Charles Babbage and Difference Enginge No. 2 | Doron Swade | Talks at Google

Charles Babbage and Difference Enginge No. 2 | Doron Swade | Talks at Google — Talks at Google

Talks at Google

AI-assisted summary of Charles Babbage and Difference Enginge No. 2 | Doron Swade | Talks at Google — original source. Check the original for full context.

Source published:

Key ideas

  • A working replica of Charles Babbage's Difference Engine No. 2, comprising roughly 8,000 components and weighing over 5 tons, was successfully built after a 17-year project led by Dr. Doron Swade at the Science Museum in London. (0:09)
  • Babbage's calculating engines embodied modern digital architecture over a century early, using discrete decimal states, mechanical pulse-shaping wedge locks, pipelined addition cycles, and anticipatory polling carries. (24:04)
  • Constructing the engine revealed minor archival design bugs in Babbage's original 1840s drawings, including reversed carry gearing and components depicted suspended in midair without mounting brackets. (55:40)
  • While Babbage's original Difference Engine No. 1 effort consumed £17,500 of British Treasury funds before cancellation, Swade estimated modern replication costs at roughly $2.5 million to $3 million. (59:46)

Detailed notes

## Background This talk was delivered by engineer and computing historian Dr. Doron Swade at Google's Mountain View headquarters as part of the Authors@Google program, and published on YouTube by Talks at Google on May 12, 2008. According to the page description, the session explores Charles Babbage's 19th-century quest to eliminate human arithmetic error through mechanical calculation and recounts Swade's seventeen-year project to construct a working Difference Engine from original Victorian designs.

## TL;DR Dr. Doron Swade details the life, character, and mechanical innovations of Charles Babbage, framing his calculating engines as the philosophical extension of the Industrial Revolution from muscle to mental labor. Swade explains how Babbage anticipated foundational computer science concepts—including digital decimal architectures, pipelining, mechanical pulse shaping, and microprogrammed control—decades before electronic computing emerged. He then chronicles the ambitious seventeen-year reconstruction of Difference Engine No. 2, demonstrating how his team resolved drafting errors, adapted nineteenth-century metallurgical standards, and engineered the machine's intricate programmable printing and stereotyping apparatus.

## Atmosphere & style The presentation carries the warm, erudite charm of a seasoned British curator addressing an audience of modern software engineers, balancing dense mechanical exposition with sharp, dry humor. Swade delivers his narrative with infectious scholarly passion, delighting in Babbage's eccentricities and polemical ferocity while treating the ironwork with deep technical reverence. The room feels engaged and intellectually captivated, bridging a 150-year technological divide as Swade turns complex cams, gear trains, and Victorian archives into a vivid, relatable engineering saga.

## Bulletin - [00:09](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=9s) A working replica of Charles Babbage's Difference Engine No. 2, comprising roughly 8,000 components and weighing over 5 tons, was successfully built after a 17-year project led by Dr. Doron Swade at the Science Museum in London. - [24:04](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1444s) Babbage's calculating engines embodied modern digital architecture over a century early, using discrete decimal states, mechanical pulse-shaping wedge locks, pipelined addition cycles, and anticipatory polling carries. - [55:40](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=3340s) Constructing the engine revealed minor archival design bugs in Babbage's original 1840s drawings, including reversed carry gearing and components depicted suspended in midair without mounting brackets. - [59:46](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=3586s) While Babbage's original Difference Engine No. 1 effort consumed £17,500 of British Treasury funds before cancellation, Swade estimated modern replication costs at roughly $2.5 million to $3 million.

## Key takeaways - [04:11](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=251s) The Victorian era treated mechanism not merely as engineering, but as an epistemological model where understanding any phenomenon equated to explaining its mechanical logic. - [05:50](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=350s) Babbage aimed to extend industrial mass production from physical muscle labor to mental labor, treating calculation as the industrial 'manufacture of numbers'. - [13:55](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=835s) Errors in mathematical and navigation tables stemmed from four human vulnerabilities: arithmetic calculation, manuscript transcription, manual typesetting, and visual proofreading. - [13:55](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=835s) To achieve absolute tabular accuracy, Babbage envisioned an end-to-end automated system that combined calculation with automatic mechanical output to stereotype printing plates. - [19:48](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1188s) Babbage viewed the universe deterministically and regarded God as a divine programmer whose natural laws operated without ongoing intervention. - [21:27](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1287s) While errata in navigation tables provided a utilitarian pitch for government funding, Babbage's deeper motive was establishing a new exploratory instrument for pure mathematics. - [26:33](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1593s) Difference Engine No. 2 operates as a strict digital system, using sword-shaped wedge locks to correct rotational drift under 2.25 degrees and intentionally jam the machine if errors exceed that threshold. - [29:25](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1765s) Instead of immediately propagating carries, the engine uses an anticipatory carry mechanism that flags a warned state when a wheel passes 9 and polls it via a sweeping helical spiral. - [33:30](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2010s) Conceived in 1834, the Analytical Engine anticipated key principles of modern computing, including separation of the central processing unit (mill) and memory (store), conditional branching, loops, and punched-card control. - [36:28](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2188s) Ada Lovelace's essential historical contribution was recognizing that numbers in calculating engines could represent non-numerical symbols, transitioning calculation into universal symbolic computation. - [39:25](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2365s) Difference Engine No. 2 executes calculation and printing through an internal mechanical microprogram driven by 28 conjugate cams. - [39:53](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2393s) The reconstructed machine calculates to 31 decimal digits across 7 orders of difference using odd-even pipelined addition cycles. - [42:26](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2546s) The modern build strictly observed historical authenticity by matching metallurgical compositions available in Victorian Britain to prove Babbage's design could have functioned in his era. - [54:40](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=3280s) Debugging the monolithic mechanism required adding a modern clutch and independent handle to cycle the printer separately from the 11-foot calculating frame.

## Detailed walkthrough ### [00:09](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=9s) Introduction: A Journey into a Past That Never Existed

Opening the session, the host Marian invites the audience to consider the concept of time travel, noting that just about one mile away from Google’s Mountain View campus, visitors can effectively step back 170 years into history. There sits the physical realization of the world's first mechanical, automatic calculating engine based on early computing concepts—a machine composed of 8,000 parts and weighing roughly 11,000 pounds (over 5 tons). Yet, as Marian points out, this is a journey into an alternate history: a past that never actually existed in Victorian Britain. Charles Babbage originally conceived Difference Engine No. 2 in the 19th century, but he never managed to construct it during his lifetime. It took roughly 150 years and one dedicated individual to turn Babbage’s unrealized dream into functional iron and bronze: guest speaker Dr. Doron Swade.

Marian provides a detailed overview of Swade’s distinguished multidisciplinary career. Holding a bachelor’s degree in physics and electronics engineering, a master’s in control engineering, and a PhD in the history of computing, Swade founded the Computer Conservation Society in the UK to focus on preserving early computing systems. At the time of this talk, he served as the assistant director and head of collections at the Science Museum in London, an honorary research fellow at Royal Holloway, University of London, and a guest curator at the Computer History Museum. Marian highlights Swade’s authoritative book on the project—published as *The Cogwheel Brain* in the UK and *The Difference Engine* in the United States—confessing that out of the roughly 700 computing history volumes lining the shelves of his own apartment, Swade’s book comfortably ranks among his top five favorites.

- Physical dimensions of Difference Engine No. 2: roughly 8,000 mechanical components, weighing approximately 11,000 pounds (5 tons). - Original design era: mid-19th century England; realized through a 17-year build project led by Doron Swade. - Doron Swade’s credentials: BSc in physics/electronics, MSc in control engineering, PhD in computing history, co-founder of the Computer Conservation Society, former assistant director/head of collections at the Science Museum in London.

---

[02:37](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=157s) Arrival in Silicon Valley and Roadmap of the Talk

Taking the podium, Doron Swade explains the timely catalyst for his presentation: the recent arrival of the five-ton calculating engine in Silicon Valley. Installed down the road at the Computer History Museum, the machine is slated for a one-year public exhibition, scheduled to run live demonstrations three times a week—provided, Swade quips, that it successfully recovers from its transatlantic "jet lag." Built starting six years earlier, the machine was never designed to travel. Its flight across the Atlantic subjected the delicate mechanism to severe vibration and altitude-induced low-temperature condensation that produced surface corrosion. Nevertheless, the engine is operational, and Swade emphasizes that no video or lecture can truly substitute for witnessing the massive apparatus moving in person.

Swade outlines the roadmap for the talk, tailoring it to an audience of software engineers and technologists. Rather than delivering a conventional, purely biographical history, he divides the presentation into three parts: the cultural and intellectual zeitgeist of the 19th century, an overview of what Babbage’s engines were conceptually designed to accomplish, and an in-depth mechanical exploration of Difference Engine No. 2 itself. Acknowledging the technical inclinations of the Google audience, Swade promises to skew his discussion toward mechanical logic, engineering constraints, and kinematic elegance, highlighting the sophisticated internal economy of Babbage’s design.

---

[04:11](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=251s) The Victorian Zeitgeist: The Age of Machinery

Swade draws an explicit parallel between modern Silicon Valley culture and the Victorian world of Charles Babbage. Today's society prides itself on living through an unprecedented era of relentless invention, where everyday news cycles trumpet novel technological breakthroughs, silver bullets, and accelerating rates of change. The victorians experienced an identical, frenzied sense of transformation. The middle decades of the 19th century witnessed an unbridled expansion of engineering ambition, dominated by an obsession with contrivances, mechanisms, and industrial contraptions. New materials, manufacturing processes, and mass-produced goods poured out of factories in overwhelming quantities, rapidly reshaping consumer life and retail shelves.

```mermaid flowchart TD accTitle: Victorian Mechanical Worldview accDescr: Diagram illustrating how external industrial machinery shaped the internal Victorian mindset and mechanical philosophy.

subgraph ExternalWorld["External Industrial Sphere"] A["Industrial Revolution"] --> B["Steam Power & Machine Tools"] B --> C["Relief of Physical Labor"] B --> D["Flood of Manufactured Goods"] end

subgraph InwardMindset["Inward Epistemology"] C --> E["Mechanical Philosophy"] D --> E E --> F["Understanding = Mechanical Explanation"] end ```

This mechanical obsession was far more than an outward economic shift; it became the Victorian epistemological framework for interpreting reality itself. Mechanical operation was the primary lens through which thinkers conceptualized the universe, human behavior, and nature. To illustrate the depth of this worldview, Swade quotes essayist Thomas Carlyle:

> “Thomas Carlyle the essayist said that it was the age of Machinery in every outward and inward sense of the word.”

As Swade elaborates, mechanical explanation became co-terminal with human understanding: if a natural or social phenomenon could not be modeled as a mechanism, the Victorian mind felt it had not been fundamentally explained.

---

[05:50](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=350s) Extending the Industrial Model: From Thing to Thought

While the familiar Industrial Revolution focused on mechanizing physical tasks and replacing muscle power, Swade argues that a second, less recognized revolution was brewing concurrently: the mechanization of mental labor. This effort sought to automate calculation and table-making through mechanical logic. The governing concept was radical: if an autonomous, automatic mechanism could physically embody mathematical rules, an operator could achieve results through raw physical exertion—turning a crank—that had previously demanded intense, fallible human intellect. To contemporary observers, this appeared to be nothing short of a physical thinking machine. Although the historical movement ultimately faltered during Babbage's lifetime, Charles Babbage (1791–1871) pioneered this intellectual leap. Describing Babbage’s grand objective, Swade explains:

> “His purpose in a sense was to extend the model of industrial production from thing to thought.”

Babbage’s son later summarized this operational philosophy by describing the calculating engine as a veritable “manufacture of numbers.” The machine took the industrial division of labor and factory production and applied it directly to the problem of numerical supply.

```mermaid flowchart LR accTitle: Bridging Technologies Concept accDescr: Two examples showing how radical new inventions were packaged inside familiar metaphors to ease public adoption.

subgraph Victorian["19th Century Bridging"] V1["Steam Street Railway"] -->|"Front-end disguise"| V2["Mechanical Wooden Horse"] end

subgraph Modern["20th Century Bridging"] M1["Electronic Pocket Calculator"] -->|"Marketing metaphor"| M2["'Electronic Slide Rule'"] end ```

To illustrate how societies cope with jarring technological leaps, Swade displays an image of a bizarre 19th-century "steam horse"—a street railway locomotive encased inside an artificial horse facade, designed specifically to avoid terrifying real horses on the road. Swade categorizes this as an example of a "bridging technology": an effort to disguise radical novelty within the comfort zone of familiar artifacts. He notes that modern computing repeatedly utilizes the exact same psychological bridge, pointing out that when Hewlett-Packard and others introduced the earliest electronic handheld calculators, marketing teams deliberately advertised them as "electronic slide rules" to make unfamiliar digital electronics palatable to practicing engineers.

---

[07:33](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=453s) Portraits of Charles Babbage: Radical Youth to Irascible Genius

Swade traces Babbage's intellectual and psychological evolution through a sequence of surviving visual portraits. He begins with an 1813 portrait of Babbage as a Cambridge undergraduate studying mathematics. Already exhibiting a radical, reforming streak, Babbage despised Cambridge’s conservative, hidebound mathematical curriculum and organized rebellions against it. Swade points out that Babbage’s hairstyle and posture in the painting intentionally mirror Napoleon Bonaparte—a defiant aesthetic statement in 1813, given that Britain was actively engaged in the Napoleonic Wars. This rebellious spirit extended to his personal life; in 1814, Babbage married Georgiana Whitmore in direct defiance of his father, a wealthy banker who insisted that marrying without independent wealth was reckless folly.

```mermaid flowchart LR accTitle: Chronological Evolution of Charles Babbage accDescr: Timeline showing the shifting stages of Charles Babbage's life from youthful radical to aged genius.

A["1813: Cambridge Radical<br>(Napoleonic defiance)"] --> B["1830s: Scientific Maverick<br>(Attack on Royal Society; 'Economy of Machinery')"] B --> C["1847–1849: Mature Designer<br>(Difference Engine No. 2; political battles)"] C --> D["1860: Venerable Grump<br>(Irascible genius; 4th Statistical Conference)"] ```

By age 53, captured in an oil portrait by Sir Thomas Lawrence hanging in the National Portrait Gallery, Babbage had assumed the mantle of a prominent gentleman of science, though not without controversy. In the early 1830s, he had alienated Britain's scientific establishment by publishing vicious polemics against the Royal Society, accusing its leadership of corruption and intellectual bankruptcy. Swade underscores Babbage's relentless, combative nature:

> “He was capable of incontinent savagery in these public attacks he would impugn the personal probity of its senior officers and he had no compunction and but to do this in public.”

Babbage managed to rehabilitate his reputation largely through his 1832 masterpiece, *On the Economy of Machinery and Manufactures*, a foundational text in political economy that analyzed factory systems and production efficiency.

Swade then presents an Antoine Claudet photographic portrait from the late 1840s, the precise window (1847–1849) when Babbage drafted the complete blueprints for Difference Engine No. 2. Though visibly weathered and bruised from his political battles—including a disastrous, bitter falling-out with Prime Minister Sir Robert Peel that terminated government funding—Babbage retains an intense, defiant energy. It was during this period of political isolation that he executed his most refined, economical mechanical masterwork. Finally, Swade shows the last known photograph of Babbage, taken in 1860 at the Fourth International Statistical Congress in London. Looking like a "venerable Grump" with a wild, Beethoven-like mane of hair, Babbage had settled into the persona his first biographer would famously christen: the "irascible genius."

[10:20](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=620s) Charles Babbage’s Character and Contrarian Temperament

Doron Swade paints a vivid portrait of Charles Babbage’s complex, volatile personality, noting that the twin traits of irascibility and towering genius remain the most compelling lens through which to understand him. Despite the extensive scholarship since 1963, Babbage stands out as a figure in whom “protest always triumphed over persuasion.” At the same time, he was a brilliant raconteur, socialite, and one of the most sought-after dinner guests in Victorian London; securing “Mr. Babbage” for an evening was considered a major coup for any hostess, guaranteeing entertainment and sparkling conversation. He was a bold thinker of startling sweep and original vision, but also a man of unyielding loyalties: his friends could do no wrong, and his enemies could do no right.

```mermaid flowchart TD accTitle: Contradictions of Babbage's Character accDescr: Chart outlining the public, intellectual, and psychological contrasts in Charles Babbage. subgraph Personality["Public vs. Private Self"] A["Celebrated Raconteur & Dinner Guest"] <--> B["Fierce Grudges & Public Denunciations"] end subgraph Ambition["Ambition & Bitterness"] C["Craved Honors (Lucasian Chair)"] --> D["Refused Knighthood; Bitter Over Incomplete Engines"] end subgraph PublicPosture["Operating Philosophy"] E["'Behaved as though being right entitled him to be rude'"] end ```

He desperately craved national honors, complaining that holding the prestigious position of Lucasian Professor of Mathematics at Cambridge was the only recognition his country had ever bestowed upon him. When offered a knighthood, he haughtily refused it, asserting that any such honor belonged rightfully to the calculating engine, not to him personally. Because he had been unable to complete his machines, he was left profoundly embittered, isolated in his inability to convey his vision to his contemporaries. As Swade summarizes his aggressive public posture:

> “He behaved as though being right entitled him to be rude and I think that's perhaps characterizes his public behavior.”

- Described as a radical, principled man constantly tilting at windmills and never hesitating to publicly denounce those he believed were in error. - Refused a knighthood on the grounds that the honor belonged to the engine itself. - Babbage felt that failing to construct Difference Engine No. 2 denied his contemporaries the ability to grasp his mechanical vision, fueling lifelong resentment.

---

[11:30](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=690s) Satirical Wit: The British Mind, American Enterprise, and Alfred Tennyson

To illustrate Babbage's sharp, biting literary style, Swade quotes from an obscure 1851 paper Babbage wrote on taxation, which touched upon the entrepreneurial climate of the mid-nineteenth century. Babbage drew a stark, humorous contrast between the conservative British temperament and dynamic American commercial ingenuity:

> “Propose to any Englishman any principle or any instrument however admirable and you will observe that the whole effort of the English mind is directed to finding a difficulty, a defect, or an impossibility in it. If you speak to him of a machine for peeling a potato, he will pronounce it impossible; if you peel a potato with it before his eyes, he will declare it useless because it will not slice a pineapple. Impart the same principle, show the same machine to an American or to one of our colonists, and you will observe that the whole effort of the mind is to find some new application of the principle and some new use for the instrument.”

This passage anticipates the enduring trope of European technical innovation clashing with American market exploitation. Swade then shares an infamous letter Babbage addressed to the poet Alfred, Lord Tennyson regarding a line in his poem *The Vision of Sin*:

> “Sir, in your otherwise beautiful poem *The Vision of Sin*, there is a verse which reads quote ‘Every moment dies a man, every moment one is born,’ close quote. It must be manifest that if this were true, the population of the world would be at a standstill. I would suggest that in the next edition of your poem you have it read quote ‘Every moment dies a man, every moment 1 and 1/16th is born.’ The actual figure is so long that I cannot get it onto a line, but I believe that the figure of 1/16th is sufficiently accurate for poetry.”

Swade notes that while modern audiences desperately want to believe Babbage wrote this tongue-in-cheek, “the terrifying thought is that he was deadly serious.”

---

[13:55](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=835s) The 1821 Genesis: “Calculations Executed by Steam” and Sources of Error

The genesis of Babbage’s computing machinery occurred in the summer of 1821 in London. Babbage and his close friend, the astronomer John Herschel, were laboriously checking astronomical tables computed manually by human "computers"—clerks assigned to carry out low-level arithmetic. Dismayed by the rampant discrepancies between the manuscripts, Babbage famously clasped his head in his hands and cried:

> “I wish to God these calculations had been executed by steam!”

“Steam” operated both as a metaphor for the unerring infallibility of machinery—the idea that the mechanical certainty of an apparatus would guarantee results free from human frailty—and as an industrial model of mass production to resolve the critical shortage of reliable tables. Asteroids were being discovered and mistaken for new planets, requiring immediate trajectory calculations. While astronomers pressured the government to compel the Royal Observatory at Greenwich to publish these tables, Greenwich refused, claiming it fell outside its royal charter. Consequently, the Astronomical Society (which Babbage had helped establish) commissioned Babbage and Herschel to do the calculations themselves.

```mermaid flowchart LR accTitle: Four Vulnerabilities of Mathematical Tables accDescr: Flow diagram tracing the four human sources of error in nineteenth-century mathematical table publishing. A["1. Manual Calculation"] -->|"Human arithmetic slips"| B["2. Transcription"] B -->|"Copying onto manuscript"| C["3. Typesetting"] C -->|"Setting loose metal type"| D["4. Verification"] D -->|"Visual proofreading failures"| E["Corrupted Printed Tables"] ```

Babbage identified four distinct stages where human error corrupted tables: 1. **Manual calculation:** Errors introduced by the human computers doing raw arithmetic. 2. **Transcription:** Mistakes made copying the calculated figures into intermediate manuscripts for the printer. 3. **Typesetting:** Setting loose individual type pieces by hand. Unlike text, where a reader naturally notices a misspelled, omitted, or transposed word, numbers offer no intuitive clues to error upon inspection. 4. **Verification:** Proofreading printed pages against originals, a task where human attention wanders and errors slip past undetected.

Babbage’s radical concept was not merely a calculating machine, but an end-to-end automated system that would calculate values and mechanically imprint them directly into stereotype plates, bypassing all human manual intervention.

---

[16:32](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=992s) Difference Engine No. 1 and the Philosophy of the 1832 Fragment

Beginning in 1821, Babbage spent a decade designing Difference Engine No. 1, an apparatus engineered to handle 16 digits and six orders of difference. It employed the mathematical technique of finite differences: while evaluating higher-order polynomials normally requires multiplication, division, subtraction, and addition, the method of differences reduces all computation to repeated addition alone. This allowed Babbage to bypass multiplication and division, which were mechanically difficult to implement at the time. The full design called for 25,000 components, an estimated weight of 15 tons, an 8-foot height, and even railway tracks beneath it to support its immense mass.

```mermaid flowchart TD accTitle: Mathematical and Philosophical Impact of Difference Engine No. 1 accDescr: Diagram illustrating the operational logic and philosophical influence of the 1832 trial piece. subgraph Mechanism["Mechanical Architecture"] A["Method of Finite Differences"] --> B["Replaces Multiplication & Division with Repeated Addition"] B --> C["Crank-Operated 1832 Trial Piece (1/7th of Engine)"] end subgraph Philosophy["Philosophical Breakthroughs"] C --> D["Transference of Thought to Mechanism"] C --> E["Demonstration of Miracles via Programmed Discontinuities"] E --> F["Scientific Defense for Geologists & Pre-Darwinian Evolutionists"] end ```

Although Babbage never finished the entire machine, he completed a operational trial portion in 1832—roughly one-seventh of the whole—which survives today in the Science Museum in London. Swade emphasizes that its historical weight cannot be overstated:

> “This is the first successful machine which has any sort of autonomy. This is the first successful transference of human intelligence to mechanism, the first machine that successfully embodies mathematical rule in mechanism. Reversing that, it's the first—it's the first machine that allows the ingression, if you like, of machinery into psychology.”

Babbage famously used this working fragment at his London soirees to illustrate his mechanical theory of miracles to the intellectual elite. He would configure the gear trains to follow a regular arithmetic rule for thousands of cycles before introducing a planned mechanical discontinuity that abruptly altered the output. Babbage argued that what natural theologians interpreted as a divine violation of natural law (a miracle) was in reality a pre-programmed discontinuity governed by a higher, deterministic law unknown to mortal observers. This mechanical demonstration provided a compelling metaphor for contemporary geologists and pre-Darwinian evolutionary theorists, offering a rational framework to explain long epochs of species stasis, sudden evolutionary appearances, and cataclysmic earthquakes without abandoning causal determinism.

[19:48](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1188s) The Thinking Machine and Physical Autonomy

Doron Swade explains that Charles Babbage saw the universe as deterministic, driven by natural phenomena that could be accounted for without constant appeals to divine intervention: “So for Babbage, God was a programmer, and those in the software industry may be flattered to know this.” The philosophical implication that a physical mechanism was performing mental labor was immediately felt by Babbage and his circle. His junior colleague Harry Wilmot Buxton observed that the marvelous pulp and fiber of the brain had been replaced by brass and iron, marveling that Babbage had taught wheelwork to think, or at least perform the office of thought. In 1833, Ada Lovelace’s mother, Lady Byron, similarly noted in her diary: “Last week we saw the thinking machine, for such it seems.”

This development represented not only the conceptual roots of artificial intelligence but also the birth of automatic computation. Swade emphasizes that both domains fundamentally depend upon mechanical autonomy: “You did not need to understand how the machine worked or the principle upon which it's based in order to get useful results, and this was unprecedented.” The machine constituted a monumental leap in logical architecture and scale, emerging with virtually no contemporary parallel. Displaying a slide depicting Babbage, Ada Lovelace, and John Herschel alongside the calculating engine, Swade playfully notes the slightly blurred image by joking that nineteenth-century digital photography was not particularly good, explaining that it depicts Babbage demonstrating his famous theory of miracles during one of his fashionable Saturday evening soirées.

[21:27](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1287s) The True Motive: A New Technology of Mathematics

Swade reveals that the traditional narrative of Babbage being motivated primarily by typos in printed tables is largely a historical misconception: “The vignette I've painted of Babbage and Herschel is misleading, and historians, myself included, endlessly charmed by it, have been misled into thinking that errors in tables was not only the jumping-off point, which it was, but that it was the enduring motive for Babbage's designs.” Babbage’s earliest private manuscripts make no mention of errata. Instead, errors in tables served as a rhetorical and utilitarian sales pitch to win financial backing from the British government. For Babbage, the calculating engine was fundamentally a brand-new technology for exploring pure and applied mathematics across three major fronts:

```mermaid flowchart TD accTitle: Mathematical Capabilities of Babbage's Engines accDescr: Three mathematical purposes conceived by Babbage beyond calculating tables. subgraph M1["New Technology of Mathematics"] A["Systematic Solution Method"] B["Heuristic Exploration"] C["Numerical Analysis"] end A -->|"All-zero column detection"| D["Halting criteria for equation roots"] B -->|"External cross-coupling shafts"| E["Generating series with no nth-term formula"] C -->|"Algebraic rearrangement"| F["Minimizing arithmetic machine operations"] ```

First, he conceived of mechanical computation as a systematic method to solve equations lacking analytic formulas. By monitoring the final column of the machine until it dropped to all zeros—the root of an equation—the machine would automatically halt. Swade draws a direct line between this mechanism and modern computer science: “Touring—pre-echo of Turing in a solution to the problem of computability using halting criteria; this is explicit in Babbage.” If an equation possessed no solution, Babbage noted that the operator could crank the machine *ad infinitum*. Second, Babbage treated the engine as a heuristic research tool. Using external shafts, columns could be cross-coupled forward and backward, allowing numerical feedback loops to generate mathematical series that lacked any known closed-form expression for the $n$-th term. Third, Babbage foresaw the creation of numerical analysis: an entire mathematical discipline dedicated to restructuring formulas to optimize execution on calculating hardware. In one of Babbage's own examples, an evaluation requiring 35 multiplications, 10 divisions, and 2 additions was algebraically restructured into an identical mathematical form needing only 5 additions, 2 multiplications, and 1 division.

[24:04](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1444s) Decimal, Digital Architecture and Early Logic Features

The Babbage engines are fundamentally decimal digital machines. They operate in base 10 using values from $0$ to $9$, and they are strictly digital because only discrete integer positions are mathematically valid. Swade explains: “Jams are good—it's a form of error detection; it says the integrity of the result has been compromised, a wheel is in an indeterminate position.” A non-integer value such as $2.5$ is never represented by positioning a wheel midway between two and three; instead, distinct decade wheels individually represent $2$ and $5$. Because the machine relies on mechanical bronze and iron parts rather than bistable electronic flip-flops that snap between states across a threshold, its digital nature had to be enforced entirely by a rigid mechanical control apparatus.

Swade illustrates Babbage’s early engineering experiments with spring-loaded half-round pawls resting in wheel indents. Although the dwell angle on early prototypes proved too shallow to force snap transitions cleanly, they demonstrate Babbage’s determination to lock wheels into definite positions and reject transitional states. Projecting modern computer terminology backward onto Difference Engine No. 1, Swade highlights several recognizable hardware and systems concepts embodied in the ironwork: - A parallel data bus - Microprogramming (the automatic execution of a sequence of micro-instructions triggered by a single macro-instruction) - Pipelining and polling - Pulse shaping (mechanical edge restoration) - One-bit binary latching - Mechanical input/output (I/O) - Bounce catchers operating analogously to Schmitt triggers to suppress mechanical contact bouncing

[26:33](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1593s) Mechanical Pulse Shaping and One-Bit Carriage Latches

```mermaid flowchart TD accTitle: Wedge-Shaped Lock Control Mechanism accDescr: Operating outcomes of the wedge lock mechanism determining alignment, stability, and jamming. subgraph Action["Lock Advances into Rack"] A["Sword-like wedge lock inserts"] --> B{"Tooth alignment status"} end subgraph Outcomes["Mechanical Response"] B -->|"Slight offset (< 2.25°)"| C["Self-corrects and re-centers wheel"] B -->|"Fully seated"| D["Locks wheel stationary during cycle"] B -->|"Drift > 2.25°"| E["Strikes tooth edge and jams engine"] end ```

Swade plays video footage demonstrating how Babbage achieved mechanical pulse shaping—the physical analogue of restoring degraded electrical signal edges. The mechanism relies on vertical, sword-shaped locking blades tipped with wedge-shaped ends that thrust between the teeth of horizontal racks. The lock performs three vital functions: 1. **Self-correction:** If a gear wheel has drifted slightly out of alignment, the wedge engages the flank of the tooth and mechanically forces the wheel back into true center. 2. **Immobilization:** When fully seated, the blade locks the wheel stationary so it cannot be disturbed during active computational strokes. 3. **Jam-on-error safety:** If a wheel drifts more than $2\frac{1}{4}$ degrees off-center due to wear, play, or mechanical error, the point of the wedge strikes the flat tip of the gear tooth rather than entering the gap. The machine immediately jams solid, warning the operator that the engine's integrity has been compromised and preventing erroneous output.

Swade then demonstrates the mechanical carriage handling. When two numbers are summed across columns, all decades add simultaneously. When an individual figure wheel passes $9$ to roll over past $10$, it does not carry immediately into the next decade—an action that would cause cascading mechanical resistance across thirty digits. Instead, reaching $10$ trips a one-bit mechanical latch into a "warned" state. During the subsequent carriage phase of the cycle, a spiral helix sweeps sequentially upward through the column, causing carry levers to poll each decade register in turn. When the mechanism interrogates a latch that has been warned, it triggers an increment on the next higher decade wheel before resetting.

[29:25](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1765s) The Mechanics of Anticipating Carriage and Interrogation

Doron Swade demonstrates the intricate mechanical operation of the carry mechanism on the engine, directing the audience's attention to the physical intersection between the shoe and the mechanical levers. He pauses the machine mid-cycle to illustrate how a digit wheel transitions between states, pointing out how the mechanism mechanically differentiates between a digit that needs to carry and one that does not. When an individual decade wheel rolls past nine, it flips a latch from an unwarned to a warned position, physically shifting an arm outward into the path of an interrogation shoe.

```mermaid flowchart TD accTitle: Mechanical Carry Cycle accDescr: The mechanical interrogation and carriage process in Babbage's calculating engine. subgraph S1["Digit Addition Phase"] A["Figure Wheel Rotates"] --> B{"Passes 9 to 0?"} B -- "Yes" --> C["Latch Flips to Warned State"] B -- "No" --> D["Latch Remains Unwarned"] end subgraph S2["Interrogation Phase"] C --> E["Arm Projects Outward into Sweep Locus"] D --> F["Arm Remains Retracted"] E --> G["Rotating Shoe Intersects Arm"] F --> H["Rotating Shoe Misses Arm"] end subgraph S3["Carriage & Reset"] G --> I["Next Decade Wheel Increments (+1)"] I --> J["Mechanism Resets Latch"] H --> K["No Carriage Occurs"] end ```

After catching a brief issue with an audio plug, Swade successfully freezes the apparatus right at the transition point. He highlights how the differing physical positions of the mechanical arms govern the logic of the calculation: an arm that is set sticks out further into space, allowing the sweeping shoe to intersect its trajectory and physically nudge the next decade wheel upward by one before resetting the latch. As Swade explains this purely mechanical conditional logic: “So that's a conditional device which interrogates each latch: if set, carries; if doesn't, you can see the two, and then resets.”

- The carry mechanism relies on intersecting loci between sweeping shoes and levers. - A wheel tripping from 9 to 0 moves the latch from an unwarned state to a warned state. - Mechanical interrogation probes each latch sequentially and executes the carry only if warned, before restoring the latch to rest.

---

[31:47](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1907s) The Analytical Engine and the Victorian Soap Opera

Transitioning from calculating machinery to general computation, Swade contrasts the difference engines with Babbage's next breakthrough: the Analytical Engine conceived in 1834. While difference engines are specialized calculating devices, the Analytical Engine astoundingly anticipated the fundamental principles of the modern electronic digital computer. Displaying Babbage's 1840 design drawings, Swade details a machine of staggering physical scale, towering 15 feet high and stretching 6 to 8 feet across for a single module. Its storage columns of figure wheels supported 50-digit precision arithmetic, with double-precision results reaching 100 digits. Although the displayed drawing shows roughly 20 registers measuring 10 feet in length, Babbage envisioned entry-level production machines with 100 registers and ambitious configurations holding up to 1,000 registers—a mechanical construct far larger than a contemporary steam locomotive.

Swade recounts the personal tragedy that framed this intellectual leap. Babbage’s original project to construct Difference Engine No. 1 had collapsed in 1833 when his master engineer, Joseph Clement, walked out following financial and managerial disputes, leading the British government to officially cancel its funding in 1842. The abandonment remained an open wound for the rest of Babbage’s life, filling his diaries with bitter disbelief and despair. Swade humorously summarizes the collapse as a quintessential period drama: “If you like, the story is of Victorian soap: there was a beautiful Countess, a flaming row with the Prime Minister, public accusations of profligate and wasteful spending of public monies, experts who disagreed about the fundamental utility of the machines, and a walkout by Babbage's chief engineer, Joseph Clement.” Swade quips that the modern team relived every single one of these tribulations while rebuilding the engine, dryly adding that he is still waiting for the beautiful countess.

- The 1840 Analytical Engine design stood approximately 15 feet high and 6 to 8 feet wide. - Precision was designed for 50 decimal digits, producing 100-digit results in double precision. - Entry-level configurations were planned around 100 registers, dwarfing railroad locomotives. - Joseph Clement walked off the Difference Engine No. 1 project in 1833, leading to government cancellation in 1842.

---

[33:30](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2010s) Modern Computing Concepts and Mechanical Notation

Examining the architecture of the 1840 Analytical Engine, Swade catalogues its modern logical attributes. The machine was programmable using four distinct types of punched cards and possessed a clear von Neumann architecture that cleanly partitioned the processor (the "mill") from memory storage (the "store"). It featured a serial fetch-execute cycle, conditional branching via if-then decision structures, iterative loops, microprogramming, pipelining, and an anticipating carriage mechanism serving the role of modern carry-lookahead adders. It performed parallel processing, automated multiplication and division, and outputted results via a printing and stereotyping unit, punched cards, or an automated graph plotter. Swade passionately refutes any suggestion that historians are merely reading modern ideas into crude mechanical artifacts, asserting: “All these features are explicitly embodied in the detail of design, and what I'll give you is just some flavor of the extent and detail of what he did.”

```mermaid flowchart LR accTitle: Analytical Engine Architecture and Notation accDescr: Architecture of the Analytical Engine and Babbage's three-part mechanical notation system. subgraph AE["Analytical Engine Architecture"] direction TB Cards["Punched Cards<br>(4 Distinct Types)"] --> Mill["The Mill<br>(Processor / ALU)"] Store["The Store<br>(Memory Registers)"] <--> Mill Mill --> Output["Output Subsystems<br>(Printer, Stereotype, Plotter)"] end subgraph MN["Mechanical Notation Language"] direction TB Tab["Tabular Form<br>(Alphabetical indices, motion types)"] Time["Timing Diagrams<br>(50-division cycles, phasing, pipelining)"] Flow["Flow Diagrams<br>(Data isolated from control)"] end ```

To coordinate this colossal apparatus without modern drafting tools, Babbage invented a symbolic Hardware Description Language termed "Mechanical Notation." Because Babbage was a mathematician who naturally reasoned in symbolic abstractions, he formulated three distinct design representations: 1. A **tabular system** tracking components through alphabetic labels with up to six superscripts and subscripts detailing connectivity, driving points, and whether motion was circular, linear, continuous, or intermittent. 2. **Timing diagrams** laying out component phasing and mechanical pipelining across 50 discrete cycle divisions, separating even and odd axes. 3. **Flow diagrams** that deliberately separated data lines from control sequences within a unified schematic.

The printing and stereotyping mechanism alone—shared between Difference Engine No. 2 and the Analytical Engine—contained roughly 4,000 parts, and Swade notes that his modern construction team had to rely on Babbage's mechanical notation sheets to resolve ambiguities in the engineering drawings.

- Features embodied: punched-card control, store/mill separation, conditional branching, loops, carry look-ahead, and automated plotting. - Mechanical notation employed up to six indices per variable to define physical dynamics and linkages. - Machine cycles were partitioned into 50 timing increments rather than 360 degrees. - The printer and stereotyping apparatus comprised roughly 4,000 components.

---

[36:28](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2188s) Ada Lovelace: Debunking Myths and Celebrating True Vision

Swade tackles the contested legacy of Ada Lovelace, carefully dismantling popular exaggerations while elevating her actual, profound contribution to computer history. Popular accounts typically assert four claims: that Lovelace was an innate mathematical genius, that she made influential architectural contributions to the Analytical Engine, that she was the world's first computer programmer, and that she served as the prophet of the computational age. Swade methodically dismisses the first three. Lovelace was an enthusiastic beginner in mathematics, not an accomplished genius; she only commenced studying elementary differences in 1839, five years after Babbage had already conceived the Analytical Engine. While Babbage was undeniably flattered by the 17-year-old Lovelace when they met in 1833 and warmly encouraged her interest, she did not alter the design or mechanical concepts of the engine. Furthermore, although she published the first detailed algorithmic program in her famous 1843 notes, the underlying computational routines were written by Babbage himself for his 1841 lectures in Turin.

Swade argues that stripping away these false accolades is essential to appreciating Lovelace's genuine intellectual triumph. Her real breakthrough was conceptual: while Babbage viewed his machine strictly as a mathematical calculator or an algebraic processor, Lovelace grasped that numbers could serve as arbitrary operational symbols representing non-numerical entities, such as musical notes or letters of the alphabet. She recognized the qualitative leap from mere arithmetic calculation to universal symbolic computation governed by rules. Summarizing her unique historical significance, Swade explains: “She was the only one who saw what the fundamental significance of Babbage's analytical engine was and saw it—saw further than Babbage did.”

- Lovelace began studying elementary differences in 1839, five years after the engine's conceptualization. - The algorithmic tables in Lovelace's 1843 publication were devised by Babbage during his 1841 Turin presentations. - Lovelace pioneered the concept of universal symbolic manipulation, distinguishing generalized computation from numerical arithmetic.

---

[39:25](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2365s) Introducing Difference Engine No. 2

Swade brings the narrative forward to Difference Engine No. 2, the design that the modern reconstruction team ultimately turned into physical reality. Babbage drafted the engine between 1846 and 1849, leaving behind a portfolio of 20 detailed engineering drawings. Among these, Swade singles out drawing 163A—the main elevation—describing it as so vividly evocative of the machine's architecture that it remains permanently burned into his memory. Cranking the main handle drives the engine through its cycles via mechanical microprogramming, implemented through stacked assemblies of 28 conjugate cams.

- Difference Engine No. 2 was originally documented by Babbage across 20 design drawings. - Drawing 163A contains the comprehensive main elevation drawing used by the modern replica team. - The machine's physical microprogram is executed through a drive assembly containing 28 conjugate cams.

[39:53](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2393s) The Physical Specifications and Programmable Printer

The Difference Engine No. 2 is a massive mechanical masterpiece consisting of 8,000 parts made of bronze, cast iron, and steel. It stands 7 feet high, stretches 11 feet long, and operates using seven orders of difference. As the machine cycles, the tabular values and their first through seventh differences are calculated, culminating in a 31-digit number that appears on the final column. This end product is then sent to the printing apparatus, which simultaneously prints a hard copy and produces stereotype plates.

Doron Swade highlights that the printer is of particular significance because it is entirely mechanically programmable. Users can program the page format, adjusting for variable line heights, variable column widths, and custom gaps between groups of lines to make the tables easier to read. It can print in one, two, three, or four columns at will, and it can stereotype in two different fonts (small and large) simultaneously, automatically adapting the line height to the font size. It even handles automatic line wrapping and column wrapping, regardless of whether the printing goes down or across the page. Swade jokes about its lack of modern features: “The print is of particular significance because it is programmable; you can program the format on it. It doesn't run Windows, and it's not portable.” He adds with a smile that it still "crashes nearly as often though," but notes that mechanical crashing is actually a positive feature because it serves as immediate error detection.

- **Total Parts:** 8,000 parts (bronze, cast iron, and steel) - **Dimensions:** 11 feet long, 7 feet high - **Precision:** Calculates to 31 decimal digits - **Formatting Capabilities:** Variable line height, variable column width, 1 to 4 columns, dual-font stereotyping, automatic line and column wrapping

---

[41:11](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2471s) A Seventeen-Year Build and the "Multiple Original"

The completed engine shown in Swade's presentation was finished in London in 2002, capping off a project that spanned 17 years from its start in 1985. Swade reflects on the immense human effort required to bring Charles Babbage's designs to life, noting that the timeline was stretched by factors far beyond pure engineering. “Strictly speaking, it was built over a period of 17 years; it was not just the logic of the building—there was something called people that were involved, and they are much less logical than solving technical difficulties,” Swade explains, pointing to the constant demands of fundraising, publicity, and team management. The machine currently housed at the Computer History Museum in Mountain View is an exact duplicate of the London engine. Because the team manufactured two sets of printer parts at the same time, the California machine is technically a "multiple original."

The physical structure of the engine features a prominent operating handle, a massive cam stack, the printing mechanism, and a series of ten columns. Numbers are stored in figure wheels on these columns, making the engine a decade decimal digital machine. The digits are arranged vertically, moving from units at the bottom up through tens, hundreds, and thousands to achieve 31-digit precision. The core design relies on three main components: the figure wheel, the sector wheel, and a helical polling device that rotates to poll each decade.

---

[42:26](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2546s) Authenticity, CAD Drawings, and a Pitch to Google

To build the engine, the modern team had to translate Babbage's original 19th-century drawings into modern solid-model CAD drawings. While Babbage's drawings were incredibly detailed—fully specifying the shapes of the parts and what they needed to do—they lacked crucial manufacturing information. There were no details on manufacturing methods, tolerancing, choice of materials, or surface finishes. The modern team had to fill in these blanks by studying surviving 19th-century machines and performing chemical composition analyses on the metals Babbage actually used. This ensured they did not use materials of a higher specification than what was available in Babbage's day. Swade emphasizes the strict discipline of the project: “Authenticity was the watchword of the project; we had to resist the charge absolutely of somebody saying, 'You built the engine, but Babbage could not have.'”

Swade uses this moment to make an open pitch to Google. He shows a 3D simulation of the machine and expresses a desire to get Google interested in creating a complete, highly detailed simulation of Babbage's even more complex Analytical Engine. Furthermore, he urges the digitization of the Babbage archive in London. Currently, the physical archive is located 70 miles outside of London, and only three people have ever thoroughly studied it: one who is now deceased, one who is in his 90s, and Swade himself. "This stuff needs to be made available to the public," Swade pleads, jokingly adding, "I think Google should adopt Babbage—end of advertisement."

---

[43:41](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2621s) Running the Engine: "Big Iron" in Motion

Swade shares a video demonstration of the Difference Engine No. 2 operating, noting that "there is no substitute for seeing this machine run, especially with its sound." The video captures the heavy, rhythmic mechanical symphony of the engine. The audience hears the sharp, slapping sound of the "sword blades"—the mechanical locks—clamping into place to secure the wheels, followed by the distinct sound of the printing mechanism taking an ink impression.

The video highlights the massive cam stack turning and shifting to execute its mechanical microprogram. The machine utilizes pipelining to optimize its calculations, splitting the work between odd and even columns. The odd axes (columns 1, 3, 5, and 7) perform their additions in one half-cycle, while the even axes calculate during the other half-cycle. The values are added, the results are transferred down the columns, and the final output is sent to the printer. Swade points out that the cam stack in the video is heavily coated in a dark grease, making it look a bit messy, whereas the duplicate machine at the Computer History Museum is much cleaner because it uses a modern transparent grease. "This is heavy machinery; this is Big Iron," Swade remarks, pointing out how quickly the locks must slide in and out, and noting that the pressure angles were so steep that the team couldn't even turn the handle when they first assembled it.

```mermaid flowchart TD accTitle: Babbage Difference Engine No. 2 Printing and Stereotyping Process accDescr: Flowchart showing how numerical values are mechanically transferred from figure wheels to print wheels, and then simultaneously printed on paper and pressed into stereotype trays.

subgraph ValueTransfer ["1. Mechanical Value Transfer"] A["Figure Wheel (Horizontal Plane)"] --> B["Sector Wheels & Racks"] B --> C["Spindles & Vertical Racks"] C --> D["Print Wheels (Vertical Plane)"] end

subgraph DualOutput ["2. Simultaneous Output"] D --> E["Ink Rollers & Paper Feed"] D --> F["Stereotype Heads (Small & Large Fonts)"] end

subgraph FinalProducts ["3. End Products"] E --> G["Ink Print Copy (For Record Keeping)"] F --> H["Plaster of Paris Trays (Stereotype Plates)"] end ```

---

[45:00](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2700s) Slicing Pineapples and Secondary Carries

As the video demonstration transitions to the front of the machine, the shifting, alternating motion of the odd and even columns becomes clearly visible. Swade marvels at the sheer visual beauty of the mechanism, joking, "Who cares whether it works when it looks like that!" The video then focuses on the back of the engine to show the carry mechanism. These parts are incredibly sharp and move with immense mechanical force. Swade warns of the physical danger of the machine, noting that "if you held a pineapple up to it, it would slice it," and jokes that "you can tell a Babbage engineer by his hand."

The carry mechanism works by polling each decade in turn. Crucially, the machine is designed to handle "secondary carries"—which are carries that occur as a cascading result of previous carries. The mechanical logic must propagate these carries up through the columns of figure wheels to ensure absolute mathematical accuracy. After demonstrating these movements, Swade pauses to check the time, offering to either take questions or continue explaining the printing and stereotyping mechanisms.

---

[46:53](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2813s) The Mechanics of Printing and Stereotyping

Swade continues by explaining the mechanical genius behind the printing and stereotyping system. The fundamental engineering challenge Babbage faced was spatial: “The problem is: how do you get a value from a figure wheel in that plane to a print wheel in that plane?” To bridge this gap, Babbage designed an intricate mechanical train consisting of sector wheels, horizontal racks, vertical spindles, and vertical racks. When the figure wheels turn, this system of gears and racks translates the horizontal rotation into vertical movement, transferring the calculated value directly to the print wheels. While the engine calculates to 31 digits of precision, it prints the results to 30 digits.

A 3D simulation illustrates the step-by-step printing cycle. The paper stock is fed through rollers, and the newly calculated result on the print wheels is inked. The paper carriage then moves up to take an ink impression. The physical action of the carriage rising automatically winds the paper forward to prepare for the next line. The print wheels then reset to zero, receive the next calculated result, and the cycle repeats. Swade notes with a smile that the virtual simulation is completely silent, unlike the clattering reality of the physical machine.

---

[48:24](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2904s) Stereotyping in Plaster and the Ink Bath

At the same time the machine is printing on paper, it is also performing stereotyping. Because these actions happen in parallel, there is absolutely no printing overhead. The stereotyping process uses trays filled with a soft material—plaster of Paris was used for the team's very first experimental runs. The stereotype heads, which feature both small and large fonts, press directly down into these plaster trays to create molds. The trays index automatically at different rates to adjust for different line heights. Because the trays have two degrees of freedom, they automatically jump across to start a new column when they reach the end of a page.

The ink print on paper is intended only for immediate record-keeping and proofing; the actual mass production of mathematical tables is done by casting metal printing plates from these plaster stereotype molds. Swade explains the revolutionary nature of this design: “You've eliminated the human from the loop entirely, from calculation to the production of stereotype plates.” To ensure perfect alignment and formatting, the trays are driven by falling weights rather than the manual turning of the operator's handle, removing human inconsistency from the indexing process. The entire layout is programmed mechanically via pattern wheels. Finally, Swade highlights the details of the ink bath, which features a cover, thick "gooey" ink, a print roller, and a spreader roll. Driven by a snail cam, the spreader roll slides back and forth to smear the ink both horizontally and vertically, ensuring an even coat for every single print.

[50:56](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=3056s) Printing Simulation and Mechanical Timing

Doron Swade walks through the final stages of the Difference Engine simulation, pointing directly into the printing mechanism to illustrate how it accepts and outputs data. Rather than generating output all at once, the machine operates incrementally. The print wheels receive the computed results from the calculating columns, and the audience can see them in action as they press down to produce a hard copy. Swade highlights the exceptionally rapid timing cycle required by the machinery: “watch that print wheel come down now boom boom it has to get out of the way before it comes out.” A sweep arm automatically advances the paper roll once each impression is made, following a precise sequence where each new result is inked, pressed into paper, and then cleared as the mechanism resets to zero.

The printer can also be configured to include an index column alongside the calculation. As shown in the simulation, the independent variable $x$ increments sequentially from $131$ to $132$, while the adjacent print wheels register a high-precision result extending up to $27$ digits (within a $33$-digit capacity). The visual simulation cleanly conveys the internal mechanical logic of how the calculating engine and printing section coordinate their movements step by step.

```mermaid flowchart LR accTitle: Difference Engine Output Cycle accDescr: Three-step pipeline showing the flow from calculation to index stepping and printing. subgraph Calculation ["Calculating Section"] A["Figure Wheels"] -->|"Transfer result"| B["Print Wheels"] end subgraph Printing ["Print Mechanism"] B -->|"Rapid cycle (boom-boom)"| C["Inked Impression"] C -->|"Paper sweep"| D["Advance Paper & Reset to Zero"] end subgraph Indexing ["Table Sequence"] D -->|"Increment x (e.g., 131 to 132)"| E["Next 27-Digit Result"] end ```

---

[51:54](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=3114s) Debugging a Monolithic Engine and “The Cogwheel Brain”

Swade explains the immense practical difficulties of maintaining and troubleshooting Difference Engine No. 2. The entire apparatus is a hardwired, monolithic system whose complex internal motions are synchronized exclusively by a central cam stack. Because Babbage never built a physical machine of this scale, his drawings did not include provisions for uncoupling components or isolating assemblies. When the engine jammed, Swade and his team could not simply detach the printer from the calculating frame. Instead, they had to resort to a painstaking process of elimination, using what Swade jokes was “the first world's first logic probe”—a simple mechanical pointer used to prod components, lever up the sector wheels and figure wheels, and wiggle the locks to detect where the mechanism was binding solid and where it still had mechanical play.

Turning to the project's documentation, Swade presents the history of the endeavor, written for the 1991 bicentennial of Babbage’s birth. The original volume covers the completion of the calculating engine in 1991, whereas the 2002 American edition—published under the title *The Cogwheel Brain*—features an updated account and photographs of the completed printing section. Swade explains that the title *The Cogwheel Brain* was deliberately selected to evoke Thomas Carlyle's concept of the mechanical age, reflecting both the internal and external lenses through which nineteenth-century society began to understand the world. It captures the emergence of a computational model of consciousness that only became possible once humanity conceived of autonomous machines. Swade concludes his formal presentation with a slide of the engine, describing it as a “sumptuous piece of engineering sculpture.”

---

[53:36](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=3216s) Q&A: Setting Initial Values and the Choice of Decimal Arithmetic

Opening the floor to questions, an audience member asks how data was originally entered into the engine. Swade clarifies that the engine has no dynamic input during a run; operators can only input initial conditions by hand-setting numerical values directly onto the figure wheels. However, before touching the wheels, the operator must deliberately disengage the engine's internal security locks. Babbage had engineered the system to be virtually tamper-proof: “Babbage challenges you and says walk up to the machine even when it's functioning and attempt to derange it externally and you won't be able to.” To set up a calculation: - The operator calculates a setup table in advance by hand, determining the initial value of $x$ and the corresponding higher-order differences for the chosen polynomial (incorporating any positive or negative coefficients up to terms in $x^7$ or $x^6$). - The setup table explicitly accounts for the engine’s alternating odd-even pipelining offsets. - Using manual override levers, the operator releases the internal interlocks. - The machine is rotated to the odd axes setup point, where the odd-numbered columns are keyed in by hand and locked. - The engine is advanced to the even axes setup point to set columns 2, 4, 6, and 8. - The machine is reset to zero, the manual locks are re-engaged, and from that point onward, every completed cycle of the drive handle computes and prints the next sequential value in the mathematical table.

Another attendee asks whether Babbage considered binary arithmetic, and whether using binary would have made the engine easier to manufacture. Swade responds that Babbage thoroughly evaluated almost every practical number base, including 3, 5, 7, 12, 15, 100, and binary. He ultimately opted for a decimal, digital representation strictly out of engineering efficiency and physical economy. In mechanical computing, lower bases such as binary require far more digits to represent large values, which exponentially multiplies the total count of mechanical linkages and moving components. A higher base such as duodecimal or hexadecimal would have required complex radix conversions for practical use. Decimal offered the optimal trade-off: it eliminated the need for human base conversion while keeping the overall part count manageable.

---

[56:01](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=3361s) Engineering Bugs, Babbage’s Omissions, and Modern Adaptations

Responding to a question about whether Babbage had designed test equipment like the “logic probe,” Swade notes that Babbage never had the opportunity to build the full machine and discover the necessity of diagnostic access points. Convinced that Babbage would have introduced decoupling mechanisms had he built the engine, Swade insisted on adding an independent handle and clutch to the printer drive. Because the printer sits roughly 11 feet away from the main calculating crank, fine-tuning print tolerances was impossible for a single operator. Furthermore, operating the printer to prime the ink roll previously required cycling the entire calculating engine, which corrupted the mathematical state and forced a complete re-entry of the setup table. The added clutch allows operators to disconnect the drive train and cycle the printer independently.

When asked if there were design bugs in Babbage’s original schematics, Swade explains that while the fundamental mathematical logic was sound, the archival drawings represented an incomplete, arrested engineering project rather than workshop-ready blueprints: - **Missing and floating parts:** Babbage's elevation drawings showed the inking rollers suspended unsupported in midair without brackets or frames. - **Inconsistent dimensions:** Identical parts differed in dimensions across various design sheets, requiring the team to harmonize measurements. - **Mechanical omissions:** The drawings featured omitted linkages alongside completely redundant mechanisms. - **A profound carry flaw:** In the drawings for the carry mechanism, the gearing was laid out in reverse: “The carry mechanism would not work not because it's not logical but because the wheels go in the wrong direction I mean it's that fundamental.” The team resolved this by mirror-imaging the original design, debating whether Babbage had made a simple drafting error, lacked oversight of his draftsmen, or intentionally inserted an anti-espionage trap to protect his intellectual property. - **Clutch redesign:** The paper rewind clutch, which automatically arrests the crank at the end of a page and rewinds the drive weights, suffered from excessive friction because it had to slide along a shaft while simultaneously resisting heavy rotational torque; the team redesigned it with a modified mechanism within the exact same physical envelope. - **Reversible modern additions:** All team modifications—including vertical counterbalancing weights on the racks, frame-reinforcing brackets, the printer crank handle, and a lower counterbalancing spring modeled after an authentic original spring—were made completely removable to safeguard historical authenticity.

---

[59:46](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=3586s) Project Budgets: Nineteenth-Century Treasury vs. Modern Reconstruction

To conclude the session, an audience member inquires how Babbage’s historical budget compares to the cost of the modern replica. Swade recounts that between 1822 and 1832, Babbage had what amounted to a blank check from the British government, ultimately spending £17,500 of public funds. Swade calculates that sum to be roughly equivalent to $200 million today. In return for this vast expenditure, the British Treasury received virtually nothing functional—only a completed assembly representing roughly one-seventh of Difference Engine No. 1, which the government eventually returned to Babbage.

While Swade cannot provide an exact, all-inclusive balance sheet for the modern reconstruction—explaining that his own institutional salary and historical research hours were absorbed into the Science Museum’s general operational budget rather than billed separately—he reveals the machine's true financial scale through its replacement appraisal. For full loss, destruction, or irreparable damage, the insurance valuation of Difference Engine No. 2 stands at between $2.5 million and $3 million.

[01:00:33](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=3633s) Costs, Comparisons, and Taking Orders for Google's Foyer

Reflecting on the financial scale of recreating Charles Babbage's mechanical marvel, Doron Swade explains that with the knowledge and expertise gained from their historical reconstruction, a Difference Engine could now be built from scratch for approximately that sum. He playfully contrasts their modern achievement with Babbage's original, ill-fated government-funded endeavor, noting that Babbage managed to spend roughly the equivalent of $200 million in contemporary terms without ever finishing a complete machine. By comparison, the modern team achieved a fully working engine for a fraction of that astronomical budget. Swade wraps up the lecture with an invitation to the tech giant's audience, quipping: “We did it for substantially less and if Google wants one for the foyer I'm taking orders.”

With the talk drawing to a close, the Google host steps up to bring the session to an end, acknowledging the tight schedule: “We're actually out of time so I want to thank you so much for coming to Google today thank you.” The presentation concludes amid enthusiastic applause from the audience.

- **Comparative expense:** Babbage expended the equivalent of roughly $200 million without completing the engine, whereas the modern team successfully engineered and built a working unit for substantially less money. - **Commercial readiness:** With the accumulated manufacturing drawings, tooling knowledge, and engineering insights, the team was positioned to replicate the engine from scratch upon request.

## Notable quotes > [42:26](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2546s) Authenticity was the watchword of the project; we had to resist the charge absolutely of somebody saying, 'You built the engine, but Babbage could not have. — Doron Swade

> [19:48](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=1188s) You did not need to understand how the machine worked or the principle upon which it's based in order to get useful results, and this was unprecedented. — Doron Swade

> [53:36](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=3216s) Babbage challenges you and says walk up to the machine even when it's functioning and attempt to derange it externally and you won't be able to. — Doron Swade

> [48:24](https://www.youtube.com/watch?v=7K5p_tBcrd0&t=2904s) You've eliminated the human from the loop entirely, from calculation to the production of stereotype plates. — Doron Swade

## Mentioned Charles Babbage, Doron Swade, Difference Engine No. 1, Difference Engine No. 2, Analytical Engine, Science Museum in London, Computer History Museum, Computer Conservation Society, Royal Holloway, University of London, The Cogwheel Brain, Thomas Carlyle, Hewlett-Packard, Napoleon Bonaparte, Georgiana Whitmore, Royal Society, Sir Thomas Lawrence, National Portrait Gallery, On the Economy of Machinery and Manufactures, Antoine Claudet, Sir Robert Peel, Fourth International Statistical Congress, Lucasian Chair of Mathematics, Alfred, Lord Tennyson, The Vision of Sin, John Herschel, Royal Observatory, Greenwich, Astronomical Society, Harry Wilmot Buxton, Ada Lovelace, Lady Byron, Alan Turing, Joseph Clement, Google, Mechanical Notation, British Treasury

## Video description In 1821, Charles Babbage was reviewing a set of mathematical tables with a colleague in preparation for a scientific presentation when, after finding a wealth of errors, he exclaimed in frustration, "I wish to God these calculations had been executed by steam!" With this outburst, Babbage began to envision an end to human errors in the numerical tables upon which finance, trade, science, and navigation relied. The Difference Engine is the fascinating story of his heroic quest, against all odds, to build the first computing machine more than one hundred years before the modern computer we use today was invented.

Doron Swade, an engineer, technology historian, and leading authority on Charles Babbage, was previously the assistant director and head of collections at London's Science Museum and is now Director of the Babbage project. He is also the mastermind of a seventeen-year project to construct a Babbage calculating engine from original nineteenth century designs.

This talk took place on May 8, 2008 at Google's Mountain View, CA headquarters as part of the Authors@Google program.

[Source](https://www.youtube.com/watch?v=7K5p_tBcrd0)

中文简报

查尔斯·巴贝奇与二号差分机 | 多隆·斯韦德 | 谷歌演讲录 多隆·斯韦德博士详细介绍了查尔斯·巴贝奇的生平、性格与机械创新,将他的计算机阐释为工业革命从体力劳动延伸至脑力劳动的哲学拓展。斯韦德解释了巴贝奇如何在电子计算技术出现数十年前,便预见到了计算机科学的基础概念——包括数字十进制架构、流水线、机械脉冲整形以及微程序控制。随后,他回顾了长达十七年的二号差分机雄心重构项目,展示了其团队如何解决图纸错误、适应19世纪的冶金标准,并制造出该机器复杂的微程序控制打印与立体铸版装置。 在伦敦科学博物馆多隆·斯韦德博士的领导下,历经长达17年的项目,由大约8000个零件组成、重逾5吨的查尔斯·巴贝奇二号差分机工作复制品成功建成。 (0:09) 巴贝奇的计算机提前一个多世纪体现了现代数字架构,采用了离散十进制状态、机械脉冲整形楔锁、流水线加法周期以及先行轮询进位。 (24:04) 制造该机器的过程揭示了巴贝奇19世纪40年代原始图纸中存在的细微档案设计缺陷,包括反向的进位齿轮以及未配备安装支架而悬浮在空中的零件描绘。 (55:40) 巴贝奇最初的一号差分机项目在取消前耗费了英国财政部17,500英镑,而斯韦德估计现代复制的成本大约在250万至300万美元之间。 (59:46)

Source quotes

[42:26] Doron Swade: Authenticity was the watchword of the project; we had to resist the charge absolutely of somebody saying, 'You built the engine, but Babbage could not have. [19:48] Doron Swade: You did not need to understand how the machine worked or the principle upon which it's based in order to get useful results, and this was unprecedented. [53:36] Doron Swade: Babbage challenges you and says walk up to the machine even when it's functioning and attempt to derange it externally and you won't be able to. [48:24] Doron Swade: You've eliminated the human from the loop entirely, from calculation to the production of stereotype plates.

Original transcript

[0:09] hello everybody um so as probably many [0:12] people around here I'm sort of [0:14] fascinated by TR time travel and I was [0:17] thinking of how to introduce doron and [0:19] this project and it occurred to me that [0:21] if you just walk in this direction about [0:23] one mile away from this campus you can [0:26] actually travel back in time 170 years [0:29] and see [0:30] uh was was the first mechanical [0:32] calculator based on the automatic [0:35] calculator based on the first computer [0:37] technology consisting of 8,000 parts and [0:40] weighting 11,000 pounds so uh this is a [0:45] quite a travel to the past and what's [0:47] more interesting is actually a past that [0:48] never existed uh Charles bage that that [0:51] conceived of this machine of the [0:53] Difference Engine Number Two he never [0:55] actually finished it uh during his [0:57] lifetime in the 19th century England so [0:59] so um and it took 150 years and one [1:04] dedicated person to actually you know [1:06] make bab's dream a reality and uh that [1:10] person is Doran sued standing here I'm [1:13] going to read this paragraph because [1:15] it's kind of complicated so D SW is an [1:17] engineer a historian and a museum [1:19] professional he he has a bachelor's [1:21] degree in physics and electronics [1:23] engineering master's degree in control [1:25] engineering and PhD in a history of [1:28] computing uh founded a computer [1:31] conservation Society specializing in [1:34] preservation of early computers uh he's [1:36] currently the assistant director and [1:38] head of collections of the science [1:40] museum in London he's a guest creator of [1:42] the aformentioned Computer History [1:44] Museum and an honorary research fellow [1:47] at the Royal hallay University of London [1:50] and since this is the authors at Google [1:52] Talk um Doran is you know you can [1:55] suspect he's also an author and his [1:56] latest book called the kwick hog wheel [2:00] brain in the UK or the Difference Engine [2:03] in the US uh talks about his experiences [2:06] in uh well talks about the B's life and [2:09] his experiences in recreating the [2:11] Difference Engine Number Two and uh [2:13] people who know me know that I have an [2:15] insane amount of computer history books [2:17] at my apartment it's like 700s by now [2:20] doran's book is still one of the ones [2:22] that I love most you know it's probably [2:25] in the top five so um yeah it gives me a [2:28] great honor and pleasure introduce to [2:30] you Dr Doran [2:37] sued um Maran thanks very much indeed um [2:40] as marchan mentioned um what prompts [2:43] this talk is the arrival in Silicon [2:45] Valley of this 5ton calculating engine [2:47] which is down the road at the Computer [2:48] History Museum it will be on display [2:50] there for one year it'll be demonstrated [2:52] three times a week if it recovers from [2:53] as jet lag um it was flown over the [2:56] Atlantic and suffered very badly there [2:58] was corrosion from low temperature high [3:00] altitude um it was shaken up very badly [3:02] it was not um actually intended when it [3:04] was built started building it 6 years [3:05] ago to travel um but it is running and [3:09] uh there is no substitute for actually [3:10] seeing the engine [3:12] live um well what can I do in 30 40 [3:16] minutes for you I thought what I would [3:17] do is tell you something about bab's [3:19] life and times what was the spirit of [3:20] his times what was the Zeitgeist of the [3:22] 19th century there are common features [3:24] with our own um tell you something about [3:27] his engines just generally what was he [3:28] trying to accomplish when he set about [3:31] spent his life's Mission as it were from [3:33] 1821 onwards attempting to design well [3:36] designing succeeding and attempting to [3:38] build these extraordinary these vast [3:40] machines and finally um and the bulk of [3:43] the talk will really be on Difference [3:45] Engine Number Two um I sense both from [3:47] feedback from Maran and other otherwise [3:49] that the audience is has technical [3:52] interests so I will keep the history not [3:54] to a minimum but I've tended to bias the [3:57] whole treatment towards the engine and [3:59] to show you and say things to an [4:01] audience um that could is perhaps more [4:03] appreciative the technical aspects than [4:04] would be um too strictly to dwell on the [4:06] historical [4:07] elements so [4:11] um B's world the 19th century we pride [4:15] ourselves now that we live in an age of [4:17] prolific invention barely a day goes by [4:20] when there isn't some report in a [4:21] newspaper about some breakthrough some [4:23] silver bullet some new [4:25] innovation um it seems to be a signature [4:27] of our times pace of change accelerated [4:29] rate of innovation and so on this is a [4:31] daily Motif particularly for Silicon [4:33] Valley and elsewhere um the victorians [4:36] shared this rather freni sense of change [4:38] the middle Decades of the 19th century [4:40] were a time of unprecedented engineering [4:42] ambition it they were obsessed with [4:44] contrivances devices Contraptions [4:46] mechanical things machines uh Thomas [4:49] Carlile The sist said that it was the [4:51] age of Machinery in every outward and [4:54] inward sense of the word outward and [4:56] inward it was not just the conquest of [4:58] Nature and the external World by techn [4:59] ology it was also an inward sense they [5:01] were the terms in which mechanism were [5:03] the terms in which people understood the [5:05] external World it seemed that there was [5:07] no predicate in nature or in human [5:09] experience that could not be exhausted [5:11] by mechanical explanation understanding [5:14] became co-terminal with mechanical [5:16] explanation and this was Baba's world [5:18] this was the world he lived [5:20] in there seemed no end to [5:23] the prospects for the industrial Arts [5:25] with new materials and new processes and [5:28] manufacturers and products poured out of [5:30] the manufactures and burdened Shop [5:32] shelves with new products new things so [5:35] it was a time as I say of frenzy change [5:37] and huge sense of [5:42] innovation the devices of the Industrial [5:44] Revolution have one thing in common they [5:46] were designed to relieve or replace [5:47] physical [5:50] labor but there was another revolution [5:53] in the making that is less well known [5:55] possibly because it failed and that is [5:56] the revolution that is the replacement [5:58] of mental labor specifically the [6:00] mechanization of calculation through [6:03] machines the idea is that if a machine [6:05] was autonomous if it was automatic and [6:07] you could embody mathematical rule in [6:09] mechanism then you could achieve by [6:12] physical exertion with something you [6:14] could only achieve up to that point in [6:16] time by mental effort and the idea was [6:19] that this was a thinking machine the the [6:22] movement failed the pioneer of that [6:23] movement was Charles babage English [6:25] mathematician born [6:27] 1791 died 1871 [6:30] and his purpose in a sense was to extend [6:34] the model of industrial production from [6:36] thing to thought from thing to thought [6:39] to extend the model of industrial [6:40] production that as his son said his [6:42] machine would be a manufacturer of [6:44] numbers this was an [6:46] extension of the industrial model the [6:49] model of production the solution to the [6:51] problem of Supply into the realm of [6:54] mental effort [6:57] so what have we got here with this steam [6:59] horse this is an example of something [7:01] Slightly bizarre the front end it's a [7:03] steam it's a steam Railway a street [7:06] Railway with a front end built on with a [7:08] horse this is literally not to frighten [7:10] the horses we have here an example of [7:12] bridging technology of absorbing the new [7:15] into the comfort zone of the familiar [7:17] and we have similar examples in modern [7:19] times the earliest electronic [7:20] calculators were advertised as [7:22] electronic slide rules this is a way if [7:24] you like of leveraging something new [7:26] into the comfort zone into the [7:28] familiarity of something already [7:31] known so very [7:33] briefly a thumbnail sketch of babage who [7:36] was he what was he trying to do where [7:37] did he fit into into London and English [7:43] society this is babage as a student in [7:46] Cambridge in 1813 he graduated [7:48] mathematics he was a crack man in [7:49] mathematics he tried to reform [7:51] mathematics at Cambridge which he [7:52] regarded as very stuffy so we have here [7:55] already the notion of a reformer someone [7:57] who was Radical the resemblance in a [7:59] Napoleon is not accidental England was [8:01] at war with France babage supported [8:03] Napoleonic France this is 1813 the [8:05] Napoleonic Wars ended in 1815 so already [8:08] you have a sense of radicalness sense of [8:09] boldness a sense of defiance in him [8:12] that's Georgiana whom he married in 184 [8:15] uh 1814 the year after this this is a a [8:18] locket in the family um in in the babage [8:21] family um in defiance of his father who [8:24] thought it was imprudent for a man to [8:26] marry without independent means his [8:28] father was a wealthy Banker [8:33] we have babich here age 53 as the man [8:37] the gentleman of science the man of [8:38] science enjoying the esteem of his peers [8:39] this is an oil painting by Lawrence in [8:41] the National Portrait Gallery he had [8:43] already been partially rehabilitated the [8:45] stage he had um shocked British [8:49] scientific Society by his sarcastic diet [8:51] tribes his public attacks against the [8:53] Royal Society for their poor governance [8:55] of Science in England he was capable of [8:57] incontinent savagery in in these public [8:59] attacks he would impune the personal [9:01] Proby of its senior officers and he had [9:03] no compunction and but to do this in [9:06] public this was about 1830 1832 he some [9:09] sense redeemed himself by writing an an [9:11] encyclopedic volume called the economy [9:13] of machinery and manufacturers which is [9:15] a turning point in political economy [9:17] where he examined the political economy [9:19] of factories this was a groundbreaking [9:21] piece of work so already by this stage [9:22] he'd been rehabilitated 1830 he was the [9:24] inant of of London Society even though [9:27] at that stage he was aged [9:30] 40 this is babage and its particular [9:32] interest to us because it's around the [9:34] time he designed Difference Engine [9:35] Number Two which was designed between [9:36] 1847 1849 so this is a portrait by Cloud [9:40] a photographic so there's some [9:41] confidence of realism um within the [9:44] constraints of Photography and there we [9:46] have somebody still beaten up he' [9:48] finished with his he' had his flaming r [9:49] with peel he'd more or less given up any [9:52] Prospect of completing any of his [9:53] machines but still there is some fight [9:55] left in him he's still designing he's [9:57] still tinkering he's still producing [9:58] different Engine 2 which is perhaps a [10:01] Masterwork in sophisticated and um um [10:04] the economy of elegant [10:06] design and this is babage that's the [10:09] last known portrait of babage taking at [10:10] the fourth statistical conference in [10:11] London 1860 this is what I call his [10:13] Beethoven look this is the venerable [10:15] Grump this is the um this is the um his [10:19] first biographer called him the erasable [10:20] genius and for all the scholarship since [10:22] this was 1963 for all the scholarship [10:25] since those two twin things the [10:26] erasability and the genius are perhaps [10:29] the most um compelling um portrait of [10:33] babage um despite all the other things [10:35] that are now known about [10:36] him um protest always triumphed over [10:39] persuasion he was a rur a socialite he [10:41] was a sort after dinner guest Mr babby [10:43] just coming to dinner was a coup for any [10:45] Hostess you were assured of [10:47] entertainment and sparkling conversation [10:49] this was a bold man of great originality [10:51] of startling sweep he was a man of [10:54] fierce loyalties his friends could do no [10:55] wrong his enemies could do no right he [10:57] achd for honors and and complained that [11:00] the Lucasian professor of mathematics [11:01] was the only honor bestowed on him by [11:03] his country he was offered a Knighthood [11:05] and refused it saying this was to go to [11:07] the engine not to him and since he had [11:09] not built the engine then um he was [11:11] unable to communicate um his vision and [11:15] we can see when we do get into the [11:16] Difference Engine 2 the Sumptuous piece [11:18] of engineering sculpture of just how [11:21] bitter he must have been to have had [11:22] this vision and being unable through the [11:25] through his failure to complete his [11:26] engines um to communicate to any of his [11:28] contemporary is so let me give you some [11:30] flavor of his [11:31] writing this is babage in 1851 writing [11:35] in an obscure paper on taxation one of [11:38] the factors that affected whether he did [11:40] or did not complete his machines was the [11:42] entrepreneurial climate climate in the [11:44] mid middle Decades of the 19th century [11:46] so this is babage writing on [11:47] entrepreneurial climate again in some [11:49] obscure paper propos to any Englishman [11:52] any principle or any instrument however [11:54] admirable and you will observe that the [11:56] whole effort of the English mind is [11:57] directed to finding a difficulty a [11:59] defect or an impossibility in it if you [12:02] speak to him of a machine for peeling a [12:04] potato he will pronounce it impossible [12:06] if you peel a potato with it before his [12:09] eyes he will declare it useless because [12:11] it will not slice a [12:12] pineapple in part the same principle to [12:15] show the same machine to an [12:16] American or to one of our colonists and [12:19] you will observe that the whole effort [12:20] of the mind is to find some new [12:22] application of the principle and some [12:23] new use for the instrument so already [12:25] you have the familiar refrain at least [12:27] in Europe that we have European [12:28] inovation versus if you like United [12:31] States commercial exploitation of [12:33] markets this is babage to Tennison the [12:35] poet [12:36] Tennison excuse [12:45] me Bish to Tennison sir in your [12:49] otherwise beautiful poem the vision of [12:50] sin there is a verse which reads quote [12:53] every moment dies a man every moment one [12:55] is born close quote it must be manifest [12:58] that if this were true the population of [12:59] the world would be at a [13:02] standstill I would suggest that in the [13:04] next edition of your poem you have it [13:06] read quote every moment dies a man every [13:08] moment one and 116th is [13:11] [Music] [13:12] Born the actual figure is so long that I [13:15] cannot get it onto a line but I believe [13:17] that the figure of 116th is sufficiently [13:19] accurate for [13:23] poetry now we have to believe that [13:25] babies wrote this tongue and cheek the [13:27] terrifying thought is that he was deadly [13:30] serious onto his engines you have some [13:33] flavor of who this man was a bold [13:35] vigorous radical man proud principled [13:39] easily taking offense tilting at [13:41] windmills never hesitating to publicly [13:44] denounce people whom he think it wrong [13:45] he he behaved as though being right [13:47] entitled him to be rude and I think [13:50] that's perhaps characterizes his public [13:52] Behavior so [13:55] engines the Genesis episode is 1821 [13:58] babby Jen great friend hersel are [13:59] sitting in London checking mathematical [14:02] tables that is tables computed by [14:05] computers computers were human beings [14:06] they did the lowlevel calculations [14:08] they're sitting checking astronomical [14:10] tables um babage becomes increasingly [14:12] dismayed by the number of Errors between [14:15] the two U lists manuscripts that were [14:17] being compared and clasps his hand to [14:19] his head and says I wish to God these [14:22] calculations had been executed by steam [14:25] steam being a metaphor both for the [14:27] infallibility of the Machinery the aing [14:29] certainty of mechanical agency would [14:31] ensure the Integrity of results machines [14:34] don't make mistakes humans do but it's [14:35] also metaphor for a model of production [14:38] it would solve the problem of Supply [14:40] astronomers were seeing asteroids and [14:42] thought they were new planets and they [14:43] needed to plot the trajectories and a [14:44] pressing government to insist that the [14:46] Gren Ro Observatory produced these [14:48] tables grench Observatory was resistant [14:50] they said it was outside its mandate and [14:52] Charter so the Astronomical Society [14:54] which Bach was actually instrumental in [14:56] founding um charged commissioned hersel [14:59] and babage to do these tables so there [15:01] was a huge political context in doing [15:02] this why were they sitting there [15:03] checking tables but this was the [15:05] mechanical Epiphany from that point on [15:07] bab's life was dominated by the wish the [15:10] need the aspiration to build these vast [15:13] mechanical calculators that would [15:14] eliminate all sources of error from [15:16] printed producted production of [15:20] tables this the errors in tables were [15:22] three were four-fold firstly there was [15:23] the error of calculation manual [15:25] calculation secondly there was the error [15:26] of transcription you had to copy the [15:27] results down so that a type Setter could [15:30] set them in loose type secondly there [15:32] thirdly there is the problem of setting [15:34] each digit in loose type this is not [15:36] like language where there's some [15:37] intuitive idea of what a spelling [15:38] mistake or an omitted word or a [15:40] scrambled word or reversed syllable um [15:43] there is no intuitive way of knowing [15:45] whether a number is wrong by inspection [15:48] although types Setters of course were [15:50] encouraged not to read what they were [15:51] types setting they did it actually quote [15:53] mechanically they encourage not actually [15:55] because one mind actually skips it's [15:57] very difficult so um um finally there [16:00] was the question on verification how [16:02] what confidence did you have once the [16:03] tables were printed that these were [16:06] accurate so there were four sources of [16:07] possible error and the idea was the [16:09] great the genius if you like was a [16:12] system a complete system that would [16:13] eliminate all forces there there would [16:15] be an automatic typ set it would print [16:16] automatically but not just a single ink [16:18] copy it would print stereotypes plates [16:20] automatically as a mechanical automatic [16:22] outcome that would allow you to [16:24] reproduce results without human [16:25] intervention and we shall see that the [16:27] implementation in different two [16:29] accomplishes precisely this so 1821 baby [16:32] sets about and he spends 10 years [16:33] designing and developing Difference [16:34] Engine number one that has um 16 digits [16:37] and six orders of difference it's a [16:38] Difference Engine because of the [16:39] mathematical principle on which it's [16:41] based with the me method of finite [16:43] differences the beauty of the method is [16:45] to evaluate a polinomial ordinarily you [16:47] need to do multiplication division [16:48] subtraction and addition the beauty of [16:50] the method of differences is you can get [16:51] the same results and tabulate the [16:53] answers by repeated addition only so it [16:56] does you you can eliminate the need for [16:58] mulp application IND division which are [16:59] very very difficult to implement [17:00] mechanically he did subsequently but [17:02] certainly not in 1820s so that's the [17:04] plan 1830s plan of Difference Engine one [17:07] it um would have weighed 50 it had [17:10] 25,000 Parts would have weighed an [17:12] estimated 15 Tons it was something like [17:15] 8 ft High um and ran ran on railway [17:18] tracks actually it was so heavy um this [17:21] is all he ever accomplished this is 17th [17:23] of his um of the full machine and this [17:26] exists it's in the science museum in [17:27] London and the significance of this [17:29] particular machine cannot be overstated [17:32] this is the first automatic mechanical [17:34] calculator you operate a [17:38] handle up there you crank it you set the [17:41] initial values on the wheels you crank [17:43] the handle and it's um you get results [17:46] which up to that point in time you can [17:47] only get by thinking this is the first [17:49] successful machine which is any sort of [17:51] autonomy this is the first successful [17:53] transference of human intelligence to [17:55] mechanism the first machine that [17:56] successfully embodies mathematical rule [17:58] in mechanism reversing that it's the [18:01] first it's the first machine that allows [18:03] the ingression if you like of Machinery [18:05] into [18:06] psychology there are other significance [18:08] to this machine he used it to [18:10] demonstrate his theory of Miracles he's [18:12] he he he he programmed the machine for [18:15] discontinuities that is to say after a [18:17] fixed number of Cycles he would [18:18] apparently break the rule and do [18:19] something there would be a disjunction [18:20] in what it did and babage um he [18:23] demonstrated this at his sarees um to [18:25] the glitter of London and he would argue [18:28] that um that discontinuities in nature [18:32] which religion called miracles [18:33] discontinuities in nature actually were [18:35] programmed were programm discontinuities [18:37] and while it seemed a violation of law [18:39] to us to of natural law to us The [18:42] Observers it was actually a [18:43] manifestation of a higher law God's law [18:45] unknown to us unknown to us and by [18:48] analogy he had programmed the thing for [18:50] a discontinuity fixed number of cycles [18:52] and by analogy therefore miracles in [18:54] nature were not uh discontinuities in [18:56] nature were not necessarily Miracles but [18:58] the product of causal um determinism and [19:02] this sent a massive Ripple through [19:03] geologists at the time who um and [19:06] predian evolutionists who were [19:08] struggling for a model of discontinu why [19:11] you could have eons with no new species [19:13] and the sudden appearance of a species [19:15] while you had earthquakes which forces [19:17] which could only be attributed to some [19:19] deity given their magnitude how would [19:21] you account for this how would science [19:23] compete with natural theology if it was [19:25] wrestling for ownership of an account of [19:27] the world and this model that is a very [19:30] good example of how an object has [19:32] influenced the history of ideas the fact [19:34] that you could have an autonomous [19:35] machine not just a speculation about [19:37] mathematical Singularity but a machine [19:39] that behaved um in a programmed way but [19:42] exhibited [19:44] discontinuity seem to to to provide [19:46] compelling evidence for the fact that [19:48] the world could be deterministic and [19:50] they are and they are a natural [19:52] phenomena that you could account for [19:53] without necessarily appealing to a deity [19:56] so for babage God was a program [19:58] and those in the software industry may [20:00] be flattered to know [20:05] this the significance that the machine [20:07] was thinking was not lost on babage or [20:09] his [20:11] contemporaries um Harry Wilmer buckton a [20:13] junior colleague of babbes wrote The [20:15] Marvelous pulp and fiber of the brain [20:17] had been replaced by brass and iron he [20:19] babage had taught wheelwork to think or [20:22] at least to do the office of thought [20:24] lady Byron that's adah love LA's mother [20:26] wrote in a diary in 183 three last week [20:29] we saw the thinking machine for such it [20:31] seems so this was not lost on them this [20:33] was if you like the beginning of What [20:35] Not only artificial intelligence but [20:37] also the beginning of automatic [20:39] computation you cannot have either of [20:41] these Without The Machine being [20:43] autonomous in some way and this is the [20:45] machine that symbolizes that fundamental [20:47] transition from speculation to actual [20:50] physical autonomy you did not need to [20:53] understand how the machine worked or the [20:54] principle upon which it's based in order [20:56] to get useful results and this was [20:57] unpreceded precedented the machine [20:59] itself is a is a is a Quantum Leap in [21:02] logical conception and physical size [21:04] there was nothing comparable anything [21:06] like it it came in a sense from [21:08] nowhere this is babage this is AD love [21:11] this is herel that is the actual very [21:13] machine and this is babage demonstrating [21:14] his theory of miracles in his very [21:16] famous sarees on Saturday and the reason [21:18] it slightly blurred is because digital [21:20] photography was not very good in the [21:22] 19th [21:27] century the vignette I've painted of [21:29] babage and hersel is misleading and [21:31] historians myself included endlessly [21:33] Charmed by it have been misled into [21:35] thinking that errors in tables was not [21:37] only the jumping off point which it was [21:39] but that it was the enduring motive for [21:41] babbage's designs babage for Babbage the [21:43] engines were a new technology of [21:45] mathematics his earliest writings do not [21:47] mention errors errors enter only as a [21:49] device of persuasion as a rhetorical [21:51] device when he had to justify these [21:53] machines to the outside [21:55] world in U utilitarian terms terms in [21:58] order to get support bab's really [22:00] interest was mathematics he saw the [22:02] engine as a new technology to do [22:06] mathematics and there are at least three [22:08] respects in which this was specifically [22:10] so he saw computation as a systematic [22:13] method of solution that you could find [22:15] the solution for equations for which [22:16] there was no analytic [22:18] formula he saw the machine as a [22:21] heuristic device you can cross [22:24] couple you can cross couple both forward [22:28] and back elements from columns you can [22:30] see their external shafts here which [22:31] allow a value from that wheel to be [22:33] coupled to a value from that wheel using [22:35] these external shafts of the external [22:36] coupling so you could cross feedback you [22:38] could produce series for which there was [22:39] no General solution for the nth term so [22:42] you saw this as a heuristic device he [22:44] saw for the first time computation of [22:46] sastic meth method of solution so the [22:48] root of an equation is when the result [22:49] goes to zero all you need to do is to [22:51] detect the all zero state in the last [22:53] column and the Machine will halt halt [22:55] resonances of touring he talks about the [22:57] machine halting when it finds a root he [23:00] says crank it if there is no solution to [23:02] the thing it will crank at infinitum [23:05] this is touring pre-echo of touring in a [23:07] solution to the problem of computability [23:09] using halting criteria this is explicit [23:11] in bage he also predicts a new branch of [23:14] mathematics called numerical analysis [23:16] which will be devoted entirely to the [23:18] optimization of mathematical formula for [23:20] the purposes of machine computation and [23:22] the example he uses he takes a formula [23:24] and he said to find the value of this [23:25] you will do 35 multiplications 10 [23:26] divisions and two additions he then [23:28] rearranges the formula which is [23:30] mathematically identical and says you [23:31] can do the same thing with five [23:33] additions and two multiplications and [23:34] one Division and he said there would be [23:36] a branch of mathematics that would [23:37] ensure that that that mathematics these [23:40] formula were pre-prepared to optimize [23:42] machine computation so this was [23:44] babbage's version new it has been if you [23:47] lik as I say historians been misled by [23:49] the vignette and other things there were [23:51] errors in tables the erat sheets in [23:52] published tables um absolutely evidenced [23:55] this but this was not bab's original [23:57] conception [24:04] the machines are decimal digital [24:05] machines they're decimal in the obvious [24:07] sense they use the numbers n to N9 they [24:08] are digital in the sense that only [24:10] discrete values of whole numbers are [24:12] legitimate so if a wheel is intermediate [24:15] between two numbers then the machine [24:19] jams it is designed to Jam jams are good [24:21] it's a form of error detection it says [24:24] the Integrity of the result has been [24:25] compromised a wheel is in in termined [24:27] position so the the number 2.5 is not [24:29] represented by a wheel numbered in a [24:32] value between two and three you have a [24:35] wheel which has five and a wheel which [24:36] has two and we'll show you that [24:37] presently each of the digits actually [24:39] has its own wheel it's a decimal digital [24:41] machine now what makes it digital this [24:43] is not inherently digital this is not [24:45] electronic flip-flop logic these are not [24:47] by Stables where you exceed a threshold [24:49] and the thing flips because the [24:50] transitional states are not stable bab [24:53] it's it's it's digital because the [24:56] control mechanism makes it so and I'll [24:58] show you an example of babbage's early [25:00] attempts here you can see babage trying [25:02] to digitize this you can see that these [25:04] half wounds that thing's on a spring it [25:06] has two actions it locks when it's fixed [25:08] and freezes that wheel but also as the [25:09] wheel goes the spring attempts to buy [25:12] okay now the dwell angle over there the [25:14] dwell angle is actually not sharp enough [25:17] to make a decisive transition but you [25:18] can see the beginnings of attempting to [25:20] digitize that this is stable that is [25:22] stable and that is Transitional and you [25:24] will know when that is proud of the [25:27] indent whether it is actually in the [25:29] correct position or not so you can see [25:31] the beginning of an attempt to digitize [25:33] uh this machine even though it was [25:34] desimal I'm now going to show you okay [25:37] um let me read you just some of the [25:39] features let me read you some of the [25:41] features of different engine one these [25:43] are logical and system features and I [25:44] should emphasize nowhere did babage ever [25:46] use the terms I'm about to use they're [25:48] not abusive um but I'm this is a kind of [25:52] backward backwards projection from our [25:54] own age to actually make it intelligible [25:55] so this is different engines and these [25:58] are circuit what we would Now call [25:59] circuit and system functions it includes [26:01] a parallel bus microprogramming that is [26:03] the automatic execution of the Min [26:04] instructions given a macro instruction [26:06] pipelining pulse shaping binary latching [26:10] polling IO certainly and a bounce [26:12] catcher schm trigger to actually stop [26:14] something bouncing which was added so [26:16] what I'm going to show you now is okay [26:17] how do you do this mechanically again [26:19] nowhere did babit use these word pulse [26:21] shaping the cleaning up of an edge which [26:23] becomes degraded which is something we [26:25] commonly do in electronics so I'm going [26:27] to you a video now of how this thing [26:28] pulse [26:33] shapes okay let me stop it and explain [26:35] something this is a lock it's a sword [26:38] blade thing with a wed shaped device at [26:39] the end that's wed shaped at the end [26:41] here and the lock is going to go in and [26:43] actually into the between the teeth of [26:45] these horizontal racks now the lock has [26:47] three functions in babby designs so the [26:49] locks are the mechanism he used to [26:51] digitized to keep this machine digital [26:53] it's part of the control system the lock [26:54] has three effects firstly if you get a [26:56] wheel that deranges SL slly the wed [26:58] shaped lock comes in it has a [26:59] self-correcting action right if it's a [27:02] slight derangement it will correct and [27:03] reenter it the lock when it's fully [27:05] inserted will stop the wheel moving so [27:07] therefore it's not vulnerable at [27:08] particular part of the cycle where it [27:09] might otherwise be deranged finally if a [27:12] wheel drifts more than two and a quarter [27:13] degrees either through drift derangement [27:15] wear or anything else the lock will not [27:17] go in and will hit the to end of the [27:19] tooth and the Machine jams the machine [27:21] is telling you the wheel is in [27:22] interminate position it's between two [27:24] digits and you should stop because the [27:26] Integrity of the result has been [27:27] compromised this is one example of how [27:29] this works so this is the vertical lock [27:31] and it's going to insert into those [27:33] racks now if you watch that very [27:34] carefully as it inserts you can see the [27:36] self correction [27:39] action there should be sound on [27:49] this [27:51] okay I don't know if you can see that [27:53] correcting action there as it goes [27:56] in [27:58] you can see that slight [28:01] shift [28:10] yeah that's it okay let me run it again [28:13] the sound is actually quite [28:14] interesting okay so as the lock goes in [28:16] if you just watch one of these you'll [28:17] see there see the [28:23] correction correction right it's self [28:26] correcting each time slightly [28:27] Arrangement it goes in okay that's [28:29] fairly [28:31] clear um right I mentioned one bit [28:34] latching okay one of the key things [28:36] about it is when when a figure wheel [28:38] goes above 10 it has to increment the [28:39] next decade to carry Carriage of 10 [28:42] crucial mechanism now what's interesting [28:44] about this is that he he adds all the [28:46] wheels together in one go and if a wheel [28:48] goes above 10 instead of incrementing as [28:50] part of that action of going above 10 it [28:52] sets a latch and says this wheel has [28:53] gone above 10 in the next part of the [28:55] cycle you must increment the next one [28:56] otherwise you'll get incorrect result so [28:58] what it does is has a one bit latch the [29:00] latch says this wheel in this wheel went [29:02] above 10 in the last cycle so these are [29:05] the latches these are the wheels so [29:07] that's unwarned that's warned and that's [29:09] carried what we're going to do is run [29:10] this what happens is that this Helix [29:12] sweeps around and these um Carry levers [29:16] pole Each decade in [29:18] turn right it interrogates each latch in [29:21] turn up the up the register and if it is [29:23] set if the latch is set it will [29:25] increment by one if it doesn't it misses [29:26] it completely so if we watch the [29:28] intersection the intersection of the [29:29] locus of this shoe on that lever that [29:32] lever these levers then when they're [29:35] warned they will intersect and carry [29:37] when they're not warned and we'll run it [29:38] now and you can see how that [29:41] goes okay now watch watch this one for [29:48] instance carried right I'm going to stop [29:50] it and try and catch it in the [29:56] middle okay that's scary I'm going to [29:58] try and catch it so you [30:03] can too [30:11] light got it okay it jumped from there [30:15] to there the the latch is set now you [30:16] can see that the extent the these arms [30:19] are in different [30:20] positions right one's further out one's [30:22] further in one will set one will not as [30:25] it sweeps around the intersection of the [30:27] Lo [30:27] so that's a conditional device which [30:30] interrogates each latch if set carries [30:33] if doesn't you can see the [30:35] two and then [30:37] resets I don't that was very clear if I [30:40] can catch one then you can see it but [30:42] that's what's happened and the sound is [30:43] gone again [30:45] um yeah is it this [30:56] plug [31:18] okay got it [31:21] okay it's caught there and the car in [31:24] the warn position so unworn to warn so [31:26] the latch is set and you can see that [31:28] this one is set that one's half carried [31:31] this one isn't set [31:33] so these arms are different you can see [31:35] that sticks out slightly further it will [31:37] then intersect or not intersect [31:38] depending on whether the latch is set or [31:39] not so it will either sweep and the [31:41] intersection of the two lokai of those [31:43] two trajectories will increment the next [31:44] decade [31:47] up okay different engines are [31:50] calculating machines they're not [31:51] strictly general purpose computers in [31:53] 1834 babage conceived the analytical [31:54] engine which startlingly embodies almost [31:57] every significant feature of a modern [31:59] digital electronic computer this is his [32:01] 1940 plan this would be 6 to 8 ft long [32:04] 15t high these are registers these are [32:07] are columns of figure Wheels if you look [32:09] down on them each of them has 50 digits [32:10] in them Precision 50 digigit Precision [32:12] double Precision arithmetic results up [32:13] to 100 digits um there are 19 or 20 [32:17] registered here babage talks about [32:19] machines with entry-level machines with [32:21] 100 registers and Machines of a th [32:23] registers a 20 register machine would be [32:26] 10t long the entry level is 100 [32:29] registers so this is a machine of [32:31] monstrous proportions you're talking [32:32] about something vastly bigger than a [32:34] steam [32:35] locomotive um let me read you some of [32:38] the features um of well firstly the um [32:42] the the his attempts to build Difference [32:44] Engine Number One failed this was the [32:46] biggest Central trauma of his life he [32:48] never really recovered he returns time [32:50] and time again in his writings in dismay [32:51] in despair in anger in disbelief as [32:53] though unable to reconcile himself to [32:56] the um dismal outcome this was in [32:59] 1833 when Joseph Clement his engineer [33:01] walked out project was never resumed [33:03] again government withdrew funding in [33:05] 1842 if you like the S is of Victorian [33:08] soap there was a beautiful Countess a [33:09] flaming ra with the Prime Minister [33:11] public accusations of Prov could and [33:13] wasteful spending of public monies [33:14] experts who disagreed about the [33:16] fundamental utility of the machines and [33:18] a walk out by bab's chief engineer [33:20] Joseph Clement we relived all of these [33:23] in the uh modern reconstruction uh [33:25] except that I'm still waiting for the [33:27] beautiful [33:28] [Music] [33:30] Countess uh this is the plan of the [33:31] 1840s analytical engine conceived in [33:34] 1834 and um let me just read you some of [33:37] the logical features that this machine [33:39] embodies it's programmable using punched [33:42] cards up there we see the punch cards of [33:44] which there were at least four separate [33:45] types it has an internal architecture [33:47] foman architecture it separates store [33:50] and Mill that is processor from memory [33:52] it has a Serial fetch execute cycle it [33:54] has conditional branching if then [33:56] statements it is capable of iterative [33:58] looping it is capable of [34:00] microprogramming pipelining anticipating [34:03] Carriage mechanism which is not what we [34:04] call look ahead internal repetive [34:06] operations multiplication indivisual [34:08] automatic and internal uh uh uh features [34:11] a 50 to 100 digit prision it had [34:13] parallel processing it output was [34:16] printed stereotyped punch card and had a [34:18] graph plotter this is all in explicit [34:21] detail in his design now the point is [34:23] again nowhere does babage use these [34:25] terms so it is not a question of [34:28] suggestive hints this is not the coded [34:29] vagueness of Nostradamus this is not a [34:32] penu industry dignifying Itself by [34:35] claiming a false ancestry this is not [34:36] the backwards projection from our Modern [34:38] Age onto the blank canvas of the past [34:40] all these features are explicitly [34:42] embodied in the detail of design and [34:43] what I'll give you is just some flavor [34:45] of the extent and detail of what he did [34:48] this is the plan of the printing [34:49] mechanism which is shared between the [34:51] analytical engine and the Difference [34:53] Engine this is a device we've already [34:54] built um that's the one you can see down [34:56] the roow at the Computer History Museum [34:58] um up there is an example of his [35:00] mechanical notation this is a symbolic [35:03] Hardware description language developed [35:04] specifically to Aid and assist and [35:07] optimize the designs he was a [35:08] mathematician he thought naturally in [35:12] abstractions this is a um the tabular [35:15] form of his design all these alphabets [35:17] mean something they are anything up to [35:18] six superscripts and subscripts which [35:20] tell you the driving points the nature [35:22] what each part is connected to whether [35:24] the motion is circular linear continuous [35:26] disc continuous and so on this is so [35:29] there's three forms there's the tabular [35:30] form I've just shown you there's timing [35:32] diagrams this is his mechanical notation [35:33] there's the timing diagram which shows [35:35] how each parts are F the parts are [35:37] phased you can see the pipelining effect [35:38] here this is mirrored down there odds [35:40] and even axes are dealt with separately [35:42] that may not mean much now but actually [35:44] this is how he worked it all out he [35:45] doesn't use 360° he uses 50 divisions in [35:48] the cycle um which was actually not [35:50] precise enough to deal with the very [35:52] tight timing cycle so we produced [35:55] another one of these which has much [35:57] finer [35:58] phasing um and finally this is his flow [36:01] diagram this is um part of the this the [36:04] third form of the mechanical notation [36:06] and interestingly he has separated data [36:08] from control but they're on the same [36:10] drawing that is data input and that's [36:12] control and this device is um 4,000 [36:16] parts and that's the physical device [36:18] that that formulation corresponds to and [36:20] this is the one that's down the road [36:21] that's the printing and stereotyping [36:22] mechanism and we Ed the notation [36:24] actually to resolve issues that were not [36:26] clear from the [36:28] drawings we can't talk about the [36:31] analytical engine without talking about [36:32] AA love lace and much has been claimed [36:35] for her contribution four things are [36:38] usually claimed for aah firstly she was [36:40] a mathematical genius secondly that she [36:43] had an in influential contribution to [36:46] the analytical engine the baby's [36:47] analytical engine finally she was the [36:50] first thirdly she was the first [36:51] programmer and finally that she was a [36:53] prophet of the computer age there's only [36:54] one of these that has any substance at [36:56] all [36:58] firstly she was not a mathematical [36:59] genius she may have been but it was [37:01] unknown she was a novice she was a [37:03] promising beginner she made a [37:05] fundamental contribution to the [37:06] analytical engine LEL started studying [37:08] Elementary methods of differences in [37:11] 1839 this is 5 years after babet [37:14] conceived the analytical engine so there [37:16] is no possibility as encouragement [37:19] certainly um bage met love lace in 1833 [37:22] when LEL was 17 bage was over 20 years [37:25] older there's no question he was [37:27] flattered and he she encouraged her she [37:29] was fascinated by the machines but that [37:31] she made any substantial contribution to [37:33] the concepts of the analytical engine [37:34] the chronology does not back up finally [37:37] she was the first computer program this [37:38] is excusable that people should think [37:41] this because she was the person who [37:42] published the first record of an [37:44] algorithmic procedure her notes 1843 on [37:48] bab's analytical engine is the most [37:49] comprehensive description of his [37:51] analytical engine but the examples were [37:54] bages they were used in Terin in 1841 so [37:57] babich was the first programmer she was [37:58] the first published program the first [38:00] thing that corresponds to stepwise [38:01] algorithmic procedure she is deserved of [38:04] her Fame I'm not rubbishing love Lacy [38:06] I'm rubbishing the people who have [38:07] lionized her for false reasons she's [38:09] absolutely deserved of her place in the [38:12] Hall of Fame she was the only one who [38:14] saw what the fundamental significance of [38:16] bab's analytical engine was and saw it [38:19] saw further than babage did she was the [38:21] one who understood the fundamental [38:22] transition from calculation to [38:24] computation that a computer was a [38:26] machine that could manipulate symbols [38:27] according to rules that number could [38:29] represent some entity other than [38:31] quantity that if you can have a machine [38:33] that manipulated number and the number [38:35] represented notes of Music letters of [38:37] the alphabet you had a computer and [38:40] she's it's not again this is not a [38:41] suggestive hint she is banging on the [38:43] table and saying this is the [38:45] significance of the analytical engine [38:47] general purpose computation through [38:48] symbolic manipulation according to rules [38:51] babage never saw the machines outside [38:53] the domain and envelope of mathematics [38:55] he could see an algebra machine machine [38:56] but still within mathematics where it [38:58] could manipulate symbols where the [39:00] specific um value of each symbol was not [39:02] necessarily [39:04] critical so that's my rant about love [39:08] lace she deserves a place but because [39:10] she saw Beyond she was fundament [39:13] responsible for that fundamental [39:14] transition she saw that the machine was [39:17] different in kind that a computer was [39:19] different in kind from any machine that [39:21] had gone [39:25] before finally Difference Engine Number [39:28] Two this [39:31] is one of 20 drawings that babish left [39:35] excuse [39:40] me one of 20 drawings that bab most [39:42] evocative of its overall shape that's [39:43] the main elevation bab 163a imprinted on [39:46] my mind it's operated by cranking a [39:49] handle we have here the micro program a [39:50] series of Cam Stacks that's 28 conjugate [39:53] cams geometric conjugates of each other [39:55] inversions of each other [39:57] um seven orders of difference tabular [39:59] value over there first difference second [40:01] third fourth seventh difference down [40:02] here the end product of the whole [40:04] process of a single cycle is that a 31 [40:06] digigit number appears on this last [40:08] column that is the printing apparatus [40:10] which prints hard copy and um produces [40:14] stereotype plates the machine has 8,000 [40:16] Parts a bronze Cast Iron and steel it is [40:19] 11t long it is 7 foot high and it's made [40:22] of bronze Cast Iron and Steel um the [40:26] print is a particular significance [40:28] because it is programmable you can [40:29] program the format on it it doesn't run [40:30] Windows and it's not [40:32] portable um it crashes nearly as often [40:36] though um but crashing is good as I say [40:39] it's Error detection um the printer is [40:42] interesting because I mean the whole [40:44] thing is interesting of course um but [40:46] the printer is particularly interesting [40:47] because you can actually program the [40:48] format on the page you have variable [40:49] line height varium column width you can [40:51] leave gaps between groups of lines for [40:52] ease of reading you can print in 1 two [40:54] three or four columns at will you can it [40:57] it stereotypes in two fonts [40:59] simultaneously small and large and has [41:01] line height that adapts according to the [41:03] size of the font and all this is [41:04] entirely mechanically programmable it [41:06] will automatically line wrap or column [41:08] wrap whether you print down the page or [41:09] across the page and I'll show you [41:10] examples of that now this is the machine [41:12] in all its Glory this is the one that [41:14] was finished in London in 2002 took 17 [41:16] years to build strictly speaking it was [41:18] built over a period of 17 years it was [41:21] not just the logic of the building there [41:22] was something called people that were [41:24] involved and they are much less iCal and [41:26] solving technical difficulties [41:28] fundraising publicity all that stuff so [41:30] the from start to finish 1985 to 2002 [41:32] was 17 years and that's the end result [41:35] the one down the road at the Computer [41:37] History Museum is an exact duplicate of [41:38] this in fact it's a multiple original [41:40] because we made two sets of parts for [41:41] the printer in one go so this is in fact [41:44] a duplicate the calculating section and [41:46] the print is in fact a multiple original [41:48] and that is the machine um that is down [41:50] the down the road so there we can see [41:51] the handle we can see the cam stack the [41:53] printing mechanism um and the series of [41:56] 10 [41:57] columns numbers are stored in figure [41:59] Wheels decade decimal digital machine [42:03] stored in columns so it goes units T [42:04] hundreds thousands from the bottom to [42:05] the top 31 digit [42:08] Precision um that's an example of bab's [42:10] design um that's a figure wheel a sector [42:13] wheel a figure wheel this device there [42:15] that's a single mechanism which I'll [42:16] show you now and this Fair Ground device [42:18] is the Helix Helix which runs up and [42:20] down which we will see that thing that [42:22] rotated to do the polling of each decade [42:25] and that is the most intricate device [42:26] and that's a modern P Spot drawing you [42:28] can see how much more information there [42:30] is on this than is on bab's original so [42:32] although baber specifies the thing [42:33] entirely you know what the shape is you [42:35] know what it needs to do there's [42:36] information which is not present methods [42:38] of manufacturer tolerancing choice of [42:39] materials and finish are not provided we [42:42] did that from for all 8,000 parts from a [42:44] knowledge of 19th century um machines [42:46] and an inspection of bab's own machines [42:48] we also did composition analysis on [42:50] Metals used by babage to ensure that we [42:52] didn't actually um make the machine from [42:55] any higher specification material that [42:57] was available to [42:58] babage um so authenticity was the [43:01] watchword of the project we had to [43:03] resist the charge absolutely of somebody [43:05] saying you built the engine but babage [43:07] could not have so we were scrupulous in [43:09] our interpretation of [43:11] this that's a simulation by the way I'd [43:14] like to interest Google in doing a [43:15] complete simulation of the analytical [43:17] engine and please see me afterwards um [43:21] and there is major work to be done on [43:22] this also to actually digitize the [43:24] archive um that is in London to make [43:26] this available at the moment the archive [43:28] is 70 Mi outside London three people [43:29] have studied it um one is dead one is in [43:32] his '90s and me and I'm not going to go [43:34] on forever this stuff needs to be made [43:36] available to the public and um I think [43:38] Google should adopt babage end of [43:40] advertisement um that's the back of [43:43] engine um and that you can see down the [43:46] road and now I'm going to run the [43:48] machine for you there is no substitute [43:49] for seeing this machine run especially [43:51] with Dicky sound um so this is in the [43:53] absence of going walking down the road [43:55] we're just going to run the machine and [43:56] you can see how it [43:59] [Music] [44:06] goes that is those are the locks coming [44:08] in those those sword blades that [44:10] slapping sound there's taking a ink [44:12] print [44:16] copy cam stack [44:18] microprogram shifting and turning that's [44:20] pipelining deals with odds the odds axes [44:24] the one cycle half cycle the even is in [44:26] the other half so he's adding a number [44:28] from there and result ends up down there [44:30] and then transfer to the [44:33] printer okay you can see the cam step [44:35] it's all greased up so it looks horrible [44:37] the one in down the road is much cleaner [44:39] using transparent [44:41] grease this is heavy machinery this is [44:43] Big [44:47] Iron that's the Locking you can see the [44:51] synchronization of the sound with that [44:53] that's a lock coming in and out very [44:54] quickly pressure angle is still great we [44:56] couldn't turn the handle when we first [44:57] built [44:59] it okay we now go look at the front of [45:01] the [45:07] machine and you'll see that shifting [45:09] motion ODS and evens go together the [45:12] odds ax is 1357 going one H cycle evens [45:14] AIS the [45:24] next [45:36] [Music] [45:44] who cares whether it works when it looks [45:45] like [45:51] [Music] [45:52] that that's the back of the engine [45:54] that's the car mechanism [45:57] [Music] [45:58] and if you held a pineapple up to it it [46:00] would slice [46:03] [Music] [46:06] it those things are viciously sharp you [46:08] can tell a babage engineer by his [46:10] [Music] [46:18] hand yep that's it it's pulling each [46:21] decade in turn it can also deal with [46:23] secondary carries carries as a result of [46:24] carries [46:34] um that's a printing [46:40] mechanism um how we for time [46:43] um as long as yeah and we can stop and [46:48] do questions or I can just carry on [46:49] there's not a lot to go okay yeah um [46:53] okay um I'm going to explain how the the [46:55] stereotyping printing mechanism works [46:57] now through a simulation this is the [46:58] kind of thing I'd like Google to do but [47:00] better [47:02] um ah um I missed that okay I mentioned [47:06] that um that it can print down the line [47:08] and automatically column wrap or it can [47:10] print across the page and automatically [47:12] line wrap it does that automatically you [47:13] just set it up and it does it so you can [47:15] print as I say Line to Line or column to [47:17] column I'm just going to go back one um [47:20] to see if we can run [47:23] the run the simulation [47:29] okay the problem is how do you get a [47:31] figure wheel in that plane a value from [47:33] a figure wheel in that plane to a print [47:34] wheel in that [47:35] plane so what babage does is he has a [47:38] train which is sector Wheels horizont [47:41] horizontal racks spindles and vertical [47:43] racks so when that turns it just [47:44] transfers the uh and and transfers the [47:47] result to that that's the print wheel [47:49] that's where that's the bit that will be [47:50] Inked and he's got 31 digits up it [47:52] prints to 30 digits calculates the 31 [47:54] prints to 30 so when these wheels turn [47:57] it transfers the result to the print [47:58] Wheels over there and this is how it [48:00] does it so that's um feed roll Paper [48:04] Stock new result is [48:06] ined comes up takes an impression the [48:09] act of coming up winds the paper [48:11] on then resets to zero takes a new [48:14] result inks the new line of print takes [48:17] an impression and just keep cycling [48:22] through [48:24] yep there's no sound to this virtual [48:28] world is [48:31] silent [48:36] um okay what it then does is start [48:39] stereotyping at the same time it does [48:41] simultane there's no printing overhead [48:42] in [48:44] this okay you've got trays of soft [48:46] material in our case we use plaster of [48:48] Paris and there's an example of this on [48:49] display down the the very first [48:50] experimental plastro Paris tray we use [48:52] so it impresses results these um [48:55] stereotyp heads lower into the trays [48:57] that's the small font that's the Big [48:58] Font the trays move at different rates [49:00] to adjust for the line Heights they [49:02] index at different rates and you can see [49:04] that the transfer the results is then [49:05] transferred to The Stereotype plates and [49:07] we'll now watch that run so that's just [49:09] the stereotyping working now the same [49:11] vertical racks are transferring to the [49:12] print Wheels transferring to the [49:14] horizont to the um print heads and [49:16] impressing results in um this was done [49:20] in 3D Studio took too long to render [49:21] full results so we on using three digits [49:23] please help um [49:28] uh so you can see the trays index [49:30] automatically each line it's a got two [49:33] degrees of freedom and automatically [49:35] jumps across when the page is um gets to [49:37] the end of the [49:41] page and now we can see the whole thing [49:43] run together in [49:49] Symphony so single result transferred to [49:51] print wheels and stereotype heads takes [49:53] an ink impression that's for record [49:54] keeping only you can't produce it from [49:56] this you produce printing plates which [49:58] you then put in printing press you've [49:59] eliminated the human from theop Loop [50:00] entirely from calculation to the [50:02] production of stereotype plates um now [50:05] let's see if we can run this um and show [50:09] you how that actually [50:15] works the the the the trays are driven [50:19] by these falling weights so you don't [50:22] rely on the uniformity of the drive of [50:24] the person with the handle to ensure the [50:26] indexing is accurate this is what [50:28] determines the indexing that's how it's [50:29] programmed that's the patent Wheels [50:31] programmed the [50:32] format that okay we can now going look [50:35] closely at the ink bath that's a cover [50:37] there gooey ink in there that's a print [50:39] roller this is a spreader roll if you [50:40] get a gap there you can see this thing [50:42] slides backwards and forwards it's got a [50:44] snail cam on the end I'll do that I'll [50:46] accelerate it [50:47] now you can see it's sliding backwards [50:49] and forwards so if there's a gap there [50:51] it actually smears it horizontally as [50:53] well as [50:54] vertically that was done by hand that [50:56] doesn't do that by itself it does it [50:57] incrementally and now we can actually [50:59] look straight into the thing and see [51:00] those are the print Wheels where it [51:01] receives the results and we'll actually [51:03] see it printing now we'll get some hard [51:06] copy you can see how very very fast how [51:08] quick the timing cycle is watch that [51:10] print wheel come down now boom boom it [51:12] has to get out of the way before it [51:13] comes out [51:15] um and as I say that sweep automatically [51:18] advances the papers so new result ined [51:20] impression sets to [51:24] zero [51:26] now we'll see it actually picking up a [51:27] result and you can program it to put an [51:29] index there so it's actually got the [51:31] independent variable 131 we'll see the [51:33] next one is the next so X is increasing [51:35] from 131 to 132 and that'll be up to a [51:38] 30- 3digit 27 digigit result so that's [51:41] what it's doing and the simulation sort [51:44] of gets the logic of it [51:52] done [51:54] okay what happened this is [51:59] a this is a monolithic machine it is [52:02] hardwired it is phased the cam stack [52:05] orchestrates the internal motions but [52:07] there is no room for adjustment they've [52:09] never built a machine of any complexity [52:10] you cannot uncouple the printer from the [52:12] machine when he jams best you can do is [52:14] hunt and Peck you poke around with the [52:16] first world's first logic probe to see [52:19] where it is solid where there is give [52:20] where there isn't so you leave her up on [52:22] the sector Wheels on the on the figure [52:24] Wheels to see is it stuck is it jammed [52:26] you fiddle with the locks to see is it [52:27] stuck is it [52:34] jammed [52:37] um this is the book that tells the [52:39] account of trying to build a machine for [52:40] the b sentin in 1991 of bab's birth it [52:44] stops in ' 91 where the calculating [52:45] section was built the American Edition [52:47] is slightly updated because by that [52:49] stage 2002 the printing section had [52:51] already been updated so there there are [52:53] photographs of the printing section in [52:54] the American Edition the cogwell brain [52:56] was intended the title was intended to [52:57] capture that ambiguity of carile of [52:59] inner and outer sense the terms in which [53:01] you saw the world and the terms in which [53:04] um you interacted with the terms in [53:06] which you saw the world and the fact [53:07] that mind that that there was a [53:09] computational view of Consciousness [53:11] which only arose once you had autonomy [53:13] autonomous machines um but the the books [53:16] the text of the books are essentially [53:17] identical and I thought I just finish [53:19] with a Sumptuous piece of engineering [53:21] sculpture thank [53:24] you [53:31] [Applause] [53:36] I understand that the mics ought to be [53:38] used for [53:39] questions so I'm instructed to inform [53:41] you of that [53:50] um so how was exactly data input into [53:53] machine how was the initi initial data [53:56] input right the only input you can [53:58] provide is to enter initial values so [54:01] you enter the initial values by entering [54:02] the by hand onto the figure Wheels but [54:04] you have to disable the security [54:06] mechanisms that prevent you deranging [54:08] the machine babage challenges you and [54:10] says walk up to the machine even when [54:11] it's functioning and attempt to derange [54:13] it externally and you won't be able to [54:15] so you have to disable those mechanisms [54:16] and then you enter the the the numbers [54:18] by hand the numbers are pre-calculated [54:20] so the starting value of x and the value [54:22] of all the differences for the starting [54:24] value of x and for a particular a [54:25] polinomial with its coefficients mhm so [54:27] all terms x 7 x 6 etc etc with any [54:30] coefficient positive or negative um um [54:33] can be calculated and tabulated and you [54:35] do a pre-calculate by hand a setup table [54:39] taking account of the odds evens [54:40] pipelining offset and you enter these [54:42] numbers as a setup procedure at the [54:44] beginning and before you do that you [54:46] have to um disable using those levers [54:49] and this lever over there you have to [54:50] disable the security mechanism to free [54:52] the wheels up you set the odd axes in [54:54] one part of the cycle you then lock them [54:57] you then Advance the machine to the [54:58] evens axis setup point you set all the [55:00] EV axis that's axis 235 23 two 2 468 um [55:06] and then you uh lock the engine take it [55:08] to zero undo the manual locks and off [55:10] you go from that point on every cycle of [55:11] the machine preuses the next value in [55:13] the table thank [55:16] you I know that babage was aware of [55:20] binary arithmetic or I read that um do [55:23] you think it would have been easier to [55:25] use if he'd used [55:27] binary B considered all practical number [55:31] bases 357 12 15 100 and of course binary [55:36] he chose decimal because of engineering [55:38] efficiency the lower the base the larger [55:40] the number of moving parts so it's not [55:43] only just familiarity of decimal system [55:45] which is useful um but it was a good [55:47] balance if it was heximal dood decimal [55:49] he would have had to have a translation [55:50] process so he chose a familiar binary [55:52] but it's digital and he explicitly says [55:55] this is for economy this is to minimize [55:57] the number of moving parts thank [56:01] you uh was the uh logic probe specified [56:05] by [56:07] B um no but he never built the machine [56:09] and I believe had he he would have had [56:11] every single thing adjustable he would [56:12] have had what we call test points he [56:14] would have facilities for uncoupling the [56:15] printing mechanism and the uh I've [56:17] actually insisted they put a clutch in [56:19] here you can see the printing mechanism [56:20] is driven from the main Drive I put an [56:22] extra handle on and a clutch otherwise [56:25] to drive the printer you have to drive [56:26] the engine firstly you're 11t away from [56:28] the person who's operating it when doing [56:29] fine adjustments it's it's impossible [56:31] secondly there are times where you want [56:32] to run the printer once you've set the [56:34] machine up the moment you turn the [56:35] handle the internal state has changed so [56:38] if you then have to adjust the printer [56:39] for instance to get the ink flow running [56:40] you want to run the printer without [56:41] running the engine I got to redo the [56:42] calculation so you want to Prime the [56:44] printer by being able to cycle it [56:45] separately so I arranged the clutch [56:47] there which actually uncouples the drive [56:49] and it's been absolutely invaluable [56:51] every modification of that kind is [56:53] reversible and we used almost everywhere [56:57] um solutions that we' found babage had [56:59] used elsewhere um did you find any bugs [57:03] in his designs as you were building the [57:04] machine I assume there must have been in [57:06] something so big that's a very good [57:07] question um we found nothing that [57:10] compromised the basic logic of his [57:12] design there was nothing that [57:14] compromised the basic logic but there [57:15] were countless modifications um for [57:18] example the um elevation of the printing [57:20] mechanism shows the print wheel the [57:23] inking wheel um suspend in midair [57:25] there's no visible means of support so [57:26] what does it mean to build this machine [57:28] to babage designs you make the inking [57:30] roll and you hold them up and you let [57:31] them go um I mean it's not practical so [57:34] this was quite a curatorial dilemma um [57:37] in what sense would we be building bab's [57:39] engine if we made any modification at [57:40] all and the Breakthrough came with the [57:42] realization that bab's designs are not a [57:45] work of unimpeachable perfection this is [57:47] an arrested engineering project so for [57:49] example there are parts that are [57:52] inconsistently dimensioned on different [57:53] drawings so the part that one drawing [57:55] maybe one thing and something else you [57:57] have to resolve this there are omitted [57:59] mechanisms there are redundant [58:01] mechanisms there is one fundamental flaw [58:03] that is so profound it's almost [58:05] mindboggling the carry mechanism would [58:07] not work not because it's not logical [58:09] but because the wheels go in the wrong [58:10] direction I mean it's that fundamental [58:12] now we speculated was this deliberate [58:14] was this a drafting error was this an [58:15] anti-espionage [58:17] device um how much supervision did the [58:19] babage have over the so we had to [58:21] resolve it what we did was used exactly [58:23] the device but mirror imaged it there [58:25] was only one device whose design we had [58:28] to improve and that was a clutch the [58:31] rewind clutch when you crank the handle [58:34] the printer automatically halts the [58:35] machine at the end of a page so you [58:37] never have to calculate how many cycles [58:39] you do you just print do it until it [58:40] gets and it freezes the machine so the [58:42] machine will not overrun so it's ready [58:44] for the next cycle now the clutch that [58:46] rewinds and then automatically these [58:48] these weights rewind you don't have to [58:50] wind them up by hand you wind one by [58:51] hand the other one winds automatically [58:52] the clutch that does that needs to to [58:54] both turn and slide and the turning the [58:58] torque was too high to get it to slide [59:00] that was it so we designed another [59:01] clutch in the same space using a very [59:03] similar mechanism but we to relieve that [59:05] Str that was the only device we actually [59:07] had to change there are a large number [59:08] of things that we added these [59:10] counterbalancing weights for instance on [59:11] the printing vertical racks are not [59:13] babage these reinforcing brackets are [59:15] not babage the handle I've explained is [59:16] not babage this U counterbalancing [59:19] spring underneath is not babage all of [59:20] them are removable and all of them are [59:22] solutions that counterbalancing spring [59:24] is based on is original spring up there [59:26] so everywhere we we paid very careful [59:28] attention to issues of authenticity so [59:31] the drawings are all the kinds of things [59:33] you'd expect from a set of drawings that [59:34] never went to a workshop they're intact [59:36] but there was a lot of information that [59:38] had to be provideed which babage would [59:39] have had to solv we solved it with the [59:41] finest attention we could to knowledge [59:43] of practice of the [59:46] day uh how does babbage's budget [59:49] compared with what you actually spent I [59:51] know it's rather hard to compare but B [59:54] had a blank check from government he [59:57] spent 175,000 of B of government money [60:00] in between 1822 and [60:02] 1832 I estimate that to be about $200 [60:05] million now and the government got [60:06] practically nothing for it um they got [60:08] 17th of the engine which they gave to [60:09] baby and said sorry um um I'm not at [60:13] Liberty to disclose how much this cost [60:15] because actually it's not because it's [60:16] on being secretive but because it's [60:17] quite difficult to unbundle I was a [60:19] staff member my research time was not [60:21] costed but I can tell you what the [60:23] insurance value of this machine is and [60:24] that based on what it would cost to [60:26] replace in the event of ir irreparable [60:29] damage or loss total loss and it's [60:31] between 2 and a half and 3 million so we [60:33] could build one from scratch with the [60:35] knowledge we have from for about that [60:38] sum babby spent two and a half2 the [60:40] equivalent of $200 million and um didn't [60:43] finish um we did it for substantially [60:45] less and if Google wants one for the [60:47] foyer I'm taking [60:52] orders we're actually out of time so I [60:54] want to thank you so much for coming to [60:56] Google today thank [61:02] you