Who am I and why am I telling this story (1m37s)

Steve opens with a self-deprecating CV — Linux nerd, retro-computing volunteer at the Cambridge computer museum, Theremin-builder — and a clear thesis: everything I'm about to tell you, I taught myself. If I can do this, anyone can.

"I'm not doing that because I'm a professor of computer science. I'm not doing that because I have a doctorate in history"

"all of this is a very long-winded way of saying, 'if I can do this, anyone can'"

"this is the ego slide — this is where the speaker brags about themselves for 10 minutes while everyone checks their email"

The first computer program: Bernoulli numbers (5m13s)

Luigi Menabrea wrote a flawed Italian book about Babbage's analytical engine. Ada Lovelace translated it, added enough notes to triple its length, and in one appendix laid out the program that calculated Bernoulli numbers — the algorithm that has stood as the first computer program ever since.

"in the process of just translating the book, she made it three times bigger"

"back in the day, these numbers would be computed by hand by a person who would be called, quite literally, a computer"

"if anyone's seen the film Hidden Figures... the women in that film were known as computers"

How Ada's program is laid out (8m49s)

The program is a tabular sequence: an operation number (line number), the operation (one of four: ×, ÷, +, −), the two variable operands, a destination column that allows writing the result into multiple cells at once (because moving the mill's cogs once was cheaper than moving them twice), and a comment-like column for "how many times has this variable changed" as a self-check against transcription errors.

"for people that remember BASIC and remember typing '10 PRINT Thomas is an idiot, 20 GOTO 10' — it's a line number"

"I said there were four instructions; add, subtract, divide and multiply. Where's equals? Well, it wasn't there at all"

"this is almost like a self-check — as you work through the program, you can say 'I've seen V2 three times now, but this is telling me it's only been changed twice'"

The first computer bug (in the comments) (13m29s)

Walking through the first six lines, Steve spots that the printed comment says 2n − 1 divided by 2n + 1 while the actual operation column says the opposite. The arithmetic in the code is correct; the explanatory note describing it is wrong. Comments and code don't match → everything is wrong. The first computer bug is, fittingly, a documentation bug.

"not only did she create the first computer program, she's created the first computer bug — well done Ada"

"the mistake she made was actually in the comments"

"if comments and code don't match, everything is wrong. That's the very first example on line four of the first program"

Translating it to JavaScript (15m32s)

The simulator is the smallest viable interpreter: an array of { line, op, a, b, destinations[] } instructions, a switch on op that performs the arithmetic in the mill (what we'd now call the accumulator), then a loop that writes the result into each destination. The Manchester Baby of 1947 had the same constraint — no add instruction, only subtract — borrowed almost directly from Ada's design.

"I'm using a nice variable name state, because it's a nice generic term"

"back in the forties and even the fifties and sixties, people were still using the word mill to mean accumulator"

"the Manchester Baby does not have an add instruction — the only thing it can do is subtract"

The loop that could never work (22m13s)

Between lines 23 and 24 Ada writes "Here follows a repetition of operations 13 to 23". She invented the loop concept. But the loop is broken — operations only reference variables by literal index (V2, V3, V4...), so there's no way to use the value of N as the variable index inside the loop body. To run the program in his simulator Steve had to invent an indirection layer (essentially V[n]) that didn't exist in Ada's notation.

"she's created the very first loop. Well done"

"there is absolutely no way in her designation of programming languages where it would say 'go to an arbitrary variable'"

"essentially you need an array index — this loop would never have worked"

Was Ada really the first programmer? (26m24s)

The arguments Steve has spent years arguing with online: the code has a bug (every programmer writes bugs), the notation is non-standard (it conveyed her idea across 180 years to him, so it worked), the loop is broken (the intent of the algorithm is clear), she was just a translator (she added two-thirds of the final book), Babbage must have written programs already (he was a mechanical engineer, he didn't).

"the very first example on line four of the first program... Anyone here not written a bug? It's like a rite of passage"

"her document from the 1840s conveyed an idea to me nearly 200 years later — if it does the job of conveying an idea across centuries from someone who's now long dead, then it's done a pretty good job at communicating"

"labelling it as 'just a translator' would say 'well, you're not a computer programmer if you use a compiler'"

What I think the real argument is (30m57s)

Ada didn't just write a program. She introduced variables, operations, loops, bugs, a notation for storing a program, and the conceptual leap that the same machine could be used for poetry, music or images by treating numbers as symbols. She effectively foresaw ChatGPT, MIDI and computer art, in 1843. The question Steve poses: if a man had done all of that, would he have been considered the first programmer?

"she invented ChatGPT just 200 years early"

"or it could be a dot, a colour, which you could then arrange in a nice sequence — thanks for inventing Photoshop"

"if a man had done all of that, would he have been considered the first programmer? And I think he would've done — so if he would've been considered the first programmer, then she should be considered the first programmer"