/path?query#fragment"] end subgraph Step2 ["第二代:精简语法与默认机制"] B["去掉账密并简化:http://grid.domain.com/path"] C["引入默认文档概念:省略复杂路径与查询"] A -->|"剔除明文账密"| B B -->|"提取常态默认项"| C end subgraph Step3 ["第三代:摒弃 UNIX 双斜杠与端口歧义"] D["避免数字主机名与端口混淆:改用! 分隔协议"] E["http!grid.domain.com"] C -->|"抛弃文件系统 // 语法"| D D --> E end subgraph Step4 ["第四代:极致协议简写与最终定型"] F["协议名与主机同名即可省略:grid!domain.com"] E -->|"同名主机协议自动折叠"| F end ``` 蒂姆接着开始对地址做大刀阔斧的“减法”。在他看来,查询字符串(query)和片段标识(fragment)通常都是可选的,完全可以留给搜索和索引系统专用;同时他还引入了“默认文档”(default documents)机制,输入主域名即可直达首页。接着,蒂姆盯上了由 UNIX 文件系统沿袭而来的双斜杠(`//`)——既然网格系统是一个全新的东西而非传统文件系统,保留它纯属多余,理应删去。 随后有人指出:如果网络上的某台主机名本身就是一个纯数字,去掉特定分隔符后,解析器会极难区分主机名和端口号。为了解决歧义,蒂姆把语法改成了用惊叹号(bang,`!`)来分隔协议与主机:`http!grid.domain.com`。 最后,蒂姆灵光一闪提出了一个优雅的约定:如果网络上某台主机的名称恰好与它所运行的协议相同,那么协议部分就能直接省略!于是,访问 CERN 默认文档的最终全限定地址变成了极简的 `grid!cern.ch`。 当团队里有人质疑这个名字读起来怪不怪时,问他:“蒂姆,难道你不觉得 `grid!cern.ch` 听起来有点傻乎乎的吗?”蒂姆反击道:“可没 `http://www.cern.ch` 那么蠢(Not as silly as HTTP colon slash slash www dot cern dot org)。”团队成员心服口服地附和:“确实,好在我们把它给修掉了。” - 蒂姆将定位方案命名为 SURF(Standard Unique Resource Finder),巧妙契合了“Surfing the Grid”的说法。 - 早期草案彻底剔除了直接在地址中暴露用户名和密码的做法。 - 摒弃了 UNIX 风格的冗余双斜杠(`//`),改用惊叹号(`!`)避免端口与纯数字主机名混淆。 - 确立了协议同名省略规则,例如网格主机 `grid` 运行时直接缩写为 `grid!domain.com`。 --- ### [24:04](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1444s) 科学家的需求与波浪号标记语言(HTML 0.2)的诞生 搞定了寻址方式,下一步就是如何呈现这些超媒体文档。蒂姆考察了自 1980 年代初就在工业排版领域应用的“标准通用标记语言”(SGML),并尝试基于它裁剪出一个精简版草案 HTML 0.1。 然而,蒂姆当时所在的 CERN 是一家物理实验室,周围的同事全是一线科学家、数学家和物理学家。他们平时撰写和发表的论文公式里充满了大量的“小于号”(`<`)和“大于号”(`>`)。科学家们看到 HTML 0.1 方案后集体抗议,直截了当地告诉他:“请千万别用尖括号,因为我们现存的所有科学文档,一旦发布到你的网格系统上,在解析瞬间就会全部崩溃,因为尖括号在物理学里有别的用处。” 蒂姆认为这是个非常严肃的问题。于是他抓取了当时 FTP 服务器和共享网络上所有现存的纯文本论文与文档,对它们做了一次详尽的字符频次统计分析,试图在基础 ASCII 码表中找出 CERN 文档里出现频率最低的字符。最终结果出炉:ASCII 码为 126(即 0x7E)的波浪号(`~`)拔得头筹。 蒂姆拍板决定:波浪号(`~`)将成为这套新标记语言的专属标记指示符!基于这一思路,他制定了极其务实的排版规则: 1. **双换行即段落**:两个回车换行自动解析为段落标记(paragraph break)。由于此前大部分普通文本文件就是用双换行分段的,这意味着现存海量的纯文本论文无须任何改动编辑,原生就是合法的网格文档。 2. **波浪号标记法**:用包含波浪号的标签来定义标题、段落和加粗文本。 到了 HTML 0.2 草案,语法演化出了类似键值对的属性定义(如用冒号声明属性:`html:lang=en`)。蒂姆意识到,一份文档应该清晰地拆分为两大部分: - 一部分是元数据(Metadata),专供机器读取并在浏览器界面的其他区域渲染; - 另一部分是主体文档(Document),呈现于主界面供人类阅读。 在定义超链接时,蒂姆直接将其标签命名为 `link`,理由显而易见——因为它创建的就是链接(link)。面对其他人“为什么不叫它锚点(anchor,`<a>`)”的疑问,蒂姆感到不可思议地质问:“什么叫锚点?为什么会有人把它叫锚点?这根本说不通啊!(What is anchor? Why would you call it that? That makes no sense.)” 这套包含了超链接与内嵌图像支持的 HTML 规范于 1992 年 2 月在网络上正式公布,瞬间在全世界引发了疯狂的“网格热”(Grid Fever)。 商业公司如梦初醒,疯狂招募相关人才;社会上甚至涌现出大量仅参加了两天速成培训班就自封为“网格大师”(Grid Master)并大赚特赚的投机者——只要印一张体面的名片,抓住行业新兴期信息差的红利,没人知道他们其实根本不懂。Dylan 幽默地补了一句,坦白自己当年也是通过这种方式才混到人生第一份工作的。 热潮在伊利诺伊大学国家超级计算应用中心(NCSA)推出第一款开源图形化网格浏览器——NCSA Mosaic 时达到顶峰。Mosaic 功能极其全面,不仅支持已有的各类网络协议(如用来阅读 Usenet 新闻组的 NNTP、Gopher、FTP),还完美支持了全新的 Grid 协议,能够图文并茂地浏览网格页面并实现超链接跳转,瞬间轰动了全球科技界。 - CERN 科学家因物理与数学公式中广泛使用 `<` 和 `>` 而强烈抵制尖括号语法。 - 文本频次统计证实,波浪号(`~`)是基础 ASCII 字符集在学术文档中使用频率最低的字符。 - 原生继承“双回车即分段”的排版习惯,使上百万篇现有纯文本历史文献零成本无缝上线。 - 链接标签遵循直觉直接采用 `link`,完全抛弃了令人费解的 `anchor`(`<a>`)叫法。 --- ### [27:39](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1659s) 另一条时间线:Digital Research 崛起与 Steve Jobs 的 NeXT 镜头切回到硬件操作系统领域。在早期的 1980 年代,IBM 准备为其划时代的个人电脑挑选一款操作系统。在这条平行时间线中,IBM 并未找上微软,而是找到了当时个人电脑操作系统领域最顶尖的技术团队——数字研究公司(Digital Research)及其创始人加里·基尔代尔(Gary Kildall)。 某天 IBM 敲响了基尔代尔的大门,开门见山提出要购买他的操作系统。双方坐下来展开谈判。此时,基尔代尔敏锐地洞察到了 IBM 当时尚未察觉的商业真相: > “如果你们的旗舰电脑全是用从中国大陆、新加坡和台湾买来的现成现货零部件组装而成的,那么你们其实什么技术资产都没有拥有。运行应用程序的关键早就不再是硬件本身——因为硬件谁都能买到;掌控应用程序生态的关键,必将是操作系统。” 基尔代尔深思熟虑后给出了方案:他同意向 IBM 授予 CP/M(微型计算机控制程序)的许可协议,允许 IBM 随整机无限制复制和出货;作为交换,基尔代尔给 IBM 提供 50% 的巨额折扣,但他必须保留对外销售该操作系统的权利。 交易就此达成。IBM PC 搭载着 CP/M-86 隆重面世,而基尔代尔带领的 Digital Research 则继续全力演进技术。1984 年,当 IBM 推出第一台 16 位计算机 PC AT 时,Digital Research 早就准备好了跨时代的大招:MP/M-86(微机多任务处理程序)。 这是计算历史上首款真正的多任务操作系统: - 没有任何历史包袱,完全打破了 8086 处理器强加的“640KB 内存魔咒”; - 它是针对 286 处理器设计的 16 位保护模式操作系统,将硬件性能压榨到了极致; - 最关键的是,它能够运行在全世界各地雨后春笋般冒出来的米黄色 IBM PC 兼容机(兼容机克隆潮)上。 用户只要买一台便宜的兼容机,装上 Digital Research 的操作系统即可。等 IBM 意识到自己丢失了整条生态链的控制权时,一切为时已晚,Digital Research 迅速成长为全行业体量最大的巨头。 与此同时,加州的苹果公司(Apple)在 Apple II、Lisa 和初代 Macintosh 之后陷入内耗纷争。1987 年公司发生恶意收购,股东们直接把远见卓识的创始人史蒂夫·乔布斯(Steve Jobs)无情地扫地出门(“kicked him out on his ass”)。心高气傲的乔布斯索性把新公司搬到了苹果大楼的正对面,誓要让老东家好看,公司取名为 NeXT。 乔布斯彻底厌弃了千篇一律的米黄色外壳,定下了全黑的设计基调,打造出一款边长约 8 英寸的纯黑立方体计算机——NeXTcube(NeXTstation 系列)。当时 IBM 兼容机一台仅售八九百美元,而一台 NeXTstation 售价高达 10,000 美元。它瞄准的是电影后期制作、前沿医学研究以及顶尖的高性能软件开发。 巧合的是,其中几台 NeXT 电脑被 CERN 采购。蒂姆·伯纳斯-李正是借助这台顶配机器,开发出了最初的 Grid 原型应用。历史上第一款网格浏览器——遵循 NeXT 命名规范被命名为 `GridBrowser.app`——就是在这套 NeXT 机器上编译运行的。 而在操作系统霸主地位稳固之后,Digital Research 又做出了震惊业界的举动:与全球第二大软件公司——出品了统治级文字处理软件的 WordPerfect 公司强强联手合并,组建成立了“Perfect Digital”。 美国司法部在长达 18 个月的时间里急得团团转,不断叫嚷着“必须采取行动!不能让一家公司拥有这么大的权力!难道没人为孩子们想想吗?”,却在法律上完全找不到任何可以阻挠合并的实质抓手。这桩超级并购终告完成,软硬件霸主 Perfect Digital 横空出世。他们手里握着全球顶级的办公套件、垄断级别的字处理软件,以及不可撼动的旗舰级操作系统,开始秘密筹划一场更宏大的生态变革。 - 面对 IBM,基尔代尔洞悉硬件商品化(Commoditization)趋势,通过保留非排他性授权将 Digital Research 推向霸主地位。 - MP/M-86 具备真正的多任务处理与保护模式,彻底绕开了 640KB 常规内存限制。 - 乔布斯离开苹果后推出高昂的纯黑 NeXT 计算机,蒂姆·伯纳斯-李恰好用它写出了初代 `GridBrowser.app`。 - Digital Research 与 WordPerfect 合并为 Perfect Digital,集齐了操作系统与杀手级办公应用的绝对霸权,美国司法部反垄断干预无功而返。 ### [031:43] 任天堂的招揽与未曾诞生的 Mozilla 在 Dylan Beattie 构想的平行宇宙中,Perfect Digital 公司亟需一款能在其平台运行的网格浏览器(Grid Browser)。彼时,Silicon Graphics 创始人 Jim Clark 与 NCSA Mosaic 的核心开发者 Marc Andreessen 恰好双双离职,正在物色下一个项目。两人本打算联手打造一个所谓的“Mosaic 杀手”,甚至连代号都想好了——就叫 Mozilla。 就在此时,任天堂(Nintendo)打来了一通改变历史的电话。任天堂表示他们正在研发 Nintendo 64 主机,希望让玩家能直接把电话线插在墙上的插座上,与全球好友联机玩《马力欧卡丁车》或《大金刚》。Jim Clark 与 Marc Andreessen 经过深思熟虑,认定这是个绝佳的主意: > “The Mozilla project never happens they moved to Seattle they start working for Nintendo.” —— Dylan Beattie 当他们搬到西雅图、入职任天堂的第一周,就在办公室遇到了另外两位传奇人物:John Carmack 与 John Romero。几年前,这“两位 John”凭借披萨和高浓度咖啡因的支撑,在某个漫长周末把《超级马力欧兄弟 3》移植到了 MP/M-86 平台上,成功从 286 CPU 中压榨出了不可思议的横向卷轴渲染效果,任天堂见状立刻将他们招致麾下。至此,全明星阵容在任天堂西雅图集结:Silicon Graphics 的图形软硬件巨擘 Jim Clark、Mosaic 浏览器先驱 Marc Andreessen、堪称同时代最顶尖图形程序员的 John Carmack,以及擅长关卡设计且“拥有一头飘逸秀发”的 John Romero。四人由任天堂全额出资包揽,彻底远离了网格浏览器的开发。 ```mermaid flowchart LR accTitle: 任天堂西雅图全明星团队组建 accDescr: 展示 Marc Andreessen、Jim Clark 与 id Software 创始人加入任天堂放弃 Mozilla 的过程。 subgraph BrowserWorld["网格浏览器领域"] JC["Jim Clark<br/>(SGI)"] MA["Marc Andreessen<br/>(Mosaic)"] Moz["计划项目:Mozilla<br/>(Mosaic 杀手)"] JC --> Moz MA --> Moz end subgraph NintendoSeattle["任天堂西雅图梦之队"] N64["Nintendo 64 联机项目"] idBros["'两个 John'<br/>John Carmack & John Romero"] DreamTeam["任天堂终极工程团队"] end Moz -.->|"项目流产 / 接受邀约"| N64 JC --> DreamTeam MA --> DreamTeam idBros --> DreamTeam N64 --> DreamTeam ``` --- ### [034:02] Grid Perfect 与 SAFE 加密协议的诞生 失去 Andreessen 和 Clark 的 Perfect Digital 只能另寻出路。他们顺着名单寻找,发现康奈尔大学法学院的图书馆馆长 Thomas Bruce 已经为 MP/M 和 GEM 编写了一款名为 Cello 的网格浏览器。由于 WordPerfect 本就是全美法律行业的统治级字处理软件,Perfect Digital 很快凭借良好关系敲定了收购交易,买下 Cello 的源代码。 经过底层库优化、添加功能并重新贴牌后,他们推出了与家族产品(WordPerfect、LetterPerfect、DataPerfect)一脉相承的“Grid Perfect”浏览器。更具变革意义的是,他们为原始的 GRID 协议引入了名为 SAFE(Secure Application Frame Encryption,安全应用框架加密)的技术。SAFE 实现了客户端与服务器之间的端到端加密,彻底杜绝了网络嗅探,这让此前加入协议中的 `SUBMIT` 动词大放异彩。加之 Grid Perfect 原生内置了 WordPerfect 的排版引擎,实现了“所见即所得”(WYSIWYG)的双向创作: > “Anybody who can read the grid can write the grid they have publishing capability baked in from day one so anybody who acquires a copy of this thing they have full-blown WYSIWYG authoring tools going along with it.” —— Dylan Beattie - **Grid Perfect 软件套件定位**:与 WordPerfect、LetterPerfect 深度整合,无缝打通文档阅读与编辑发布。 - **SAFE 加密协议**:全称 Secure Application Frame Encryption,允许网格客户端安全提交数据,消除中间人监听风险。 - **读写一体化设计**:打破单向只读浏览模式,任何网格文档的读者都具备直接创作并回传文档的完整工具。 --- ### [036:06] 嵌入式语言之争:HyperLisp 击败 Java 随着静态文档的局限性日益凸显,业界迫切需要一种可以同时嵌入在网格客户端与服务器中的编程语言,用于交付丰富的富客户端应用(Rich Applications)。Dylan 调侃道,在 90 年代中期,要在商业软件中嵌入语言,“唯一且显而易见的竞争者”正是拥有半个世纪历史的古典哲学利器——Lisp(“如同旧时代优雅的绝地武士光剑”)。 当时 Paul Graham 正在经营名为 Viaweb 的创业公司。Viaweb 很早就发现 Lisp 是构建网格服务器的绝佳工具:由于 GRID 协议是无状态的,而 Lisp 原生支持延续传递风格(Continuation-Passing Style),无需频繁分配共享内存与维护有状态连接,就能将处理到一半的进程拍摄快照并在日后唤醒继续运行。 此时,Sun Microsystems 也带着基于 Solaris 体系研发的“Write Once, Run Anywhere”的面向对象新宠 Java 前来推销,试图游说 Perfect Digital 采用 Java,甚至提议如果保留 Lisp 也可以改名叫“JavaLisp”或“LavaScript”。然而,在 Perfect Digital 与 Paul Graham 长达一下午的会面中,Graham 深入阐释了语言设计哲学、符号类型,以及日后成文的“Blub 悖论”(所有编程语言并非生而平等): > “Every piece of HyperLisp program code is also an expression and every expression is also a data structure the language is completely mutable you will never need to serialize or deserialize anything because you can literally take a program that is half executed turn it into a data structure pipe it over the wire to a client that will basically rehydrate it and just keep running it where it left off.” —— Dylan Beattie Graham 随即算了一笔经济账打动了财务主管:这种同构能力意味着用户运行程序时实际上是在向平台贡献自己的算力,系统架构天然具备类似 DNS 的横向扩展能力,无需堆叠高昂的服务器基础设施。Java 当场出局,HyperLisp 彻底锁定了下一代网格编程统治地位。 --- ### [039:29] HyperLisp 的运行时与异步网络代码实战 HyperLisp 最终直接内嵌在 HTML 4.01(4.0 规范中存在大量拼写错误)之中。由于采用了前缀与前置表达式语法,其事件绑定代码呈现出独特的形式: ```lisp (bind (load document) (lambda (element event) (alert "Hello, World!"))) ``` 与复杂的文档对象模型不同,HyperLisp 依托符号(Symbols)直接将运行时映射至底层,省去了多余的命名空间与限定修饰。随后 Dylan 展示了一个处理异步表单交互的复杂实例: ```lisp (bind (submit contact-form) (lambda (form event) (grid-submit (get-action form) (lambda (response) (set-value! result-span response))))) ``` 当触发表单提交时,内建的网络引擎 `grid-submit` 会异步向服务器发送回调请求。一旦服务器响应返回,由于前置闭包设计,回调会自动将结果写入对应的页面节点。Dylan 辛辣地讽刺道: > “And so this is a distributed client server what in a parallel universe they take 10 years to invent and they'd call it Ajax and it would be horrible this comes out of the box this is free.” —— Dylan Beattie HyperLisp 原生带来的语言设计红利彻底规避了后世 Web 的诸多折磨: - **符号类型(Symbol Types)**:浏览器核心运行时与 DOM 节点直接映射为 Lisp 符号,免去冗长繁琐的 DOM API 调用与复杂的属性命名空间。 - **天然支持中划线(Hyphen-case)**:由于算术运算符全部采用前缀符号形式(如 `(+ a b)`),减号不会与变量名混淆,彻底终结了“驼峰命名(camelCase)还是下划线(snake_case)”的无休止圣战,变量统统优雅地使用 `kebab-case`。 - **内建异步模型**:基于延续与 Lambda 回调的异步通信机制完全开箱即用,消除了回调地狱与复杂的序列化协议。 ### [41:37](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=2497s) Perfect Digital 的巅峰与 GEM 95 盛大发布 在那个世界里,简单的编程模型才刚展现出它的巨大潜力,伴随着号角长鸣与万众瞩目,Perfect Digital 正式发布了他们有史以来最强大的桌面操作系统——GEM 95。为了这场盛大的发布派对,他们甚至把披头士乐队(The Beatles)重新召集到了一起。GEM 95 从底层就深度集成了 Grid 浏览器,并且捆绑附送了一款剥离了打印机支持的免费版 WordPerfect,让所有用户都能随心所欲地撰写文档并直接发布到 Grid 网络上。 这场发布会被办成了一场全民狂欢,披头士乐队之所以愿意“重聚”(Get back),正是因为 Perfect Digital 将浏览器上的“后退”按钮(Back button)铺天盖地用在了所有的宣传文案中。人人都为此疯狂,整个计算机行业的前景一片光明——然而,苹果公司(Apple)除外。当时的苹果早已把史蒂夫·乔布斯(Steve Jobs)扫地出门,如今处境却一落千丈,从糟糕变成极度糟糕,再跌落至彻底的深渊,高层们只能在泥潭中绝望挣扎,急切地寻找破局之道。他们需要全新的路线、全新的操作系统和全新的品牌形象,彻底重构整间公司。当时的苹果首席执行官吉尔·阿梅里奥(Gil Amelio)在一次董事会上小心翼翼地提议,或许可以尝试收购 NeXT 并请乔布斯回归,结果却被董事会当场哄堂大笑扫地出门。Dylan Beattie 调侃道:“他们把他笑出了会议室,这简直是最愚蠢的想法,在被苹果那样无情地踹出大门之后,史蒂夫·乔布斯绝不可能再回苹果。” - GEM 95 实现了操作系统与 Grid 浏览器的深度原生绑定。 - 附带的免费 WordPerfect 特意取消了打印支持,纯粹为了驱动用户创作 Grid 线上文档。 - 苹果公司陷入经营危机,董事会断然否决了重新接纳乔布斯的提议。 --- ### [43:04](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=2584s) 苹果收购 BeOS 与“香蕉机”的诞生 乔布斯并不是唯一离开苹果的有远见卓识的工程师。当时硅谷还有一家初创公司 Be Incorporated,他们开发了一款名为 BeOS 的全新高性能操作系统。BeOS 拥有专用的多线程支持,能够完美处理流媒体播放、实时通信和音频处理等严苛任务。然而 Be 公司的团队渐渐发现,系统固然出色,但他们自己制造的硬件主机却难以在市场上立足。于是他们果断调整战略,决定将 BeOS 移植到现有的硬件平台上;鉴于核心工程师全都拥有苹果背景,苹果硬件自然成了第一选择。双方一拍即合,苹果的管理层仿佛看到了救命稻草,认为这正是苦苦寻觅的救世之作,毫不犹豫全盘接纳。 苹果彻底推翻了以往沉闷的设计规范。在随后的新品线上,他们一改整个行业非灰即米黄的传统机箱风格,大胆采用了亮黄色,媒体立刻将其戏称为“香蕉机”(Banana boxes)。这种惊艳的设计在市场上掀起狂潮,人们争相把它搬进客厅。但在万事俱备之下,他们唯独缺少一个趁手的 Grid 浏览器。苹果四处求购,先打电话给 Cello 团队,对方遗憾地表示已经和 Perfect Digital 签了排他协议;打给 NCSA,得到的答复是 Mosaic 项目基本处于开源半瘫痪状态;最后,他们出人意料地联系上了挪威国家电信公司 Telenor。Telenor 惊讶地表示自家实验室正巧有一个名为 Opera 的浏览器项目,原本还不知该如何处置。苹果与 BeOS 团队随即买下 Opera 的源码、Presto 渲染引擎以及底层网络栈,将其命名为 BeLink。不过在拼写自动更正的调戏下,大家都叫它 Blink。至此,支持全部 CRUD(Create、Update、Delete、Select、Submit)协议并自带所见即所得编辑器的 Blink 面世,苹果重新拥有了正面对抗 Perfect Digital 那些米黄色办公克隆机的底气。 ```mermaid flowchart TD accTitle: 苹果操作系统与浏览器救局路线 accDescr: 展示苹果在拒绝乔布斯后,通过收购BeOS与挪威电信Opera重组软件生态的流程 subgraph OS_Acquisition["操作系统转型"] A["苹果深陷泥潭<br/>(拒绝乔布斯)"] --> B["接触 Be Incorporated"] B --> C["收购多线程系统 BeOS"] C --> D["推出黄色'香蕉机'<br/>(Banana Boxes)"] end subgraph Browser_Acquisition["浏览器补全"] D --> E{"寻找 Grid 浏览器"} E -->|"被独占"| F["Cello (已被 Perfect Digital 拿下)"] E -->|"已停滞"| G["NCSA Mosaic (开源苟延残喘)"] E -->|"达成交易"| H["挪威电信 Telenor (Opera / Presto)"] H --> I["打造 BeLink 浏览器<br/>(自动更正为 Blink)"] end ``` - BeOS 拥有面向多媒体与低延迟通信的前瞻性微内核多线程架构。 - 苹果香蕉机打破了 PC 领域长期垄断的“米黄色审美”,成为消费级热销产品。 - 苹果收购 Opera 资产后改名为 BeLink,与 Blink 渲染器的谐音梗形成了巧妙的历史错位呼应。 --- ### [45:39](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=2739s) 信息的洪流与 Stanford 车库里的 Goggle 随着支持 Grid 的个人电脑进入千家万户,大众开始疯狂编写并上传自己的网页。每个拨号上网的用户都能分到 5MB 的免费存储空间。在这个没有被发明出臃肿 JavaScript 框架的平行宇宙中,5MB 的容量巨大无比,足以容纳海量的内容。但硬币的反面是垃圾信息的大爆发:所有人都坚信只要把各种粗制滥造的内容扔到网上吸引到流量,未来总有办法把流量变现。一时间,如何在浩如烟海的废品中检索出有价值的信息,成了 Grid 技术演进中最核心的未解难题。 在斯坦福大学,两名博士生塞吉·佩奇(Sergey Page)与拉里·布林(Larry Brin)正挤在一间车库里研发名为 BackRub 的反向链接算法。该算法的核心逻辑非常优美:一个网页的权威性取决于有多少其他网页链接到它;被重要网页链接的页面,其自身的重要性也会水涨船高。在商讨算法名字时,塞吉提到自家孩子在学校学到了一个代表极大数值的名词“Googol”(1 后面跟随 100 个零),认为用来形容海量索引非常贴切,顺便可以替换掉听起来有点奇怪的“BackRub”。然而在指派专人去注册域名时,自动更正再次作祟,最终注册成功的域名变成了 goggle.com。 Goggle 拥有当时最为干净、纯粹的极简首页,与那些充斥着横幅广告的混乱门户形成了鲜明反差,技术效果也极度惊艳。但他们面临着一个致命的尴尬:完全不知道该如何赚钱。整个服务运作高效、界面极简且向公众免费开放,却无法产生任何收入。当有人提议投放网络广告时,两人对此嗤之以鼻。就在一筹莫展之际,拉里·布林接到了一个神秘电话——电话那头是史蒂夫·乔布斯。乔布斯透露自己正在秘密筹备一个惊天项目,对收购 Goggle 拥有的 PageRank 技术表现出了浓厚兴趣。 - 缺乏前端庞大框架的时代,5MB 免费空间足以搭建内容极其庞大的全功能站点。 - BackRub 算法通过网页之间的相互链接权重,确立了全新的可信度评分模型。 - Goggle 坚持极简主义与拒绝广告的商业取向,导致其深陷无法盈利的技术困境。 --- ### [48:15](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=2895s) 1999 年 5 月 4 日:NeXTBook 与惊世发布会 1999 年 4 月,整个科技界都被千年虫(Y2K bug)末日论的恐慌所笼罩,媒体成天渲染核弹因为无法识别四位年份代码而会在午夜失控爆炸的末日预言。就在这片喧嚣中,硅谷的所有知名科技记者突然打破了常规的工作节奏——每个人的办公桌上都被专人送来了一封黑色信封。拆开信封,里面躺着一张印着烫黑字样的卡片,正面写着冷峻的一行字:“为 NeXT 时代做好准备”(Get ready for the NeXT generation);背面则标注着时间和地点:1999 年 5 月 4 日,门洛帕克露天剧场(Menlo Park amphitheater)。记者们的第一反应是这或许是卢卡斯影业针对《星球大战》(Star Wars)搞的宣发噱头(取自“愿原力与你同在”的同音梗 May the 4th be with you),但卢卡斯影业在电话里坚决否认:“别开玩笑了,谁会去给《星球大战》拍续集啊?我们才不会毁掉完美的前作呢。” 到了 5 月 4 日清晨,门洛帕克剧场座无虚席。上午 9 点整,全场瞬间鸦雀无声,史蒂夫·乔布斯大步走上舞台。他畅谈了 Grid 的愿景、数字出版的变革、技术赋予个体的力量以及科技对人类社会产生的深刻冲击。就在大家猜测他的落脚点时,乔布斯从讲台后方缓缓抽出了一台让全场屏住呼吸的设备——这台被命名为 NeXTBook 的笔记本电脑,颠覆了当时所有人的认知。 整台机器厚度不足一英寸,机身由一整块铝合金整体切削而成,表面经过阳极氧化呈现出深邃的哑光黑色。令人瞠目结舌的是,它上面没有软盘驱动器,没有光驱,甚至没有任何传统的串行端口。面对全场记者关于“连软驱都没有,我们怎么插打印机、怎么安装《模拟城市》(SimCity)”的错愕质问,乔布斯展现出了极具戏剧性的杀手锏。观众席中早早安插了大约两打 NeXT 员工,他们同时从座椅下拉出 NeXTBook 并掀开屏幕——屏幕上正流畅地实时播放着舞台上乔布斯演讲的现场视频直播! 在场的所有记者瞬间陷入失控。在那个 Wi-Fi 概念从未普及、在线流媒体如同科幻、笔记本笨重如砖头的年代,在场众人彻底忘却了自己的报道职责,有人甚至为了抢占前排视野在座椅间推搡跨越。Dylan Beattie 描述道:“没人见过 Wi-Fi,没人见过流媒体视频,也没人见过这样的笔记本电脑。他们完全忘记了自己是来写报道的记者,大家为了看一眼这玩意儿甚至互相用手肘互殴、爬过前排座椅。”这台被誉为世界最快、最轻薄的笔记本电脑,搭载了 333MHz 的 ARM RISC 处理器、64MB 内存、5GB 闪存空间,运行着新一代 NeXTSTEP 操作系统,全机内置无线网络功能,并随机附带了一个名为 NeXT Antenna、大小宛如一本精装硬皮书的无线发射基站。 | 规格维度 | NeXTBook 工业配置与设计特性 | |:--- |:--- | | **外观工艺** | 一体成型切削铝合金(Unibody),黑色阳极氧化处理,厚度小于 1 英寸 | | **处理器与架构**| 333 MHz ARM RISC 处理器 | | **存储配置** | 64 MB 运行内存,5 GB 固态闪存空间(Flash Drive) | | **外设接口** | 舍弃软驱、CD-ROM、串口;仅配备 2 个 USB 接口与 1 个耳机插孔 | | **网络与生态** | 内置无线网卡,搭载新版 NeXTSTEP 系统,附带 NeXT Antenna 无线基站 | ### [51:07](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3067s) 平行时空中的 NeXTBook、Nexus 浏览器与千禧年格局 Dylan Beattie 继续描绘那个充满幻想的平行时间线:在这条世界线中,NeXT 推出了创新的硬件设备 NeXTBook。用户只需把它插进电话插孔,它内置了 CD-ROM 驱动器,甚至还自带为全家提供信号的无线网络系统,让人们再也不用蹲在电话插口旁办公。该设备搭载了全新的网格(Grid)浏览器 Nexus。它的基因可直接追溯至十年前蒂姆·伯纳斯-李(Tim Berners-Lee)在 NeXT 操作系统上开发的第一个应用程序。Nexus 响应极快,内置了有史以来最快的 HyperLisp 解释器,与底层系统紧密融合,并无缝集成了 Goggle 的搜索技术。无论是个人文件夹、本地备份、邮件还是整个网络网格,搜索窗口都能瞬间定位。史蒂夫·乔布斯(Steve Jobs)借此华丽回归,下一代计算时代正式揭幕。 千禧年危机(千年虫)平稳度过,导弹没有误射,世界也没有毁灭,大家只是在宿醉中醒来。此时的计算格局由四股力量瓜分:乔布斯的 NeXT 公司凭借搭载 Nexus 的 NeXTBook 引领潮流;黄色的 BeBox(运行 BeOS)配备了从挪威电信公司 Telenor 诞生的 Blink 浏览器;Digital Research 则掌控着 Gen 95 与 Perfect Digital 体系,其浏览器血统可追溯至法学院图书馆员开发的 Cello;而 Sun Microsystems 依然在角落坚挺,苦苦推销着支持 Java 的 NCSA Mosaic,“虽然根本没人关心”,但凭借庞大的市场预算仍持续发力。 ```mermaid flowchart TD accTitle: 假想平行时间线中的四大计算阵营 accDescr: 展示 Dylan Beattie 虚构的千禧年四方计算势力分布及其核心软硬件。 subgraph NeXT["NeXT 阵营"] A["NeXTBook 硬件"] --> B["Nexus 浏览器"] B --> C["HyperLisp 解释器 & Goggle 搜索"] end subgraph Alternative["其他竞争阵营"] D["BeBox 硬件 (BeOS)"] --> E["Blink 浏览器 (来自 Telenor)"] F["Gen 95 / Perfect Digital"] --> G["Cello 浏览器 (法学院血统)"] H["Sun Microsystems"] --> I["NCSA Mosaic (主推无人问津的 Java)"] end ``` ### [52:46](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3166s) 任天堂与“Reality”:元宇宙的提前降临 在这条时间线的两三年后,任天堂终于放出了大招。自从约翰·卡马克(John Carmack)、约翰·罗梅洛(John Romero)、吉姆·克拉克(Jim Clark)和马克·安德森(Marc Andreessen)齐聚西雅图,任天堂便全额资助他们的所有研发。在交付了一款极为成功的《超级马力欧雷神之锤世界》(Super Mario Quake World)后,该团队沉寂了一段时间。直到 2002 年 12 月圣诞前夕,任天堂正式公布了一款意图颠覆世界的平台——“Reality”。 “Reality” 是一款大型多人在线 3D 沉浸式虚拟世界,瞬间风靡全球。它直接实现了尼尔·斯蒂芬森(Neal Stephenson)在小说中描绘的“超元域”(Metaverse)以及威廉·吉布森(William Gibson)《神经漫游者》中的赛博空间(Matrix)。这套系统的分工极为梦幻:图形底层渲染、可视化与界面交互由卡马克与罗梅洛一手打造;底层高并发网络架构出自马克·安德森之手;而驱动整个虚拟世界的 3D 引擎则继承了硅图公司(Silicon Graphics, SGI)的纯正基因。Dylan 打趣地向台下发问:“现在很难想象没有它的生活对吧?你能想象一个没有 Blink、没有 HyperLisp、没有 BeBox 和 NeXTBook 的世界吗?”观众的笑声揭示了这场脑洞推演的荒诞与讽刺。 ### [54:09](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3249s) 历史真相(一):IBM 的敲门与万维网的双斜线之悔 随后,Dylan 将话题拉回现实,逐一剖析这些平行幻想背后真实的历史节点。第一处关键分岔点是 1981 年 IBM 为其初代个人电脑寻觅操作系统: - **加里·基尔代尔错失历史**:IBM 最早登门拜访的是 Digital Research 的创始人加里·基尔代尔(Gary Kildall),希望为其 PC 授权 CP/M 系统。传奇版本称加里那天因为天气晴朗开飞机兜风去了;另一个不那么浪漫的版本则是他根本不待见 IBM 并直接让他们走人。IBM 转而敲响了名单上的下一位——年轻的比尔·盖茨(William H. Gates III)。盖茨紧紧抓住了这次机会,成就了微软帝国的霸业。 - **伯纳斯-李的“双斜线”遗憾**:Dylan 引用了对万维网发明者蒂姆·伯纳斯-李的访谈。当被问及若重来一次会改变什么时,伯纳斯-李直言:“事实证明,根本没有必要保留那两个斜杠。”(“It turns out really if you think about it, you don't need this double slash. I could have designed it to not have the double slash.”)当初设计 `http://` 后的双斜线只是为了标记顶层路径,但人们在日常引用时总是连同协议一起输入,甚至在广播里被误念为“反斜杠反斜杠”(backslash backslash)。伯纳斯-李感慨,如果当初去掉这两个字符,全世界不知能节省多少打字的时间和纸张油墨。 ### [56:40](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3400s) 历史真相(二):标记语言、Lisp 遗风与被扼杀的 PC 版马力欧 现实中的很多技术选择并非最优解,而是偶然与妥协的产物: - **HTML 与标记语言的选择**:HTML 基于 SGML 继承了尖括号语法,但当时早已有 1970 年代初诞生的 LaTeX,Sun 公司内部也有大批 LaTeX 专家;微软自身则拥有富文本格式(RTF)。HTML 完全有可能基于这些不同的语法演化。 - **浏览器中的 Lisp 执念**:用 Lisp 构建 Web 并非空想。保罗·格雷厄姆(Paul Graham)创办的 Viaweb 当年正是用 Lisp 编写服务器端软件,最终卖给雅虎实现财富自由,并借此资本创办了著名的 Y Combinator。同时,AutoCAD 和 Emacs 都内置了成熟的 Lisp。网景在招募布兰登·艾克(Brendan Eich)时,原本初衷正是让他将 Scheme(一种 Lisp 方言)塞进浏览器;但迫于紧迫的项目工期和市场营销部门“语法如果不像 C 语言就会吓跑程序员”的恐慌,最终在极短时间内拼凑成了 JavaScript。 - **卡马克与任天堂的版权警告**:卡马克与罗梅洛确实曾将《超级马力欧兄弟 3》完美移植到 IBM PC 上,并满怀期待地寄了一张软盘给任天堂。然而任天堂给出的回应不是工作邀约,而是严厉警告:如果敢公开发布,法务部的起诉将让你们粉身碎骨。任天堂完全依赖卡带硬件这一极其有效的物理防盗版机制,对开放的 PC 平台毫无兴趣。 ### [58:25](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3505s) 历史真相(三):被错过的收购与“刚刚好”的技术现实 在资本运作与商业决策层面,历史同样充满了极具戏剧性的擦肩而过: - **网景的另一条出路**:吉姆·克拉克与马克·安德森确实曾无限接近与任天堂签约联合构建游戏网络系统。如果协议达成,网景公司根本不会诞生,马克·安德森也不会光脚登上《时代》周刊封面;这完全取决于顶尖技术人才在某一天对“哪个方向看起来更好玩”的随性抉择。 - **操作系统的生死劫**:Cello 浏览器确实存在,且是由康奈尔大学法学院的图书馆员托马斯·布鲁斯(Thomas Bruce)针对 Windows 平台编写的。BeOS 也是由前苹果高管创立的真实操作系统,前苹果 CEO 吉尔·阿梅里奥(Gil Amelio)在收购 NeXT 之前确实曾认真与 BeOS 团队接洽过收购事宜。如果当时苹果收购了 BeOS 而非 NeXT,史蒂夫·乔布斯就不会回归,iMac、iPod、iPhone 都不会出现,苹果也不可能从濒临破产逆袭成万亿美元市值的科技巨头。 - **Excite 错过 80 万美元的 Google**:在缺乏清晰盈利模式的早期阶段,Google 曾试图以区区 80 万美元的价格将自身技术出售给当时的门户与搜索巨头 Excite,但 Excite 拒绝了,理由是认为这家公司成不了大气候。如今 Google 已经成为庞然大物,而 Excite 却早已消失在公众记忆之外。 在演讲的结尾,Dylan Beattie 阐述了这趟历史回溯的核心启示: > “我们每天使用并依赖的许多系统,之所以存在,并不是因为有人历尽千辛万苦找到了最优解决方案、做出了最完美的决定,或是找到了最出色的团队与环境来合作;它们的存在,仅仅是因为人们找到了一种‘刚刚好够用’的东西,并在此基础上思考‘我们接下来能用它做什么’。” > (“So many of these systems that we use and we rely on every day are not there because somebody worked really really hard to find the best solution or to make the best decision or to find the best possible team or circumstances to work with. They're there because people found something that was just good enough and they thought about what can we do with it, what's gonna be next.”) Dylan 提醒技术人员,行业中最容易陷入的陷阱就是“分析瘫痪”(analysis paralysis)——面对无数可选平台和工具犹豫不决、无从下手。然而纵观技术史,今天繁荣的整个 Web 开发生态,其基石正是建立在一个个当初可能根本不会诞生的偶然产物之上。 ### [01:01:05](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3665s) 历史的分岔路与“未曾诞生的万维网” Dylan Beattie 在演讲尾声回顾了现代 Web 发展史上那些充满偶然与仓促的技术决策——设想一下,如果当年关键的人物跑去搭建游戏网络,或者整个行业并非依赖一种由开发者在来自 Sun Microsystems 的市场营销压力下短短十天内拼凑出来的编程语言(即 JavaScript),世界将会怎样。然而他指出,回望历史,技术演进往往并非取决于抉择的绝对对错,而是取决于手头拥有的方案是否足够支撑人们继续前行与探索。 正如 Dylan Beattie 所总结的那样,哪怕没有绝对完美的决定,技术世界依然在跌跌撞撞中前行:“甚至即使你没有做对那些决定,其实也没有真正的对与错,关键只在于我们是否拥有足够的基础继续走下去并取得进展;正如你们在这次演讲中所见,我们本可能走向一种截然不同的发展轨迹。”(“Even if you don't get those decisions right, there isn't really a right and wrong, all there is is have we got enough to go on and make progress, and as you see in this talk, we could have ended up making a very, very different kind of progress.”) 演讲最后,Dylan Beattie 将听众从平行的虚构历史拉回现实。也许未来某天技术的演化依然会迎来意外的转向,但在当下的真实世界里,大家依然可以安心享受 Chrome、Firefox、Safari 以及 Microsoft Edge 构筑的现代网络生态。他向全场观众致意,并以这句点题之语正式结束了整场演讲:“这就是那个未曾诞生的万维网,谢谢大家!” - **JavaScript 的诞生背景**:提及了当年在 Sun Microsystems 的商业和市场营销压力下,Brendan Eich 仅用十天仓促拼凑出该语言的历史轶事。 - **技术进化的本质**:强调技术路线没有先验的“正确与错误”,关键在于现有方案是否足以推动行业形成共识并向前发展。 - **现实世界的浏览器格局**:列举了当下现实世界中主流的几大浏览器产品,包括 Google Chrome、Mozilla Firefox、Apple Safari 以及 Microsoft Edge。 ## Notable quotes > [07:28](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=448s) it's easy to think that there is some sort of Darwinian survival of the fittest... but that is absolutely not true. — Dylan Beattie > [00:06](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=6s) Do not read the comments on YouTube they will only make you feel angry. — Dylan Beattie > [54:09](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=3249s) It turns out really if you think about it, you don't need this double slash. I could have designed it to not have the double slash. — Tim Berners-Lee > [27:39](https://www.youtube.com/watch?v=8JOD1AQGqEg&t=1659s) 如果你们的旗舰电脑全是用从中国大陆、新加坡和台湾买来的现成现货零部件组装而成的,那么你们其实什么技术资产都没有拥有。运行应用程序的关键早就不再是硬件本身——因为硬件谁都能买到;掌控应用程序生态的关键,必将是操作系统。 — Gary Kildall ## Mentioned Dylan Beattie, Douglas Adams, IBM, Digital Research, Gary Kildall, Paul Mockapetris, Tim Berners-Lee, CERN, Microcosm, Steve Jobs, Apple, NeXT, WordPerfect, Jim Clark, Marc Andreessen, Nintendo, John Carmack, John Romero, Paul Graham, Sun Microsystems, Gil Amelio, Be Incorporated, Telenor, Sergey Brin, Larry Page, Excite, Brendan Eich, Bill Gates, Microsoft, Google, ARPANET, TCP/IP, DNS, HTML, HTTP, Mosaic, BeOS, JavaScript, Lisp ## Video description The story of the web is a story about freedom. It's a story about information, about breaking down barriers, about creating new ways for people to communicate, to collaborate, and to share their ideas. It’s also a story that has as much do with marketing, money and meetings as it does with research and innovation. It’s a story of mediocre ideas that succeeded where brilliant ideas failed, a story of compromises, rushed deadlines and last-minute decisions. And it could so easily have been very, very different. What if IBM had hired Digital Research instead of Microsoft to develop the operating system for their first PC, way back in 1980? What if Marc Andreessen and Jim Clark had gone to work for Nintendo in 1993 and never founded Netscape? What if one of the team at CERN had said “Tim, won’t it sound a bit silly if everyone spends the next fifty years saying double-you-double-you-double-you all the time”? In this talk, Dylan Beattie will explore alternative history of the world wide web - a web with no Microsoft, no Windows; no Firefox, no Google and no JavaScript. A software industry from another timeline, a world of platforms, protocols and programming languages that are unmistakably alien - and yet strangely familiar. So strap in, hold tight, and join us as take you on a journey through... the web that never was. NDC Conferences https://ndcsydney.com https://ndcconferences.com [Source](https://www.youtube.com/watch?v=8JOD1AQGqEg)

Source quotes

[7:28] Dylan Beattie: it's easy to think that there is some sort of Darwinian survival of the fittest... but that is absolutely not true. [0:06] Dylan Beattie: Do not read the comments on YouTube they will only make you feel angry. [54:09] Tim Berners-Lee: It turns out really if you think about it, you don't need this double slash. I could have designed it to not have the double slash. [27:39] Gary Kildall: 如果你们的旗舰电脑全是用从中国大陆、新加坡和台湾买来的现成现货零部件组装而成的,那么你们其实什么技术资产都没有拥有。运行应用程序的关键早就不再是硬件本身——因为硬件谁都能买到;掌控应用程序生态的关键,必将是操作系统。

Source quotes

[0:06] Dylan Beattie: Don't worry about changing database providers or worry, but know that every company who used an ORM in case they needed to switch databases never actually did it. [27:36] Dylan Beattie: What Gary Kildall has realized and IBM haven't is that if your flagship computer is made out of off-the-shelf parts that you bought from China and Singapore and Taiwan, you don't actually own anything.

Original transcript

[0:06] ladies and gentlemen of the class of 17 [0:10] use flat screens if I could offer you [0:16] only one tip for the future flat screens [0:20] would be it the benefits of flat screens [0:24] have been proved in Hollywood whereas [0:27] the rest of my advice has no basis more [0:30] reliable than my own meandering [0:33] experience I will dispense this advice [0:37] now enjoy the confidence and optimism of [0:43] Greenfield projects [0:45] oh never mind you'll not appreciate the [0:47] confidence and optimism of Greenfield [0:49] till things start going wrong but trust [0:52] me when you finally ship you'll look [0:55] back at the code you wrote and recall in [0:58] a way you can't grasp now how simple [1:00] everything seemed and how productive you [1:03] really were your code is not as bad as [1:07] you imagined don't worry about changing [1:11] database providers or worry but know [1:14] that every company who used an ORM in [1:16] case they needed to switch databases [1:18] never actually did it the real problems [1:21] in your projects are the dependencies [1:23] you can't control the engineer who [1:25] unexpectedly leaves your team at 4:00 [1:28] p.m. on some idle Tuesday learn one [1:31] thing every day that scares you [1:34] optimize don't reformat other people's [1:37] code bases don't put up with people to [1:40] reformat yours rebase don't get obsessed [1:44] with frameworks sometimes they help [1:47] sometimes they hurt it's the user [1:49] experience that matters and the user [1:52] doesn't care how you created it remember [1:56] the retweets you receive forget the [1:59] flame bait if you succeed in doing this [2:01] tell me how keep your old hard drives [2:06] throw away your old network cards [2:10] refactor don't feel guilty if you don't [2:14] understand f-sharp some of the most [2:16] productive junior developers I've worked [2:18] with didn't know F sharp some of the [2:21] best systems architects I know still [2:23] don't write plenty of tests be kind to [2:28] your keys you'll miss them when they're [2:31] gone maybe you have a degree maybe you [2:34] don't maybe you have an open source [2:36] project maybe you don't maybe you wrote [2:39] code that flew on the space shuttle [2:40] maybe you worked on Microsoft SharePoint [2:43] whatever you do keep improving and don't [2:47] worry where your next gigs coming from [2:48] there's a big world out there and [2:51] they're always gonna need good [2:52] developers look after your brain don't [2:56] burnout never be scared to take a break [2:58] it is the most powerful computer you [3:01] will ever own launch even if you have no [3:06] users put your own q8 I have a plan even [3:10] if you choose not to follow it do not [3:12] read the comments on YouTube they will [3:14] only make you feel angry [3:17] [Music] [3:22] cash your package dependencies you never [3:25] know when they will be gone for good [3:27] read your log files they're your best [3:30] source of information and the first [3:32] place you'll notice something starting [3:34] to go wrong understand that languages [3:40] come and go and it's the underlying [3:42] patterns that really matter [3:44] work hard to fill the gaps in your [3:46] knowledge because the wiser you get the [3:49] more you'll regret the things you didn't [3:51] know when you were young codon 86 [3:55] assembler once but stop before it makes [3:58] you smug code in Visual Basic once but [4:01] stop before it makes you stupid [4:04] read accept certain inalienable truths [4:08] your code has bugs you will miss your [4:11] deadlines and YouTube might end up in [4:14] management and when you do you'll [4:16] fantasize that back when you were a [4:18] developer code was bug free deadlines [4:22] were met and developers tune their [4:24] database indexes tune your database [4:27] indexes don't deploy your code without [4:30] testing it maybe you have a QA team [4:33] maybe you have integration tests you [4:36] never know when either one might miss [4:37] something don't best too much with your [4:41] user interface or by the time you ship [4:44] it will look like a karaoke booth [4:47] be grateful for open source but be [4:50] careful whose coach run sharing code is [4:53] noble but there's more to it than taking [4:55] bits of your projects folder putting a [4:58] readme file on them and hoping if you [5:00] upload it to github someone else will [5:02] come along to fix all your problems but [5:06] trust me on the flat screens [5:10] [Music] [5:16] good afternoon Sydney are you all having [5:20] an enlightening and educational [5:22] conference are you enjoying yourselves [5:25] are you coming to pub conf good alright [5:33] the author Douglas Adams once came up [5:37] with a set of rules that describe the [5:40] way human beings react to the technology [5:43] around them in the world he said [5:45] anything that is in the world when [5:48] you're born is normal it's just a [5:50] natural part of the way the world works [5:53] anything invented between when you're 15 [5:56] and 35 is new and revolutionary and you [6:00] can probably get a career in it anything [6:04] invented after you're 35 is against the [6:07] natural order of things [6:09] now this is me I was born in 1978 just [6:16] too late to see Star Wars the first time [6:18] around the World Wide Web [6:21] I was born into a world where ARPANET [6:23] already existed where computers already [6:25] existed where we had color TV and you [6:28] know the idea of computer networks and [6:31] things but the Internet as we sort of [6:34] recognize it now didn't really exist yet [6:38] things invented the web was kind of [6:41] invented when I was around 15 years old [6:42] and I did go on to have quite a good [6:44] career in it I started as a webmaster I [6:46] learned HTML I did you know all these [6:48] kinds of things the things that were [6:50] invented after I turned 35 and are [6:52] therefore unnatural and against the [6:55] order of things are the Apple watch [6:57] fidget spinners and something called [7:00] dabbing which I still don't know what [7:03] that is but you know the point that [7:08] Douglas Adams is making there is that [7:10] our experience of technology is [7:14] incredibly subjective so much of it has [7:17] to do with at what point do we first [7:20] encounter these devices and sis [7:23] terms and ideas that go on to change the [7:25] world around us you know and it's easy [7:29] to forget that the things that are all [7:31] around us every day these amazing pieces [7:34] of kit that we all take for granted that [7:36] just work you know another thing Douglas [7:38] Adams he said technology is stuff that [7:39] doesn't work yet when it works it's just [7:41] stuff you know like chairs don't fail [7:45] anymore because we don't think of this [7:47] so we don't think of chairs as being [7:48] technology um but you know the point [7:51] he's making all these kinds of things [7:52] that we take for granted every day all [7:54] of them started out as hacks and [7:57] prototypes and ideas and you know things [7:59] that were held together with string and [8:01] glue and code and gaffer tape and [8:03] technical debt and eventually they [8:05] mature and you know if your first [8:06] experience here is the column on the [8:08] left you probably still think of all [8:11] these things as a pretty controversial [8:13] idea and if your first experience is the [8:15] things on the right well yeah of course [8:17] we live in a world with smartphones and [8:19] electric cars why would we not you know [8:21] imagine that and it's easy to think that [8:26] there is some sort of Darwinian survival [8:30] of the fittest that we have all these [8:32] prototypes and all these ideas and all [8:34] these solutions to problems like smart [8:37] phones and digital communication and and [8:39] the best solution is going to win but [8:44] that is absolutely not true [8:46] and what I'm going to talk about today [8:49] one of the ideas behind this talk I want [8:51] to look at the points in the history of [8:53] the web where something just kind of [8:56] happened for a dumb reason and I'm gonna [8:59] tell you a story about what might have [9:01] happened if it had gone ever so slightly [9:04] differently so let's go back to the [9:08] beginning in the beginning IBM created [9:11] the mainframe and it was good a computer [9:13] so powerful that Thomas J Watson who the [9:16] the Watson AI system is now named after [9:19] thought the world might one day need as [9:22] many as five of these computers during [9:26] the second half of the 20th century [9:28] scientists and engineers took all of [9:30] this computing and information theory we [9:32] aren't from valves to transistors to [9:34] integrated circuits we went from [9:36] in code to assembler to interpreted in [9:38] compiled languages you know these [9:40] computers made their way out of [9:42] laboratories and you know companies and [9:44] universities and started making their [9:46] way into people's homes and towards the [9:49] end of the 1970s IBM dominated computing [9:53] in a way that you know you can't quite [9:56] you think of the biggest companies in [9:57] the world today you know Google and or [9:59] alphabet and Apple and Amazon all of [10:04] them you can probably think of someone [10:05] so you know Apple are dominant but other [10:07] people are still selling laptops other [10:08] people are still selling smartphones IBM [10:10] basically owned the world in the late [10:13] 1970s when it came to information [10:15] technology and they realized that there [10:19] might be an opportunity here to sell [10:21] people a computer that you could fit in [10:23] a car and take home and actually have in [10:24] your house you know turns out Tom Watson [10:27] was wrong maybe they could sell more [10:28] than five and there was an opportunity [10:31] here now IBM in the 70s was a juggernaut [10:35] you couldn't steer it it was this [10:37] massive enterprise that was incapable of [10:40] doing anything quickly because it would [10:42] take six weeks just to approve the font [10:44] for the purchase orders and so they came [10:48] up with an interesting idea they put [10:50] together a team to create the first [10:51] personal computer the PC they were [10:54] frightened of getting left behind and [10:55] they made a radical decision they said [10:57] we're not going to manufacture this [10:58] thing we I B M are gonna buy parts off [11:02] the shelf that are being made in [11:04] factories all over the world [11:05] we'll pick stuff up where we can we need [11:07] a motherboard get one of those processor [11:09] Intel have those serial bus yeah get [11:11] some of those out of Taiwan or China or [11:13] Singapore or wherever the stuff is being [11:14] manufactured and they needed an [11:17] operating system and so they went [11:20] shopping they went to see the best [11:23] software company in the world when it [11:25] came to building operating systems for [11:27] home personal computers and who was that [11:29] in 1981 we'll find out a little later [11:33] now then while IBM are working out how [11:36] to put computers in your house the [11:38] Advanced Research Projects Agency ARPA [11:41] is coming up with this crazy idea to [11:43] connect computers together and you know [11:45] they had this good idea like if we could [11:46] connect four computers together so that [11:48] if one of them [11:49] wiped out by some kind of Cold War [11:51] military attack the other still worked [11:53] that might be a good idea let's start [11:55] engineering here for resilience and [11:57] around the same time as IBM was [11:59] launching the PC ARPA published a [12:01] specification for an intern networking [12:03] protocol the standard which became known [12:06] as ipv4 was published around 1981 this [12:09] is actually a map of a map of the [12:11] internet from 1973 of every single [12:14] device that was correct connected to the [12:16] ARPANET before it became the internet [12:18] drawn by hand by a draft person with a [12:21] pen and paper that's every single device [12:23] and what they're running you know this [12:28] was published in 1981 January 1st 1983 [12:30] became known as Flag Day that was the [12:32] day they said from now on everything [12:35] connected to our network must run tcp/ip [12:38] that's it [12:39] 1982 they ratified a standard called [12:42] SMTP the Simple Mail Transfer Protocol [12:45] we suddenly had this system where all [12:48] the computers in the world could talk to [12:49] each other and a system where the people [12:51] using those computers could send mail to [12:53] one another now the internet started to [12:57] grow and the problem with systems that [13:00] are growing is you need a way of keeping [13:02] track of them now there are only two [13:04] really hard problems in computer science [13:06] one of them is cache invalidation the [13:08] other one is naming things so let's talk [13:10] about DNS because before DNS was [13:15] invented if you set up a new server and [13:18] you put it on the Internet you would [13:20] plug it in and then you would phone Jake [13:22] and you would say Jake I got a new [13:25] computer can you add it to the hosts [13:27] file please this was the hosts file it [13:31] was a file that lived at the network [13:33] information center at Stanford and when [13:36] you know one of your team was going [13:37] there for a meeting you'd say how can [13:38] you grab a copy of the hosts file and [13:40] bring it back oh you'd FTP it overnight [13:42] when the phone calls were cheap or you [13:44] know you'd get them to send you a copy [13:45] on a tape or something but this was how [13:48] organizations managed their list of [13:51] hosts that were connected to the [13:52] internet with this one file if you [13:53] wanted to add a machine to the hosts [13:55] file on a public holiday you couldn't [13:58] because the people who had permission to [14:00] edit the file we're all off celebrating [14:01] with the [14:02] you couldn't add machines at night or on [14:04] the weekend or at Thanksgiving or at [14:06] Christmas and it kind of worked for a [14:08] while but it was obvious that this [14:10] internet thing was likely to start [14:12] growing a little bit and maybe we needed [14:14] a better system so enter this guy ass [14:19] poor maka Petrus from the University of [14:22] Southern California which also invented [14:24] UNIX and LSD so thank you [14:27] University of SoCal for those [14:28] contributions to society and they hired [14:31] him and they said look we need a better [14:32] thing better solution to this host [14:34] system we got these five different [14:35] engineering solutions could you look at [14:37] them all and you know just just pick the [14:39] best ones and keep everyone happy now I [14:41] quite like what he did next cuz I think [14:44] it's what I would have done next he [14:45] looked at all five of these solutions [14:46] went these are all crap and threw them [14:48] away and he designed and invented by [14:51] himself [14:52] DNS a hierarchical database of host [14:56] names and Network addresses which has [14:58] the fantastic quality you know it has [14:59] survived pretty much unchanged from his [15:02] original specification because he [15:05] designed it to scale with the complexity [15:07] of the network it was supporting the [15:09] more hosts you connect the more DNS [15:11] servers you've got the more you can [15:13] share the load the more stuff gets [15:14] cached the fewer root level lookups [15:16] actually have to happen to make that [15:18] system work and you know that still [15:20] survives today supporting hundreds of [15:22] millions of domains now so the scene is [15:26] set tcp/ip exists DNS exists smtp exists [15:31] we've got conventions and protocols [15:34] computer hardware you know commodity [15:35] hardware is cheap lots of people have it [15:37] and in cern on the border of France and [15:41] Switzerland this guy Tim berners-lee web [15:45] developer is he actually he uses this [15:49] job title that that's the real title [15:51] that Tim berners-lee uses and I love [15:53] Scott Hanselman Slyke he could have been [15:55] the web developer but he's like no web [15:57] developers fine and he's working at CERN [16:02] and you know he's noticed they have [16:04] these fantastic computer networks now [16:06] but most of his colleagues who are [16:08] physicists and mathematicians if they [16:10] want to publish their work so someone [16:12] else can read it they don't really have [16:15] kind of digital equivalent of journals [16:17] and he thinks you know what if we could [16:20] come up with a way of publishing [16:21] material on online on the network so he [16:25] writes up a proposal information [16:27] management Tim berners-lee and he sends [16:29] it to his supervisor who sends it back [16:31] with a three word comment vague but [16:34] exciting now it is really easy in [16:39] hindsight to think well obviously that's [16:41] the world wide web that's how it was [16:43] gonna end up you know this idea of HTML [16:45] and HTTP and links and all this kind of [16:47] stuff um I was studying at the [16:49] University of Southampton in 1997 and [16:52] one of my professors actually two of my [16:54] professors there had worked on a [16:56] competing submission for hypermedia [16:58] before the world wide web was invented [17:00] there was this big international hyper [17:02] media conference and one of my [17:05] professors he said I was sat there and I [17:07] got up and I showed them my system and [17:08] then the next guy who got up was this [17:10] chap called Tim who had come over from [17:11] CERN and I sat there and I thought yeah [17:14] we're not gonna get this one um but the [17:16] system that the the Southampton team had [17:18] was a thing called microcosm and this is [17:20] actually a screenshot of a running [17:21] microcosm system on running on Windows [17:24] 3.1 and microcosm didn't use embedded [17:28] hyper media because their philosophy is [17:30] we don't want people to have to edit [17:31] their documents because these documents [17:33] already exist we don't have to rewrite [17:34] everything what we'll do is we'll invent [17:37] a network server that can maintain a [17:39] database of links independently of the [17:42] material that's being linked so you can [17:45] put your you know your text files and [17:47] your rich text files and your [17:48] WordPerfect files and whatever on a [17:50] network file share and we will run a [17:52] indexing service over the top of them [17:54] that doesn't touch the original [17:55] documents but it looks for keywords and [17:57] contexts and it links things into other [17:59] things and it shows you all of the links [18:01] here in a separate window running [18:03] alongside your content so that you're [18:05] always presenting the original document [18:07] the way the author intended it to be [18:08] done and the hypermedia is being applied [18:10] by you know librarians and information [18:13] analysts and you know it's fascinating [18:15] just to see this now has survived [18:17] microcosm became multi cousin PLC who [18:20] now do a very good line doing discovery [18:23] and you know big lawsuits where they've [18:25] got all these documents and files on [18:27] some companies subpoenaed serve [18:29] and they use a variation of that tech to [18:31] this day to go through an index all that [18:33] material um but you know this is a [18:36] completely different philosophy on how [18:39] online hypermedia [18:40] might have worked but you won't lock [18:53] this door with the key of imagination [18:55] beyond it is another dimension a [18:57] dimension of sound a dimension of sight [19:01] a dimension of mind [19:04] you're moving into a land of both shadow [19:06] and substance of things and ideas you [19:09] just crossed over into the twilight zone [19:12] [Music] [19:14] [Applause] [19:17] we're at Sun it's the early 1990s Tim [19:21] berners-lee is in his office he's been [19:23] playing around with something called the [19:24] mesh and a couple of other people have [19:27] also got projects that are related to [19:28] mesh systems and things and he's [19:31] thinking he needs a new name for it and [19:32] he has this idea he thinks what if I [19:34] call it the World Wide Web and he's [19:37] scribbling www on pieces of paper and [19:39] ideas and everything and then something [19:41] fortuitous happens one day the network [19:43] goes down and he has to tell someone [19:46] what he's doing [19:46] over the phone and he picks the phone up [19:49] because when he goes yeah it's this this [19:50] www and they say hang on a second [19:53] that acronym is actually longer than the [19:55] thing it's short for this is gonna if [19:58] this catches on it it might sound a [19:59] little bit daft and Tim thinks you know [20:01] that's a good point so he calls his [20:04] colleagues in when they're all having [20:05] tea at four o'clock in the afternoon and [20:07] says look I got this thing and I need [20:09] some names for it what do you think and [20:10] someone is yeah well World Wide Web [20:12] yeah I see the problem with that one [20:13] what if we call it the International [20:15] information interchange the ìiî and so [20:18] Leo's name sounds like pirates and [20:19] someone who's just read Neuromancer says [20:21] what about the multinational metadata [20:23] matrix and they think oh yeah that [20:26] sounds like it could have potential [20:27] they'd play around with these ideas for [20:29] a little while eventually he comes up [20:30] with the global repository of [20:32] interlinked documents the grid he's got [20:37] a name [20:37] he needs a protocol protocols are always [20:40] called to something something something [20:41] Transfer Protocol File Transfer Protocol [20:44] Mail Transfer Protocol [20:45] he's like I'm slinging hypertext here [20:47] hypertext Transfer Protocol and he looks [20:49] around things well there's a couple of [20:51] basic operations for you know publishing [20:53] and updating documents and he looks at [20:55] structured query language SQL and he [20:58] thinks you know we can borrow some ideas [20:59] from that we'll have a verb in our thing [21:02] select means select that document from [21:04] that server bring it back to me create I [21:06] want to publish something on the grid [21:08] update I made a mistake please take that [21:11] document and update it with these [21:12] changes [21:13] delete means take that down the lawyers [21:15] have just got in touch and that's not [21:16] supposed to be there and someone walks [21:19] into his office goes you know maybe it's [21:21] just possible it might end up supporting [21:23] more than that what if we add submit as [21:25] well so you can send instructions to a [21:27] remote system get it [21:29] something for you now one of the things [21:30] that Tim really likes about these is [21:32] they're all the same length so when you [21:33] view your your weblog files or your grid [21:35] log files in a fixed-width font [21:37] it'll lines up nicely wouldn't it sound [21:39] like a nice engineering solution the [21:43] next thing they need we've got a [21:44] protocol we've got a name we've got all [21:46] these these resources these grid pages [21:49] that are going to be out there on the [21:50] internet somewhere we need a way of [21:52] identifying the ones with with looking [21:54] at the thing we want to select he plays [21:57] around for a little while he comes up [21:58] with something that becomes the standard [22:00] unique resource finder the word serf [22:03] looks like it might catch on you know [22:05] people might talk about surfing the grid [22:06] that could be interesting and he comes [22:09] up with a draft for this and the first [22:11] draft of it is HTTP colon slash slash [22:13] user app pass that grid domain.com colon [22:15] port slash path question mark query hash [22:17] fragment and someone goes why are you [22:20] putting username and password [22:21] credentials in plain text on something [22:23] that's going to be part of the user [22:24] interface and he says good point we [22:25] shouldn't support that that would be [22:26] stupid [22:27] HTTP colon slash slash grid domain comm [22:31] port path query fragment and he's [22:34] thinking you know we could probably make [22:36] this a little bit simpler lets you know [22:37] query and fragment and normally going to [22:39] be optional so we'll just we'll delegate [22:41] those to stuff that's just being used [22:42] for search and indexing and stuff and [22:44] then he comes up with will will have [22:46] default documents so we can just go grid [22:48] domain comm and that'll host a default [22:50] document and then he thinks what is the [22:53] slash slash thing that is that you know [22:55] II got it from UNIX file systems but [22:58] this isn't a file system this is a [22:59] completely new thing let's get rid of it [23:00] and then someone points out that if you [23:03] put hosts on your network whose name is [23:06] a number that could be ambiguous between [23:08] the host name and the port number this [23:12] is a problem they think what could we do [23:14] about this he changes it he goes with [23:16] HTTP bang grid domain.com exclamation [23:20] mark that separates the protocol from [23:22] the resource from that from the host and [23:24] then he has a good idea at least what if [23:26] we just establish a convention that if [23:28] there is a host on your network which [23:31] has the same host name as the protocol [23:33] you're using we can leave that part out [23:36] completely grid bang domain.com that's [23:38] it [23:39] that is the fully qualified grid [23:42] of the default document on your server [23:44] bang nailed it and one of his team [23:47] serves to hit Tim don't you think grid [23:49] bang CERN dog sounds a bit silly and he [23:52] says not a silly HTTP colon slash slash [23:55] www.tsiyon.org and they says yeah yeah [23:57] right good point it's good thing we [23:59] fixed that one hey next up we need a way [24:04] of presenting these documents a way of [24:07] presenting this hypermedia and he starts [24:10] out with a proposal he's looking at a [24:12] thing called a standard generalized [24:13] markup language which has existed since [24:15] the early 1980s in industrial [24:17] typesetting that kind of thing and he [24:19] thinks we could use a simplified version [24:20] there so he comes up with this HTML [24:22] naught point one draft he sends around [24:25] some of his colleagues now Tim works at [24:27] CERN soon as a physics laboratory most [24:29] of his colleagues are scientists [24:31] mathematicians physicists they publish [24:33] equations which have lots of less than [24:35] this and greater than that and they come [24:37] back and they're like please don't use [24:39] angle brackets like literally every [24:41] document we've got will break the second [24:43] we try and publish it on your grid [24:45] system because we use those angle [24:46] brackets for something else and he [24:49] thinks that's an interesting point so he [24:51] takes a corpus of all the text files [24:53] that are already available on ftp [24:55] servers and you know online on various [24:57] network shares and he does a frequency [24:59] analysis to find the least common ASCII [25:02] character in all the publications at [25:04] certain anyone want to hazard a guess at [25:06] what it might be it's the tilled it's [25:11] Eskie code d7 is the least frequently [25:14] used character in the basic ASCII set he [25:17] thinks right that's it the tilled will [25:19] be the the mark up signifier character [25:21] in my new language plays around with it [25:23] okay first rule let's take a double line [25:26] break that's a paragraph break that [25:29] means the vast majority of text files [25:31] are already valid grid documents because [25:33] people have been writing in plain text [25:35] they don't use tools very often and [25:36] they've been using return return to [25:38] indicate a paragraph right let's just [25:39] preserve that behavior so a lot of these [25:41] documents we can publish online without [25:42] having to edit them at all it comes up [25:46] with some ideas around markup there we [25:48] go that's a heading that's a paragraph [25:50] it's got some bold text in it comes up [25:52] with something a little more elaborate [25:53] eventually he circulates his [25:55] after HTML version not 0.2 he's got this [25:59] : key value syntax for denoting [26:01] attributes so this is an HTML HTML [26:03] document the language is English he [26:05] realizes that most of these things are [26:07] gonna have some metadata which is for [26:09] machines to read and present elsewhere [26:11] in the user interface and it's gonna [26:12] have a document part which is for people [26:14] to read which gets presented in the main [26:16] user interface there are headings those [26:18] paragraphs he invents a way of linking [26:19] which he uses linked for because it [26:22] creates links and someone says Tim what [26:24] do we call it anchor and he's like what [26:25] is anchor why would you call it that [26:27] that makes no sense he creates a way of [26:31] embedding images he publishes this [26:34] specification online at about February [26:36] of 1992 and the world goes crazy [26:40] everybody gets grid fever suddenly [26:43] companies all over the world of thinking [26:45] we need to get some of that we need do [26:46] we know anyone who can do that historian [26:48] people start taking like a 2-day course [26:50] and setting themselves up as grid [26:52] masters and running very very successful [26:54] businesses because they know how to [26:55] print a business card and they know that [26:57] no one else knows that they don't know [26:58] what they're doing because this is also [27:00] new and exciting which incidentally is [27:02] how I got my first job but it finally [27:07] peaks with the launch of the first [27:08] graphical open source grid browser thing [27:11] called NCSA mosaic from the National [27:13] Center for supercomputing applications [27:15] based at the University of Illinois this [27:18] supports all of the existing Network [27:21] things that supports nntp so you can [27:22] read Usenet with it it supports gopher [27:24] supports FTP but it also supports this [27:28] new grid protocol it can browse grid [27:31] pages it can display images it can link [27:33] to other things and the world stands up [27:36] and takes notice now you remember I said [27:39] at the beginning that in the early 1980s [27:40] IBM had this this personal computer and [27:43] they were shopping for an operating [27:45] system to run on it so they went to the [27:48] best in the business you know the [27:49] absolute finest group of engineers who [27:52] are working on personal computer [27:54] operating systems anywhere digital [27:56] research and gary kildall the head of [28:00] digital research had a knock at the door [28:02] one day it's IBM going we want to buy [28:04] your operating system they sit down they [28:06] have a bit of a chat [28:07] now what gary kildall has real [28:09] and IBM haven't is that if your flagship [28:13] computer is made out of off-the-shelf [28:14] parts that you bought from China and [28:16] Singapore and Taiwan you don't actually [28:18] own anything the key to running [28:21] applications is no longer the hardware [28:23] because anybody can buy that the key to [28:25] running applications is going to be the [28:27] operating system and so kildall thinks [28:30] for a minute and he goes alright I will [28:32] sell you a license to use my CPM a [28:35] control program for microcomputers a [28:37] license that's why BM you can copy and [28:40] ship as many copies of these as you want [28:41] and I'll give you a 50% discount if I [28:44] can have a license to do the same thing [28:45] myself and I can keep selling it the [28:48] deal is done [28:49] IBM ship the personal computer with CPM [28:51] 86 gary kildall and digital research [28:54] continue to develop it when IBM launched [28:58] their first 16-bit computer the pc 80 [29:01] and 1984 digital research already they [29:04] got a thing called MPM 86 it's the multi [29:06] multitasking program for micro computers [29:09] it's a game changer it's the first truly [29:11] multitasking operating system doesn't [29:14] have any legacy constraints there's no [29:16] 640 K memory limit imposed by the 808 6 [29:19] processors it is a protected mode 16-bit [29:23] operating system runs on 286 processor [29:25] and pushes the thing to the limit but [29:28] what really kicks them into the big [29:30] league is that it runs on any compatible [29:32] hardware there's companies all over the [29:34] world now popping up with these beige [29:35] clones of IBM boxes buy one of those put [29:38] a digital research operating system on [29:40] it by the time IBM realized their [29:42] mistake it is too late Digital research [29:44] is rapidly becoming the biggest company [29:46] in the world [29:47] now ad in California there's competition [29:51] there's Apple who have had a pretty good [29:53] run since the Apple 2 computer was put [29:55] together in the late 70s they've done [29:57] the lessor they've done the first [29:59] Macintosh everything is not healthy at [30:02] Apple headquarters there's a lot of [30:04] strife there's a lot of conflict and [30:06] eventually in 1987 there's a hostile [30:09] takeover the shareholders take Steve [30:11] Jobs the visionary genius who founded [30:13] Apple Computer and they kick him out on [30:16] his ass and he moves across the street [30:18] because he wants to rub in their faces [30:20] what he's gonna do next [30:22] and he starts a new company next they're [30:26] done with page everything they've known [30:28] is gonna be black they come up with a [30:29] thing called the next station [30:31] it is a black cube about eight inches on [30:34] a side nobody's really ever seen [30:36] anything like it yeah [30:37] the next these IBM clones are going for [30:39] eight nine hundred bucks the next [30:41] station retails for ten thousand dollars [30:43] it is aimed at cinema production it's [30:46] aimed at medical research it's aimed at [30:48] really really high power software [30:50] development a couple of these boxes end [30:52] up in Sun Tim berners-lee is using one [30:55] of these to develop his prototype Grid [30:56] applications the first grid browser [30:59] which being next is called grid browser [31:01] application actually was built and run [31:04] on on one of these next systems [31:06] meanwhile Digital research merge with [31:12] the second largest software company in [31:15] the world [31:15] the wordperfect corporation the United [31:19] States Department of Justice flaps [31:20] around for about 18 months going [31:22] something must be done we can't have a [31:24] company that powerful won't somebody [31:25] think of the children but they can [31:27] actually come up with a way to block it [31:28] and so perfect digital is born they're [31:34] planning something big they have the [31:36] best office suite in the world they have [31:38] the best word processor in the world [31:39] they have the world's leading flagship [31:41] operating system they're thinking you [31:43] know we need a a grid browser we need a [31:45] way of browsing the grid from from our [31:47] platform so they go and they have a chat [31:50] with these guys anyone recognize these [31:52] two blues blue shirted individuals so [31:56] the chap on the left is Jim Clark Jim [31:58] Clark was the founder of Silicon [32:00] Graphics spent the 1980s building these [32:02] incredibly high-powered 3d graphics [32:04] workstations the chap on the right is [32:06] Mark Anderson Mark Anderson has been [32:09] working at NCSA here's one of the [32:11] founders of mosaic he's one of the [32:13] engineers who worked on the original [32:14] mosaic grid browser and they have both [32:17] for various reasons they've both quit [32:19] their jobs and they're fishing around [32:21] for something to do next and our [32:24] thinking here maybe we could we could [32:25] start a browser might be a good idea we [32:27] could build something you know mosaic [32:29] was all right let's kill it let's build [32:30] a mosaic killer we can call it Mozilla [32:32] how about that for a name [32:34] but then they get a phone call the [32:38] Nintendo corporation ring amount and say [32:40] look we got this new you know a Nintendo [32:42] 64 console we're working on and we think [32:45] it would be really neat if you could [32:47] plug this into the phone jack on the [32:48] wall and you could play like Mario Kart [32:50] and and Donkey Kong against your friends [32:53] around the world [32:53] and Jim Clark and Mark Anderson they [32:57] think long and hard about this and [32:58] eventually I think you know this [32:59] actually sounds like a pretty good idea [33:00] the Mozilla project never happens they [33:03] moved to Seattle they start working for [33:05] Nintendo and who do they meet walking [33:07] around the office in their first week [33:09] there but these two illustrious [33:11] gentlemen the two John's John Carmack [33:13] and John Romero couple years previously [33:16] over the course of a very very long [33:19] weekend fueled by a lot of pizza and [33:21] caffeine the John's Carmack and Romero [33:23] had ported Super Mario Brothers 3 to MPN [33:27] eight-six and they put it on a disc and [33:29] they sent a copy to Nintendo and they [33:32] said check this ad we managed to get [33:33] side-scrolling out of a 286 CPU they [33:35] said nobody could do that and Nintendo [33:38] went this is brilliant come and work for [33:40] us and they did off in Seattle Jim Clark [33:44] Silicon Graphics Mark Anderson from [33:46] mosaic John Carmack probably the finest [33:49] graphics programmer of any generation [33:51] John Romero who's quite good at level [33:54] design and having good hair the four of [33:56] them sequestered away with Nintendo [33:59] bankrolling the whole project who knows [34:00] what they're going to come up with in [34:02] the meantime the team at perfect digital [34:05] are back to the drawing board they have [34:07] the gem three desktop they're 286 visual [34:10] operating system they have a whole bunch [34:11] of things they need a grid browser so [34:15] they go to the next name on their list [34:16] they are knocking and they discover that [34:19] the law school at Cornell University [34:22] some guy called Thomas Bruce he's a [34:24] librarian he runs the library and he has [34:27] written an grid browser for NPM and gem [34:31] systems now the perfect digital [34:33] corporation has fantastic relationships [34:36] with the legal community because [34:37] WordPerfect is the de-facto word [34:39] processor for every legal firm in the [34:41] United States so they very rapidly [34:44] hammer out a deal they acquire the cello [34:46] source code they rebranded [34:48] they optimize a couple of the libraries [34:49] they add a couple features they come up [34:51] with grid perfect they got WordPerfect [34:53] they got letter perfect they've got no [34:54] perfect bank grid perfect there it is [34:56] their grid browser and they include [34:59] something which goes on to change the [35:01] face of distributed computing they [35:03] extend it as well as the grid global [35:06] repository of Internet insolent [35:08] documents protocol they come up with [35:09] something called safe secure application [35:11] frame encryption for the first time your [35:15] Grid client can talk to a grid server [35:16] and no one else can sniffing on what [35:18] you're doing no one can see Audion [35:20] suddenly the submit verb that they put [35:22] in the original protocol takes off [35:24] because people can do all kinds of [35:26] things with it it's not just get me a [35:28] document update that document delete [35:29] that document they've included word [35:32] perfect in their grid browser so anybody [35:35] who can read the grid can write the grid [35:37] they have publishing capability baked in [35:40] from day one so anybody who acquires a [35:43] copy of this thing they have full-blown [35:44] WYSIWYG authoring tools going along with [35:46] it but it isn't long before people start [35:49] thinking having to do everything in [35:52] terms of static documents here is [35:53] proving really really restrictive do you [35:56] think there's some way could we find [35:57] like a programming language we could [36:00] embed in the grid client and on our grid [36:02] servers so we could actually start [36:04] delivering rich applications and [36:06] behavior on this thing now in the [36:09] mid-1990s if you are going to embed a [36:12] programming language in a piece of [36:13] business software there is only one [36:16] choice there is an obvious contender for [36:18] this and that contender is lisp half a [36:23] century old that still has a perfect [36:25] timeless era bennett a few coders from [36:28] each new generation rediscovering the [36:30] lisp arts [36:31] your father's parentheses elegant [36:34] weapons for a more civilized age paul [36:39] graham is running a startup company [36:42] called via web and what via web have [36:44] realized quite early on is that lisp is [36:47] a fantastically good language for [36:48] building grid servers the grid protocols [36:51] are stateless and Lisp natively supports [36:54] a concept called continuation passing [36:56] it's a way of capturing execution [36:58] halfway through some chain of processes [37:01] capturing its nap shopping it and then [37:03] resuming it at a later date in a way [37:04] that you don't need to constantly be [37:06] allocating state and shared memory and [37:08] all kinds of overheads and resources and [37:10] things Lisp is the obvious choice now [37:13] around this time Sun Microsystems [37:17] who have been happily turning out [37:18] Solaris boxes they've invented a [37:20] language as well they come up with a [37:21] thing called Java which is a run [37:24] everywhere blah blah blah [37:25] object-oriented whatever and they're [37:28] pushing this thing hard they want [37:29] everyone to be using Java for everything [37:31] there ever is and they come knockin they [37:34] go and see perfect digital and they say [37:36] hey you know we'd love you guys to get [37:38] on board with this Java thing we're [37:40] doing and perfect digital light actually [37:41] we've got this language already and they [37:45] say well maybe you could rename it what [37:47] are you calling at the moment and they [37:48] say oh gotta call it hyper Lisp [37:50] could you call it Java Lisp or love a [37:55] script or something java re' so they [37:59] call Paul Graham in for a meeting and [38:02] over the course of a long afternoon he [38:05] explains his philosophy on programming [38:07] language design he tells them about the [38:08] history of Lisp he tells them about [38:10] symbol types he goes through something [38:12] that he would later publish online it's [38:14] called the blub paradox about how all [38:15] programming languages are not created [38:17] equal [38:18] by the end of the afternoon Java is [38:21] basically dead hyper Lisp has won out [38:24] the thing that clinches it is when he [38:26] points out that every piece of java [38:30] hyper Lisp every piece of hyper liske [38:34] program code is also an expression and [38:37] every expression is also a data [38:39] structure the language is completely [38:41] mutable you will never need to serialize [38:43] or deserialize anything because you can [38:46] literally take a program that is half [38:49] executed turn it into a data structure [38:51] pipe it over the wire to a client that [38:53] will basically rehydrate it and just [38:54] keep running it where it left off it [38:56] gives you this fantastically powerful [38:58] expressive model for running distributed [39:01] client-server applications he turns to [39:03] the finances and the bean counters he [39:05] says look this basically means everyone [39:07] who starts using your application is [39:09] effectively donating their computer to [39:11] you so you can run code on that so you [39:13] don't need to invest in so much [39:14] infrastructure [39:15] it's like DNS it scales with the number [39:18] of people who are trying to use your [39:20] solution that is it Java is out the [39:24] future of client-server Grid programming [39:26] belongs to hyper Lisp and I'll see what [39:29] it looks like [39:30] sure you do so we have HTML which by [39:36] this point is on 4.0 1 because 4.0 had [39:39] spelling mistakes in the specification [39:42] you can embed blocks of hyper Lisp code [39:45] directly into your code the simplest [39:47] possible thing you can do is you can say [39:50] alright we're going to now this reads [39:52] backwards because it uses a prefix [39:54] notation so we're going to bind and [39:56] we're going to bind the load event of [39:58] the document I'm not going to say [40:00] document object because this is Lisp and [40:02] it's cleverer than that we are going to [40:04] bind the load event of the document to a [40:07] lambda this is a callback and we're [40:09] gonna say when it happens pass the [40:11] element that triggered it and the event [40:13] that happened and run some code and that [40:15] code is just gonna say alert hello world [40:17] a a hello world in hyper Lisp let's look [40:19] at a slightly more sophisticated example [40:21] now so now we have a form on our page [40:25] and we're going to bind the submit event [40:27] for that to a lambda now reading it from [40:31] the back we've got this grid submit [40:32] method that you can see in here and what [40:35] that's going to do is it's gonna use the [40:37] network code baked into it and send a [40:38] callback back to the server and get some [40:40] response back from that and when it [40:42] receives the response from that because [40:43] this whole thing is completely [40:44] asynchronous by design because why would [40:47] it not be it will set the value of the [40:49] span to whatever came back from the [40:51] server and then down here we got a [40:52] little bit of kind of boilerplate markup [40:54] syntax and so this is a distributed [40:57] client server what in a parallel [40:59] universe they take 10 years to invent [41:01] and they'd call it Ajax and it would be [41:02] horrible this comes out of the box this [41:05] is free because we have symbol types in [41:09] list we can map the elements and the the [41:11] sort of fundamental runtime of the [41:13] browser directly onto symbols so they [41:15] don't need name spacing or qualification [41:17] we can have hyphens in our variable [41:20] names because all of our arithmetic [41:21] operators use postfix notation so this [41:24] convention around styling which is [41:25] growing up about spaces versus camelcase [41:28] underscores hyphens is the thing that [41:30] everybody settles on that just works the [41:32] lambda callbacks built-in give us a [41:34] synchronous client-server communication [41:35] out-of-the-box it is a beautifully [41:37] simple powerful programming model we're [41:41] just beginning to explore the potential [41:42] of it when there is trumpets and fanfare [41:45] and perfect digital announce gem 95 [41:47] their most powerful desktop operating [41:50] system ever they get the Beatles back [41:53] together for the launch party they're [41:55] using the grid browser is baked into [41:58] this operating system from the ground up [41:59] they're shipping a free version of [42:01] WordPerfect with no printer support that [42:03] will let you write an author documents [42:05] and put them on the grid they have this [42:07] massive party The Beatles do get back [42:09] because they're using this back button [42:11] from the browser in all of their [42:12] marketing literature and everybody loves [42:15] it the future is bright the future is [42:17] fantastic unless you are working for [42:20] Apple who have kicked Steve Jobs out and [42:22] are now kind of wishing maybe they [42:25] hadn't because things have gone from bad [42:26] to really bad to worse to really worse [42:29] and they are floundering around trying [42:31] to work out what they can do they need a [42:34] new approach they need a new operating [42:36] system they need a new brand they [42:38] basically need to overhaul everything [42:40] that they are doing and Gil Amelio who [42:43] is the CEO of Apple at the time he goes [42:45] into a board meeting one day and he says [42:47] you know maybe we could try and buy next [42:50] and get Steve Jobs to come back they [42:52] laugh him out of the room this is the [42:53] dumbest idea and there is no way Steve [42:56] Jobs is ever going back to Apple after [42:57] the way they kicked him across the [42:59] street but he's not the only visionary [43:04] engineer who's left Apple during their [43:05] tenure there there's a start-up be [43:10] incorporated be have a thing called the [43:12] B Oz it is a completely brand new [43:15] ground-up high performance operating [43:18] system it has dedicated threading [43:20] support for playing streaming media [43:21] doing real-time communication audio [43:23] processing all these kinds of things and [43:25] what Bea have realized is their [43:28] operating system is fantastic but the [43:30] boxes the physical computers they were [43:32] building to run it on those are proving [43:35] really difficult to pin down so they [43:37] have pivoted they've changed their [43:38] strategy they are going to try [43:40] and find hardware and their first step [43:43] because both of the engineers behind it [43:45] have an apple background is to port B us [43:47] onto Apple hardware they set up a [43:51] meeting they go and see the Apple people [43:53] and they say we've got this brilliant [43:55] high performance multi-threaded you know [43:57] real mode or a protected mode operating [43:59] system would you be interested in you [44:01] know acquiring it an Apple think this is [44:03] brilliant this is exactly the thing that [44:05] they've been looking for they jump all [44:08] in they love the styling they love the [44:10] branding they love it so much that the [44:12] next line of Apple computers that come [44:13] out which are instantly dubbed banana [44:16] boxes by the press these things are [44:19] brilliant up till now computers have [44:20] been grey and beige something that they [44:22] can make them in yellow no one had ever [44:24] thought of doing this before you know [44:25] imagine a computer that wasn't beige [44:27] people want this in their house it's [44:29] beautiful it's running this amazing new [44:32] BIOS operating system the one thing they [44:34] don't have is a grip browser so they [44:37] make a couple of phone calls and they [44:39] ring the cello people and they're like [44:40] we already did the deal with perfect [44:41] digital you missed that one and they [44:44] ring and CSA and NCSA say welcome a [44:46] Lakes kind of open source and limping [44:48] along that's not really doing very much [44:49] and then out of the blue they phoned the [44:52] Norwegian telephone company Telenor and [44:56] Telenor say you know this is funny we [44:58] have a grid browser we've been working [45:00] on here it's a thing we call opera and [45:02] we're not quite sure what to do with it [45:05] so the be us people and the Apple people [45:08] and the Norwegian telephone company make [45:10] a deal and they acquire opera the source [45:12] code for that the presto rendering [45:14] engine the network stack that runs it [45:16] they renamed it be link to go with B [45:18] boxes and be us and of course it [45:20] instantly gets renamed to blink by it [45:22] autocorrect it supports all the grid [45:26] protocols create update delete select [45:28] submit it's got a WYSIWYG editor [45:30] suddenly you know there is a first-class [45:32] competitor again to the the juggernaut [45:34] which is perfect digital running on [45:36] these beige office clones now people all [45:40] over the world have got these systems [45:42] and they are writing and posting grid [45:44] pages left right and centre everyone is [45:46] getting dial-up and everyone is getting [45:47] five megabytes of grid space free when [45:50] they do this because you can publish a [45:51] lot of stuff in in five megabytes [45:53] because we never [45:54] invented those massive JavaScript [45:55] frameworks in this universe and it's [45:59] becoming very very difficult to find the [46:02] good stuff because people are just [46:03] putting crap online because they think [46:05] if they get users maybe later they can [46:08] work out how to make money out of all of [46:09] that traffic and the big unsolved [46:12] problem in grid computing has become how [46:15] do we find content how do we search for [46:17] the things that we want to look at now [46:19] at Stanford there's a couple of PhD [46:22] students Sergey Paige and Larry Brin and [46:24] they're working out of a garage on an [46:27] algorithm they called backrub and they [46:29] call it back rub because it uses links [46:31] back to documents to decide how credible [46:34] they are if lots of pages on the grid [46:37] are linking to you then clearly you're [46:38] important and then if you link to [46:40] somebody that link means they're [46:42] probably important as well it's really [46:44] really clever and you know they're [46:47] bouncing this idea around and thinking [46:49] this could could really go somewhere [46:51] what are we gonna call it and Sergey [46:54] says you know my kid was in school the [46:55] other day and and they were telling him [46:57] about this this number called a Google [46:58] which is the biggest like number it's [47:01] like 10 followed by a million zeros or [47:02] something maybe we should call our new [47:04] search engine that because back rub [47:06] sounds a little bit questionable and [47:08] remember we haven't been invented yet so [47:10] no one's gonna be able to find us so [47:12] they call it Google and they ask someone [47:14] to go and register the domain and of [47:15] course what happens is autocorrect [47:17] happens and they end up with google.com [47:21] now it's a great idea you know the [47:23] technology works works really well they [47:25] have this wonderfully minimalist [47:27] homepage [47:28] everyone else is plastering as much [47:29] advertising as they can onto their grid [47:31] portals goggle has this really nice [47:33] stripped-down page but there is just one [47:35] problem they have no idea how they are [47:38] ever going to make any money out of [47:40] something that works well is minimalist [47:42] and they are providing for free and so [47:45] they start kind of looking around for a [47:48] business plan and someone suggests maybe [47:50] you could run advertising they didn't [47:52] like that idea at all and then Larry [47:55] Page takes a call it's Steve Jobs he's [47:58] working on something something big [47:59] something secret he might be interested [48:02] in acquiring goggles PageRank technology [48:05] now we don't know exactly [48:07] Lee what took place next we don't know [48:09] what was discussed in those meetings we [48:11] don't know who paid who for what or how [48:13] much but we know what happened after [48:15] that [48:16] computer journalists April 1999 everyone [48:20] is talking about the millennium bug [48:22] we're all going to get killed on the [48:24] stroke of midnight by all these nuclear [48:25] missiles that can't handle four-digit [48:28] dates or or whatever there is a break [48:31] from the millennium bug hysteria when [48:33] all of the computer journalists in [48:35] Silicon Valley get into work one day and [48:36] on their desk they find a hand-delivered [48:38] black envelope they open it up and [48:41] inside there is a card black printed [48:44] card on the front it just says get ready [48:46] for the next generation on the back [48:48] Menlo Park amphitheater 4th of May 1999 [48:52] they immediately think this must be a [48:54] Star Wars thing may the 4th be with you [48:56] so someone phones Lucasfilm Lucasfilm [48:58] say don't be ridiculous why would anyone [49:00] make a sequel to Star Wars just we are [49:03] not gonna mess with perfection and the [49:07] rumor mill starts working overtime you [49:09] know what is it why are we going what is [49:10] this next generation what are we talking [49:12] about on the day may the fourth comes [49:15] around the Menlo Park amphitheater is [49:16] the auditorium is packed a little before [49:19] 9:00 a.m. hush falls over the audience [49:21] Steve Jobs takes the stage and he talks [49:24] for a while he talks about the grid he [49:25] talks about digital publishing about [49:27] empowering people about technology about [49:29] the social impact of all of the [49:31] technology and just as everyone wonders [49:33] where he's going with this he reaches [49:34] behind his podium and he pulls out his [49:37] it is that a laptop it's called the next [49:42] book nobody has ever seen anything quite [49:44] like it it's less than an inch thick [49:46] it's machined from a single piece of [49:48] aluminium [49:48] it's anodized lustrous black it's got a [49:50] no floppy disk drive [49:51] it's got no cd-rom drive it's got no [49:53] serial port how do you plug your printer [49:55] in and they're going you don't need [49:56] printers this is the future you know um [49:58] it's got two USB ports and a headphone [50:01] jack the questions start buzzing through [50:03] the crowd how do you how do you connect [50:04] that to anything how do I get my games [50:06] how do I run SimCity it doesn't have a [50:07] floppy disk drive and that's when he [50:09] unveils his masterstroke because [50:10] throughout the audience there are plants [50:12] about two dozen Apple employees reach [50:15] under their seats and they pull out [50:17] their next books they open the lid [50:19] and they are already playing the live [50:21] video stream of jobs on stage doing his [50:24] presentation [50:24] nobody has seen Wi-Fi before nobody has [50:27] seen streaming video before nobody's [50:29] seen a laptop like this before they [50:30] forget they're supposed to be [50:31] journalists people are elbowing each [50:33] other in the face and clambering over [50:34] seats trying to get a a look at this [50:36] thing they are so blown away that they [50:38] completely forget they're supposed to be [50:40] there writing about this thing because [50:41] nobody has ever seen anything like it [50:45] it's the fastest lightest laptop the [50:48] world has ever seen [50:48] 333 megahertz arm RISC processer 64 [50:52] megabytes of RAM five gigs a flash drive [50:56] space it's got the next version of the [50:59] next-step operating system built into it [51:01] it has built-in wireless internet it [51:03] ships with a box called the next end or [51:05] a thing about the size of a hardback [51:07] book you plug this in your phone Jack [51:09] it's got a cd-rom drive in it it [51:11] supplies the wireless networking for [51:13] your house so you don't need to sit next [51:14] to a phone jack every time you're doing [51:15] anything and of course next step ships [51:19] with its own brand-new grid browser we [51:23] say brand-new it's called Nexus and the [51:26] DNA of Nexus goes all the way back to [51:29] the very first grid browser dot [51:31] application developed by Tim berners-lee [51:33] on the next operating system a decade [51:36] earlier it's responsive it's got the [51:38] fastest hyper Lisp interpreter anyone's [51:40] ever seen it's wired right into the [51:42] fabric of the new operating system and [51:44] the whole thing is seamlessly integrated [51:46] with goggle search technology you can [51:48] bring up a window you can search for [51:50] what you're looking for it doesn't [51:50] matter if it's in your folders in your [51:52] backups in your mail if it's on the grid [51:54] somewhere it will find it for you Steve [51:56] Jobs is back with a bang the next [51:59] generation has arrived the millennium [52:01] bug comes and goes the missiles do not [52:04] go off nobody dies but we all wake up [52:06] with one hell of a hangover and the next [52:10] generation of computing looks like this [52:12] we have Steve Jobs the next corporation [52:15] the next book running Nexus based on Tim [52:17] berners-lee's operating system we've got [52:19] the banana boxes running blink the [52:21] browser they got out of the Norwegian [52:23] telephone company Telenor we've got Gen [52:25] 95 and perfect digital going all the way [52:27] back to cello invented by a librarian [52:29] working in a law school [52:31] and we've got sun microsystems who for [52:33] some reason are still hanging around [52:35] going we can run NCSA mosaic look it [52:37] supports Java which nobody cares but [52:40] they've invested heavily in their [52:41] marketing and they're gonna keep [52:42] plugging away at it a couple of years [52:46] later the buzz starts Nintendo are [52:51] coming up with something big now [52:53] nobody's heard very much from them since [52:55] John Carmack and Romero and Jim Clark [52:57] and Mark Anderson all went off to [52:59] Seattle with Nintendo bankrolling [53:01] whatever they wanted to come up with and [53:02] they did ship a very successful Super [53:05] Mario quake world but after that [53:10] everything goes a bit quiet December [53:11] 2002 just in time for Christmas Nintendo [53:14] unveiled a platformer that they think is [53:16] gonna change the world and it's just [53:18] called reality it is a massively [53:21] multiplayer online world an immersive 3d [53:24] experience takes the world by storm it's [53:27] John Carmack and John Romero's dream of [53:29] implementing the Metaverse visualized by [53:31] Neal Stephenson in his books or the [53:33] matrix from William Gibson's Neuromancer [53:35] books the graphics code the [53:37] visualization the interface is John [53:39] Carmack and John Romero the network code [53:41] is Mark Anderson the 3d graphics engine [53:43] that powers the whole thing is straight [53:45] out of the DNA of Silicon Graphics it is [53:48] virtual reality brought to life and it [53:51] seems hard now to imagine life without [53:54] it doesn't it you know I mean can you [53:56] imagine a world without blink imagine a [54:00] world without hyper Lisp [54:01] imagine a world without B boxes without [54:06] next books that's a bit of a stretch [54:09] isn't it so how much of that is true how [54:14] close did we come in 1981 IBM went [54:20] shopping for an operating system they [54:22] knocked on gary kildall door the founder [54:25] of digital research - by CPM to run on [54:27] the IBM personal computer and gary [54:30] kildall was out he had gone flying [54:32] because it was a nice day there is a [54:35] slightly less romantic version of this [54:37] story where he just didn't like them and [54:39] told them to piss off [54:41] but they gave him first call he for [54:46] whatever reason did not take it [54:48] they went knocked on the next name on [54:50] their list which was this enterprising [54:52] young man mr. William H Gates the third [54:55] of my crew - soft who in our universe by [54:58] the way are building a little hyper list [55:00] for debugging tools out of an office in [55:01] Seattle about ten fifteen people you [55:03] know nice little operation friendly [55:05] people it could have happened [55:09] you know IBM could probably one or two [55:11] meetings away Microsoft would never have [55:14] become the company they did and the IBM [55:16] PC would have taken over the world [55:17] running a digital research based [55:18] operating system Tim berners-lee the [55:21] world wide web let me let me ask you why [55:24] don't start with one historical question [55:25] if I could it again you know we live in [55:28] the world and part you created the [55:30] architecture for if you think back to [55:32] the start if you had it to do all over [55:35] again would you change anything what [55:38] would you have done differently it's a [55:41] good question and of course you can [55:44] never ask what would have happened if [55:45] you'd have done something differently [55:46] there was some I mean there's some some [55:48] simple things in fact when I designed [55:50] the URI URL that this thing which stops [55:53] HTTP colon slash slash - the slash slash [55:58] was to indicate that we are actually [56:00] starting at the top that we're not [56:01] starting down with the next slash but [56:02] the but a cool line in there is well it [56:05] turns out people never use the slash - [56:07] without the HTTP colon on it turns out [56:09] really if you think about you don't need [56:11] this knife slash I could have designed [56:12] it to not have the session boy now that [56:15] people know people all over the people [56:17] on the radio are calling it backslash [56:19] backslash means that everybody's people [56:21] haven't they use that finger so much [56:23] think of the amount of you know of print [56:25] now that would have been sage if we [56:26] wouldn't have sales if I just remove the [56:28] double slash but there you go yeah and [56:30] you can't it seems like a good idea at [56:32] the time could have been different HTML [56:40] you know it's based on sgml that's where [56:42] we got the angle brackets from sgml [56:44] wasn't the only markup language latex [56:46] has been around since what the early [56:47] 1970s this is what latex markup syntax [56:50] uses lots of people at son knew later it [56:53] could so easily [56:54] HTML could have been based on something [56:57] else Microsoft had rich text format [56:58] going back to the late 1980s HTML could [57:02] have been based on rich text format the [57:04] idea of embedding Lisp as a programming [57:06] language via web was a real company Paul [57:09] Graham and his partners really were [57:11] building web servers in Lisp in the [57:13] mid-1990s they sold via web to Yahoo and [57:15] that's what you know made him a [57:17] millionaire he went on to found Y [57:18] Combinator with the revenue he made out [57:20] of that Lisp was embedded in all kinds [57:24] of things [57:24] AutoCAD came with an embedded Lisp [57:26] system Emacs came with an embedded Lisp [57:28] system it was the natural choice of [57:31] programming language there's even one [57:33] version of the story that says you know [57:35] when when Netscape approached Brendan [57:38] Eich they said we want you to come and [57:40] work for us because we want you to put [57:41] Lisp in the browser and what ended up [57:44] being JavaScript was a combination of [57:46] time pressure and marketing pressure and [57:48] if this doesn't look like C it's gonna [57:49] frighten everybody John Carmack and John [57:52] Romero really did port super super Mario [57:55] 3 to the IBM PC and they sent a disc to [57:58] Nintendo but in our timeline Nintendo [58:01] didn't say that's fantastic come and [58:03] work for us and in tendo said that is [58:05] fantastic and if it ever sees the light [58:06] of day we are gonna sue you so hard you [58:08] won't be able to find the crater because [58:12] Nintendo had no interest at all in [58:14] people being able to pirate their games [58:16] that Nintendo cartridge the physical [58:19] Hardware was a fantastically effective [58:21] copy protection mechanism Marc [58:25] Andreessen and Jim Clarke really did get [58:27] wormers to the point of signing a deal [58:29] with Nintendo to go and build a gaming [58:30] network and if they had that's what they [58:33] would have done instead of starting [58:34] Netscape Netscape would never happen [58:36] Mark Anderson wouldn't have ended up [58:38] barefoot on the cover of Time magazine [58:39] standing in front of a pile of money you [58:42] know and that was these these two really [58:44] smart people going what do we want to [58:45] work on next that looks kind of fun that [58:46] looks kind of fun maybe we'll do that [58:48] one [58:48] now we'll do that one that's how easily [58:50] it could have happened cello was real [58:52] real web browser ran on Microsoft [58:55] Windows written by Thomas Bruce who was [58:57] the head librarian at Cornell Law School [58:59] it was a real thing B us was a real [59:02] operating system written by two former [59:04] Apple employees there were [59:06] real meetings between the the engineers [59:09] behind BIOS and Jill Emilio prior to [59:11] Apple managing to acquire next getting [59:14] Steve Jobs to come back he came out with [59:16] the the iMac the iPod the iPhone you [59:18] know Apple's resurgence from a company [59:21] who were nearly on the brink of death to [59:22] a trillion-dollar market cap is almost [59:24] entirely down to the things they did [59:26] after Steve Jobs came back into the fold [59:29] in the in the 1990s and Google being up [59:33] for sale they tried to sell to excite [59:37] for $800,000 Google they were [59:41] established they had the reputation they [59:43] had you know this formidable thing they [59:45] just did not have a business plan they [59:47] were trying to sell it to excite anyone [59:49] remember excite yeah anyone used excite [59:52] like in living memory they were an [59:55] internet portal they were search engine [59:57] they could have bought Google for 800 [59:59] grand and they passed they said now we [60:01] don't think your company is going to [60:03] amount to anything [60:04] Google of course now is basically owns [60:06] the world and it's listening to every [60:07] word we say and indexing it all in case [60:09] it needs it all later and you know the [60:14] the lesson I mean that this talk is [60:16] designed to be thought-provoking and [60:18] interesting and kind of cover a little [60:20] bit history but I think the main if you [60:22] want to take a lesson away from this so [60:24] many of these systems that we use and we [60:28] rely on every day are not there because [60:31] somebody worked really really hard to [60:34] find the best solution or to make the [60:36] best decision or to find the best [60:38] possible team or circumstances to work [60:41] with they're there because people found [60:43] something that was just good enough and [60:44] they thought about what can we do with [60:46] it [60:46] what's gonna be next cuz you know we [60:48] have this thing in our industry analysis [60:50] paralysis where you get overwhelmed with [60:52] choice you don't know what to do next [60:53] literally you could do anything you have [60:56] a dozen different platforms to choose [60:58] from and you know almost all the web [61:01] development we do now is based on top of [61:03] a browser that might never have existed [61:05] if they'd gone to make a games network [61:07] instead running a programming language [61:08] that was bolted together in ten days by [61:11] somebody under marketing pressure from [61:13] Sun Microsystems even if you don't get [61:16] those decisions right [61:17] there isn't really a right and wrong all [61:19] there is is have we got enough to go on [61:21] and and make progress and as you see in [61:24] this talk we could have ended up making [61:26] a very very different kind of progress [61:28] maybe we still will but back here in the [61:32] real world [61:32] enjoy Chrome and Firefox and Safari and [61:36] Microsoft edge but that was the web that [61:39] never was thank you