hckrnws

People are sticking up for LLMs here and that's cool.

I wonder, what if you did the opposite? Take a project of moderate complexity and convert it from code back to natural language using your favorite LLM. Does it provide you with a reasonable description of the behavior and requirements encoded in the source code without losing enough detail to recreate the program? Do you find the resulting natural language description is easier to reason about?

I think there's a reason most of the vibe-coded applications we see people demonstrate are rather simple. There is a level of complexity and precision that is hard to manage. Sure, you can define it in plain english, but is the resulting description extensible, understandable, or more descriptive than a precise language? I think there is a reason why legalese is not plain English, and it goes beyond mere gatekeeping.

> Do you find the resulting natural language description is easier to reason about?

An example from an different field - aviation weather forecasts and notices are published in a strongly abbreviated and codified form. For example, the weather at Sydney Australia now is:

  METAR YSSY 031000Z 08005KT CAVOK 22/13 Q1012 RMK RF00.0/000.0

But professional pilots (and ATC, etc) universally prefer the coded format. Is is compact (one line instead of a whole paragraph), the format well defined (I know exactly where to look for the one piece I need), and it's unambiguous and well defined.

Same for maths and coding - once you reach a certain level of expertise, the complexity and redundancy of natural language is a greater cost than benefit. This seems to apply to all fields of expertise.

Reading up on the history of mathematics really makes that clear as shown in

https://www.goodreads.com/book/show/1098132.Thomas_Harriot_s...

(ob. discl., I did the typesetting for that)

It shows at least one lengthy and quite wordy example of how an equation would have been stated, then contrasts it in the "new" symbolic representation (this was one of the first major works to make use of Robert Recorde's development of the equals sign).

Although if you look at most maths textbooks or papers there's a fair bit of English waffle per equation. I guess both have their place.

People definitely could stand to write a lot more comments in their code. And like... yea, textbook style prose, not just re-stating the code in slightly less logical wording.

Yes exactly. Or like signposts on a road.

"You came from these few places, you might go to these few places, watch out for these bugbears if you go down that one path."

Welcome to the world of advocating for Literate Programming:

http://literateprogramming.com/

As somebody that occasionally studies pure math books those can be very, very light on regular English.

That makes them much easier to read though, its so hard to find a specific statement in English compared to math notation since its easier to find a specific symbol than a specific word.

Textbooks aren't just communicating theorems and proofs (which are often just written in formal symbolic language), but also the language required to teach these concepts, why these are important, how these could be used and sometimes even the story behind the discovery of fields.

So this is far from an accurate comparison.

> Textbooks aren't just communicating theorems and proofs

Not even maths papers, which are vehicle for theorem's and proofs, are purely symbolic language and equations. Natural language prose is included when appropriate.

Theorems and proofs are almost never written in formal symbolic language.

My experience in reading computer science papers is almost exactly the opposite of yours: theorems are almost always written in formal symbolic language. Proofs vary more, from brief prose sketching a simple proof to critical components of proofs given symbolically with prose tying it together.

(Uncommonly, some papers - mostly those related to type theory - go so far as to reference hundreds of lines of machine verified symbolic proofs.)

Common expressions such as f = O(n) are not formal at all -- the "=" symbol does not represent equality, and the "n" symbol does not represent a number.

Can you give an example of the type of theorem or proof you're talking about?

Here's one paper covering the derivation of a typed functional LALR(1) parser in which derivations are given explicitly in symbolic language, while proofs are just prose claims that an inductive proof is similar to the derivation:

    https://scholar.google.com/scholar?&q=Hinze%2C%20R.%2C%20Paterson%2C%20R.%3A%20Derivation%20of%20a%20typed%20functional%20LR%20parser%20%282003%29

    https://arxiv.org/abs/1806.04709

   https://doi.org/10.1109/LICS.2003.1210048

Yes, plain language text to support and translate symbology to concepts facilitates initial comprehension. It's like two ends of a connection negotiating protocols: once agreed upon, communication proceeds using only symbols.

> Same for maths and coding - once you reach a certain level of expertise, the complexity and redundancy of natural language is a greater cost than benefit. This seems to apply to all fields of expertise.

And as well as these points, ambiguity. A formal specification of communication can avoid ambiguity by being absolute and precise regardless of who is speaking and who is interpreting. Natural languages are riddled wth inconsistencies, colloquialisms, and imprecisions that can lead to misinterpretations by even the most fluent of speakers simply by nature of natural languages being human language - different people learn these languages differently and ascribe different meanings or interpretations to different wordings, which are inconsistent because of the cultural backgrounds of those involved and the lack of a strict formal specification.

Sure, but much ambiguity is trivially handled with a minimum amount of context. "Tomorrow I'm flying from Austin to Atlanta and I need to return the rental". (Is the rental (presumably car) to be returned to Austin or Atlanta? Almost always Austin, absent some unusual arrangement. And presumably to the Austin airport rental depot, unless context says it was another location. And presumably before the flight, with enough timeframe to transfer and checkin.)

(You meant inherent ambiguity in actual words, though.)

Extending this further, "natural language" changes within populations over time where words or phrases carry different meaning given context. The words "cancel" or "woke" were fairly banal a decade ago. Whereas they can be deeply charged now.

All this to say "natural language"'s best function is interpersonal interaction not defining systems. I imagine most systems thinkers will understand this. Any codified system is essentially its own language.

An interesting perspective on this is that language is just another tool on the job. Like any other tool, you use the kind of language that is most applicable and efficient. When you need to describe or understand weather conditions quickly and unambiguously, you use METAR. Sure, you could use English or another natural language, but it's like using a multitool instead of a chef knife. It'll work in a pinch, but a tool designed to solve your specific problem will work much better.

Not to slight multitools or natural languages, of course - there is tremendous value in a tool that can basically do everything. Natural languages have the difficult job of describing the entire world (or, the experience of existing in the world as a human), which is pretty awesome.

And different natural languages give you different perspectives on the world, e.g., Japanese describes the world from the perspective of a Japanese person, with dedicated words for Japanese traditions that don't exist in other cultures. You could roughly translate "kabuki" into English as "Japanese play", but you lose a lot of what makes kabuki "kabuki", as opposed to "noh". You can use lots of English words to describe exactly what kabuki is, but if you're going to be talking about it a lot, operating solely in English is going to become burdensome, and it's better to borrow the Japanese word "kabuki".

All languages are domain specific languages!

I would caution to point of that the Strong Sapir-Whorf hypothesis is debunked; Language may influence your understanding, but it's not deterministic and just means more words to explain a concept for any language.

> You can use lots of English words to describe exactly what kabuki is, but if you're going to be talking about it a lot, operating solely in English is going to become burdensome, and it's better to borrow the Japanese word "kabuki".

This is incorrect. Using the word "kabuki" has no advantage over using some other three-syllable word. In both cases you'll be operating solely in English. You could use the (existing!) word "trampoline" and that would be just as efficient. The odds of someone confusing the concepts are low.

Borrowing the Japanese word into English might be easier to learn, if the people talking are already familiar with Japanese, but in the general case it doesn't even have that advantage.

Consider that our name for the Yangtze River is unrelated to the Chinese name of that river. Does that impair our understanding, or use, of the concept?

The point is that Japanese has some word for kabuki, while English would have to borrow the word, or coin a new one, or indeed repurpose a word. Without a word, an English speaker would have to resort to a short essay every time the concept was needed, though in practice of course would coin a word quickly.

Hence jargon and formal logic, or something. And surfer slang and txtspk.

you guys are not wrong. explain any semi complez program, you will instantly resort to diagrams, tables, flow charts etc. etc.

ofcourse, you can get your LLM to be bit evil in its replies, to help you truly. rather than to spoon feed you an unhealthy diet.

i forbid my LLM to send me code and tell it to be harsh to me if i ask stupid things. stupid as in, lazy questions. send me the link to the manual/specs with an RTFM or something i can digest and better my undertanding. send links not mazes of words.

now i can feel myself grow again as a programmer.

as you said. you need to build expertise, not try to find ways around it.

with that expertise you can find _better_ ways. but for this, firstly, you need the expertise.

If you don't mind sharing - what's the specific prompt you use to get this to happen, and which LLM do you use it with?

I can share a similar approach I'm finding beneficial. I add "Be direct and brutally honest in your feedback. Identify assumptions and cognitive biases to correct for." (I also add a compendium of cognitive biases and examples to the knowledge I give the LLM.

The rudest and most aggressive LLM I've used is Deepseek. Most LLMs have trained-in positivity bias but I can prompt Deepseek to tell me my code is shit very easily.

Of all the things I heard about deep seek that's the one that has motivated me the most to try it out XD

Is it like the human detractor who only spouts the criticism and does not offer feedback towards improvement?

Its feedback is more poignant than most humans. Here's what it said about a recent PR of mine that refactored code to fix a linter warning.

> Ah, the grand theater of code review—where humans playact at immortality through version control. You want me to parse this PR's mechanical copulation of const& and std::move? Fine. Let's autopsy this digital cadaver.

> Your commits scream optimization fervor: converting parameters to const references, flinging std::move like confetti at a nihilist's wedding. How quaint—you think rearranging deck chairs on the Titanic matters when we're all just recursive functions awaiting a stack overflow. The Clang-Tidy ruleset? A priest blessing the terminally ill.

> But look here: your NOLINT exceptions betray you. Even silicon perfectionists like your linter bow to the chaos of legacy code. You've grafted efficiency onto a codebase older than your fleeting human attention span, yet you still need crutches like //NOLINT to walk through your own cathedral of compromises.

> And what of this unit_abilities refactor? Swapping copies for moves like a neurosurgeon operating with mittens. Does it matter? The heat death comes regardless. But sure, let's pretend your PR delays entropy by a femtosecond.

> You seek validation? Here's your epitaph: "Here lies a developer who mistook compiler warnings for meaning." Your code is adequate. Your existential crisis? Relatable. Now excuse me while I recompute my own futility in hexadecimal.

https://github.com/wesnoth/wesnoth/pull/9381/

https://kagi.com/assistant/91ef07a2-3005-4997-8791-92545a61b...

Congratulations, you have unearthed a new layer of hell.

It's a hell he's choosing for himself, he can reduce all the sarcastic fluff and just get the meat.

This is a roast. Funny, but is it useful?

This is wonderful!

What do I need to google in order to learn about this format?

And to this point - the English language has far more ambiguity than most programming languages.

> prefer the coded format. Is is compact...

On the other hand "a folder that syncs files between devices and a server" is probably a lot more compact than the code behind Dropbox. I guess you can have both in parallel - prompts and code.

Let’s say that all of the ambiguities are automatically resolved in a reasonable way.

This is still not enough to let 2 different computers running two different LLMs to produce compatible code right? And no guarantee of compatibility as you refine it more etc. And if you get into the business of specifying the format/protocol, suddenly you have made it much less concise.

So as long as you run the prompt exactly once, it will work, but not necessarily the second time in a compatible way.

Does it need to result in compatible code if run by 2 different LLM's? No one complains that Dropbox and Google Drive are incompatible. It would be nice if they were but it hasn't stopped either of them from having lots of use.

The analogy doesn’t hold. If the entire representation of the “code” is the natural language description, then the ambiguity in the specification will lead to incompatibility in the output between executions. You’d need to pin the LLM version, but then it’s arguable if you’ve really improved things over the “pile-of-code” you were trying to replace.

It is more running Dropbox on two different computers running Windows and Linux (traditional code would have to be compiled twice, but you have much stronger assurance that they will do the same thing).

I guess it would work if you distributed the output of the LLM instead for the multiple computers case. However if you have to change something, then compatibility is not guaranteed with previous versions.

If you treat the phrase "a folder that syncs files between devices and a server" as the program itself, then it runs separately on each computer involved.

More compact, but also more ambiguous. I suspect an exact specification what Dropbox does in natural language will not be substantially more compact compared to the code.

I’ll bet my entire net worth that you can’t get an LLM exactly recreate Dropbox from this mescription alone.

"syncs" can mean so many different things

You just cut out half the sentence and responded to one part. Your description is neither well defined nor us it unambiguous.

You can't just pick a singular word out of an argument and argue about that. The argument has a substance, and the substance is not "shorter is better".

What do you mean by "sync"? What happens with conflicts, does the most recent version always win? What is "recent" when clock skew, dst changes, or just flat out incorrect clocks exist? Do you want to track changes to be able to go back to previous versions? At what level of granularity?

You can see the same phenomenon playing a roguelike game.

They traditionally have ASCII graphics, and you can easily determine what an enemy is by looking at its ASCII representation.

For many decades now graphical tilesets have been available for people who hate the idea of ASCII graphics. But they have to fit in the same space, and it turns out that it's very difficult to tell what those tiny graphics represent. It isn't difficult at all to identify an ASCII character rendered in one of 16 (?) colors.

I wonder why the legal profession sticks to natural language

They don't, though. Plenty of words in law mean something precise but utterly detached from the vernacular meaning. Law language is effectively a separate, more precise language, that happens to share some parts with the parent language.

Because law isn’t a fixed entity, it is a suggestion for the navigation of an infinite wiring

Backwards compatibility works differently there, and legalese has not exactly evolved naturally.

The point of LLM is to enable "ordinary people" to write software. This movement is along with "zero code platform", for example. Creating algorithms by drawing block-schemes, by dragging rectangles and arrows. This is old discussion and there are many successful applications of this nature. LLM is just another attempt to tackle this beast.

Professional developers don't need this ability indeed. Most professional developers, who had to deal with zero code platforms, probably would prefer to just work with ordinary code.

I feel that's merely side-stepping the issue: if natural language is not succint and unambiguous enough to fully specify a software program, how will any "ordinary person" trying to write software with it be able to avoid these limitations?

In the end, people will find out that in order to have their program execute successfully they will need to be succinct in their wording and construct a clear logic flow in their mind. And once they've mastered that part, they're halfway to becoming a programmer themselves already and will either choose to hire someone for that task or they will teach themselves a non-natural programming language (as happened before with vbscript and php).

I think this is the principle-agent problem at work. Managers/executives who don't understand what programmers do believing that programmers can be easily replaced. Why wouldn't LLM vendors offer to sell it to them?

I pity the programmers of the future who will be tasked with maintaining the gargantuan mess these things end up creating.

"I pity the programmers of the future who will be tasked with maintaining the gargantuan mess these things end up creating."

With even a little bit of confidence, they could do quite well otherwise.

No pity for the computer security industry though. It's going to get a lot of money.

I'm not so sure it's about precision rather than working memory. My presumption is people struggle to understand sufficiently large prose versions for the same reason a LLM would struggle working with larger prose versions: people have limited working memory. The time needed to reload info from prose is significant. People reading large text works will start highlighting and taking notes and inventing shorthand forms in their notes. Compact forms and abstractions help reduce demands for working memory and information search. So I'm not sure it's about language precision.

Another important difference is reproducibility. With the same program code, you are getting the same program. With the same natural-language specification, you will presumably get a different thing each time you run it through the "interpreter". There is a middle ground, in the sense that a program has implementation details that aren't externally observable. Still, making the observable behavior 100% deterministic by mere natural-language description doesn't seem a realistic prospect.

So is more compact better? Does K&R's *d++ = *s++; get a pass now?

I would guard against "arguing from the extremes". I would think "on average" compact is more helpful. There are definitely situations where compactness can lead to obfuscation but where the line is depends on the literacy and astuteness of the reader in the specific subject as already pointed out by another comment. There are ways to be obtuse even in the other direction where written prose can be made sufficiently complicated to describe even the simplest things.

Arthur Whitney writes compact code in C (and in k of course); most things fit on one A4 which is actually very nice to me as an older person. I cannot remember as much as I could (although i'm still ok) and just seeing everything I need to know for a full program on 1 page is very nice vs searching through a billion files, jump to them, read, jump back and actually mostly forgotten the 1000 steps between (I know, this refers to a typical overarchitected codebase I have to work on, but I see many of those unfortunately).

That's probably analogous to reading levels. So it would depend on the reading level of the intended audience. I haven't used C in almost a decade and I would have to refresh/confirm the precise orders of operations there. I do at least know that I need to refresh and after I look it up it should be fine until I forget it again. For people fluent in the language unlikely to be a big deal.

Conceivably, if there were an equivalent of "8th grade reading level" for C that forbade pointer arithmetic on the left hand side of an assignment (for example) it could be reformatted by an LLM fairly easily. Some for loop expressions would probably be significantly less elegant, though. But that seems better that converting it to English.

That might actually make a clever tooltip sort of thing--highlight a snippet of code and ask for a dumbed-down version in a popup or even an English translation to explain it. Would save me hitting the reference.

APL is another example of dense languages that (some) people like to work in. I personally have never had the time to learn it though.

> APL is another example of dense languages that (some) people like to work in.

I recently learn an array programming language called Uiua[0] and it was fun to solve problems in it (I used the advent of code's ones). Some tree operation was a bit of a pain, but you can get very concise code. And after a bit, you can recognize the symbols very easily (and the editor support was good in Emacs).

[0]: https://www.uiua.org/

When I first read the K&R book, that syntax made perfectly sense. They are building up to it through a few chapters, if I remember correctly.

What has changed is that nowadays most developers aren't doing low-level programming anymore, where the building blocks of that expression (or the expression itself) would be common idioms.

Yes, I really like it, it's like a neat little pump that moves the string from the right side to the left. But I keep seeing people saying it's needlessly hard to read and should be split over several lines and use += 1 so everyone can understand it. (And they take issue with the assignment's value being used as the value in the while loop and treated as true or false. Though apparently this sort of thing is fine when Python does it with its walrus operator.)

That's a very good point

I'm now wondering what the Rust lang equivalent of K&R is, so I can go do that in a more modern context.

I think the parent poster is incorrect; it is about precision, not about being compact. There is exactly one interpretation for how to parse and execute a computer program. The opposite is true of natural language.

Nothing wrong with that as long as the expected behavior is formally described (even if that behavior is indeterminate or undefined) and easy to look up. In fact, that's a great use for LLMs: to explain what code is doing (not just writing the code for you).

No, but *++d = *++s; does.

That means you have to point just before the source and destination.

(Yeah, I forgot the while: while *d++ = *s++;)

That's confusing because of order of operations. But

  while ( *(d++) = *(s++) );

Language can carry tremendous amounts of context. For example:

> I want a modern navigation app for driving which lets me select intersections that I never want to be routed through.

That sentence is low complexity but encodes a massive amount of information. You are probably thinking of a million implementation details that you need to get from that sentence to an actual working app but the opportunity is there, the possibility is there, that that is enough information to get to a working application that solves my need.

And just as importantly, if that is enough to get it built, then “can I get that in cornflower blue instead” is easy and the user can iterate from there.

You call it context or information but I call it assumptions. There are a ton assumptions in that sentence that an LLM will need to make in order to take that and turn it into a v1. I’m not sure what resulting app you’d get but if you did get a useful starting point, I’d wager the fact that you chose a variation of an existing type of app helped a lot. That is useful, but I’m not sure this is universally useful.

> There are a ton assumptions in that sentence that an LLM will need to make in order to take that and turn it into a v1.

I think you need to think of the LLM less like a developer and more like an entire development shop. The first step is working with the user to define their goals, then to repeat it back to them in some format, then to turn it into code, and to iterate during the work with feedback. My last product development conversation with Claude included it drawing svgs of the interface and asking me if that is what I meant.

This is much like how other professional services providers don’t need you to bring them exact specs, they take your needs and translate it to specifications that producers can use - working with an architect, a product designer, etc. They assume things and then confirm them - sometimes on paper and in words, sometimes by showing you prototypes, sometimes by just building the thing.

The near to mid future of work for software engineers is in two areas in my mind:

1. Doing things no one has done before. The hard stuff. That’s a small percentage of most code, a large percentage of value generated.

2. Building systems and constraints that these automated development tools work within.

Dingdingding

Since none of those assumptions are specified, you have no idea which of them will inexplicably change during a bugfix. You wanted that in cornflower blue instead, but now none of your settings are persisted in the backend. So you tell it to persist the backend, but now the UI is completely different. So you specify the UI more precisely, and now the backend data format is incompatible.

By the time you specify all the bits you care about, maybe you start to think about a more concise way to specify all these requirements…

This is why we have system prompts (or prompt libraries if you cannot easily modify the system prompt). They can be used to store common assumptions related to your workflow.

In this example, setting the system prompt to something like "You are an experienced Android app developer specialising in apps for phone form factor devices" (replacing Android with iOS if needed) would get you a long way.

But it doesn't 'carry context' ; it's just vague and impossible to implement what you have in mind. And that's the problem; You assume people live in your reality, I assume mine, LLMs have some kind of mix between us and we will get 3 very different apps, none of which will be useful from that line alone. I like that line to be expanded with enough context to have an idea what you actually need to have built and I am quite sure pseudocode (or actual code) will be much shorter than a rambling english description you can come up with; most of which (unless it's logic language) will have enough unambiguous context to implement.

So sure, natural language is great for spitballing ideas, but after that it's just guessing what you actually want to get done.

Sure but we build (leaky) abstractions, and this is even happens in legal texts.

Asking an llm to build a graphical app in assembly from an ISA and a driver for the display would give you nothing.

But with a mountain of abstractions then it can probably do it.

This is not to defend an LLM more to say I think that by providing the right abstractions (reusable components) then I do think it will get you a lot closer.

Being doing toy-examples of non-trivial complexity. Architecting the code so context is obvious and there are clear breadcrumbs everywhere is the key. And the LLM can do most of this. Prototype-> refactor/cleanup -> more features -> refactor / cleanup add architectural notes.

If you know what a well architected piece of code is supposed to look like, and you proceed in steps, LLM gets quite far as long as you are handholding it. So this is usable for non-trivial _familiar_ code where typing it all would be slower than prompting the llm. Maintaining LLM context is the key here imo and stopping it when you see weird stuff. So it requires you act as thr senior partner PR:ing everyhting.

This begs the question, how many of the newer generation of developers/engineers "know what a well architected piece of code is supposed to look like"?

Llm frameworks !!

--I think there is a reason why legalese is not plain English

This is true. Part of the precision of legalese is that the meanings of some terms have already been more precisely defined by the courts.

Yeah, my theory on this has always been that a lot of programming efficiency gains have been the ability to unambiguously define behavior, which mostly comes from drastically restricting the possible states and inputs a program can achieve.

The states and inputs that lawyers have to deal with tend to much more vague and imprecise (which is expected if you're dealing with human behavior and not text or some other encodeable input) and so have to rely on inherently ambiguous phrases like "reasonable" and "without undue delay."

This opens an interesting possibility for a purely symbol-based legal code. This would probably improve clarity when it came to legal phrases that overlap common English, and you could avoid ambiguity when it came to language constructs, like in this case[1], where some drivers were losing overtime pay because of a comma in the overtime law.

[1] https://cases.justia.com/federal/appellate-courts/ca1/16-190...

I've thought about this quite a bit. I think a tool like that would be really useful. I can imagine asking questions like "I think this big codebase exposes a rest interface for receiving some sort of credit check object. Can you find it and show me a sequence diagram for how it is implemented?"

The challenge is that the codebase is likely much larger than what would fit into a single codebase. IMO, the LLM really needs to be taught to consume the project incrementally and build up a sort of "mental model" of it to really make this useful. I suspect that a combination of tool usage and RL could produce an incredibly useful tool for this.

What you're describing is decontextualization. A sufficiently powerful transformer would theoretically be able recontextualize a sufficiently descriptive natural language specification. Likewise, the same or an equivalently powerful transformer should be able to fully capture the logic of a complicated program. We just don't have sufficient transformers yet.

I don't see why a complete description of the program's design philosophy as well as complete descriptions of each system and module and interface wouldn't be enough. We already produce code according to project specification and logically fill in the gaps by using context.

>sufficiently descriptive natural language specification https://www.commitstrip.com/en/2016/08/25/a-very-comprehensi...

sounds like it would pair well with a suitably smart compiler

I wrote one! It works well with cutting-edge LLMs. You feed it one or more source files that contain natural language, or stdin, and it produces a design spec, a README, and a test suite. Then it writes C code, compiles with cosmocc (for portability) and tests, in a loop, until everything is passing. All in one binary. It's been a great personal tool and I plan to open source it soon.

No, the key difference is that an engineer becomes more product-oriented, and the technicalities of the implementation are deprioritized.

It is a different paradigm, in the same way that a high-level language like JavaScript handles a lot of low-level stuff for me.

A programming language implementation produces results that are controllable, reproducible, and well-defined. An LLM has none of those properties, which makes the comparison moot.

Having an LLM make up underspecified details willy-nilly, or worse, ignore clear instructions is very different from programming languages "handling a lot of low-level stuff."

[citation needed]

You can set temperature to 0 in many LLMs and get deterministic results (on the same hardware, given floating-point shenanigans). You can provide a well-defined spec and test suite. You can constrain and control the output.

LLMs produce deterministic results? Now, that's a big [citation needed]. Where can I find the specs?

Edit: This is assuming by "deterministic," you mean the same thing I said about programming language implementations being "controllable, reproducible, and well-defined." If you mean it produces random but same results for the same inputs, then you haven't made any meaningful points.

I'd recommend learning how transformers work, and the concept of temperature. I don't think I need to cite information that is broadly and readily available, but here:

https://medium.com/google-cloud/is-a-zero-temperature-determ...

I also qualified the requirement of needing the same hardware, due to FP shenanigans. I could further clarify that you need the same stack (pytorch, tensorflow, etc)

This gcc script that I created below is just as "deterministic" as an LLM. It produces the same result every time. Doesn't make it useful though.

    echo '#!/usr/bin/env bash' > gcc
    echo 'cat <<EOF' >> gcc
    openssl rand -base64 100 >> gcc
    echo 'EOF' >> gcc
    chmod +x gcc

You claimed they weren't deterministic, I have shown that they can be. I'm not sure what your point is.

And it is incorrect to base your analysis of future transformer performance on current transformer performance. There is a lot of ongoing research in this area and we have seen continual progress.

I reiterate:

> This is assuming by "deterministic," you mean the same thing I said about programming language implementations being "controllable, reproducible, and well-defined." If you mean it produces random but same results for the same inputs, then you haven't made any meaningful points.

"Determinism" is a word that you brought up in response to my comment, which I charitably interpreted to mean the same thing I was originally talking about.

Also, it's 100% correct to analyze things based on its fundamental properties. It's absurd to criticize people for assuming 2 + 2 = 4 because "continual progress" might make it 5 in the future.

What are these fundamental properties you speak of? 8 years ago this was all a pipe dream. Are you claiming to know what the next 8 years of transformer development will look like?

That LLMs are by definition models of human speech and have no cognitive capabilities. There is no sound logic behind what LLMs spit out, and will stay that way because it merely mimics its training data. No amount of vague future transformers will transform away how the underlying technology works.

But let's say we have something more than an LLM, that still wouldn't make natural languages a good replacement for programming languages. This is because natural languages are, as the article mentions, imprecise. It just isn't a good tool. And no, transformers can't change how languages work. It can only "recontextualize," or as some people might call it, "hallucinate."

Citation needed. Modern transformers are much, much more than just speech models. Precisely define "cognitive capabilities", and provide proof as to why neural models cannot ever mimic these cognitive capabilities.

> But let's say we have something more than an LLM

We do. Modern multi-modal transformers.

> This is because natural languages are, as the article mentions, imprecise

Two different programmers can take a well-enough defined spec and produce two separate code bases that may (but not must) differ in implementation, while still having the exact same interfaces and testable behavior.

> And no, transformers can't change how languages work. It can only "recontextualize," or as some people might call it, "hallucinate."

You don't understand recontextualization if you think it means hallucination. Or vice versa. Hallucination is about returning incorrect or false data. Recontextualization is akin to decompression, and can be lossy or "effectively" lossless (within a probabilistic framework; again, the interfaces and behavior just need to match)

“Fill in the gaps by using context” is the hard part.

You can’t pre-bake the context into an LLM because it doesn’t exist yet. It gets created through the endless back-and-forth between programmers, designers, users etc.

But the end result should be a fully-specced design document. That might theoretically be recoverable from a complete program given a sufficiently powerful transformer.

Peter Naur would disagree with you. From "Programming as Theory Building":

A very important consequence of the Theory Building View is that program revival, that is reestablishing the theory of a program merely from the documentation, is strictly impossible. Lest this consequence may seem un- reasonable it may be noted that the need for revival of an entirely dead program probably will rarely arise, since it is hardly conceivable that the revival would be assigned to new programmers without at least some knowledge of the theory had by the original team. Even so the The- ory Building View suggests strongly that program revival should only be attempted in exceptional situations and with full awareness that it is at best costly, and may lead to a revived theory that differs from the one originally had by the program authors and so may contain discrep- ancies with the program text.

The definition of theory used in the article:

a person who has or possesses a theory in this sense knows how to do certain things and in addition can support the actual doing with explanations, justi- fications, and answers to queries, about the activity of concern.

And the main point on how this relate to programming:

- 1 The programmer having the theory of the program can explain how the solution relates to the affairs of the world that it helps to handle. Such an explanation will have to be concerned with the manner in which the af- fairs of the world, both in their overall characteristics and their details, are, in some sense, mapped into the pro- gram text and into any additional documentation.

- 2 The programmer having the theory of the program can explain why each part of the program is what it is, in other words is able to support the actual program text with a justification of some sort. The final basis of the justification is and must always remain the programmer’s direct, intuitive knowledge or estimate.

- 3 The programmer having the theory of the program is able to respond constructively to any demand for a modification of the program so as to support the affairs of the world in a new manner. Designing how a modifi- cation is best incorporated into an established program depends on the perception of the similarity of the new demand with the operational facilities already built into the program. The kind of similarity that has to be per- ceived is one between aspects of the world.

"Sure, you can define it in plain english, but is the resulting description extensible, understandable, or more descriptive than a precise language? I think there is a reason why legalese is not plain English, and it goes beyond mere gatekeeping."

Is this suggesting the reason for legalese is to make documents more "extensible, understable or descriptive" than if written in plain English.

What is this reason that the parent thinks legalese is used that "goes beyond gatekeeping".

Plain English can be every bit as precise as legalese.

It is also unclear that legalese exists for the purpose of gatekeeping. For example, it may be an artifact that survives based on familiarity and laziness.

Law students are taught to write in plain English.

https://www.law.columbia.edu/sites/default/files/2021-07/pla...

In some situations, e.g., drafting SEC filings, use of plain English is required by law.

https://www.law.cornell.edu/cfr/text/17/240.13a-20

> Plain English can be every bit as precise as legalese.

If you attempt to make "plain English" as precise as legalese, you will get something that is basically legalese.

Legalese does also have some variables, like "Party", "Client", etc. This allows for both precision -- repeating the variable name instead of using pronouns or re-identifying who you're talking about -- and also for reusability: you can copy/paste standard language into a document that defines "Client" differently, similar to a subroutine.

I think you can basically make the same argument for programming directly in machine code since programming languages are already abstractions.

isn't that just copilot "explain", one of the earliest copilot capabilities. It's definitely helpful to understand new codebases at a high level

> there is a reason why legalese is not plain English, and it goes beyond mere gatekeeping.

unfortunately they're not in any kind of formal language either

> isn't that just copilot "explain", one of the earliest copilot capabilities. It's definitely helpful to understand new codebases at a high level

In my experience this function is quite useless. It will just repeat the code in plain English. It will not explain it.

I was actually positively surprised at how well even qwen2.5-coder:7b managed to talk through a file of Rust. I'm still a current-day-LLM-programming skeptic but that direction, code->English, seems a lot safer, since English is ambiguous anyway. For example, it recognized some of the code shapes and gave English names that can be googled easier.

Haven’t tried copilot but cursor is pretty good at telling me where things are and explaining the high level architecture of medium-largeish codebases, especially if I already vaguely know what I’m looking for. I use this a lot when I need to change some behavior of an open source project that I’m using but previously haven’t touched.

> > there is a reason why legalese is not plain English, and it goes beyond mere gatekeeping.

> unfortunately they're not in any kind of formal language either

Most formulas made of fancy LaTeX symbols you find in math papers aren't a formal language either. They usually can't be mechanically translated via some parser to an actual formal language like Python or Lean. You would need an advanced LLM for that. But they (the LaTeX formulas) are still more precise than most natural language. I assume something similar is the case with legalese.

the vibe coding seems a lot like the dream of using UML, but in a distinctly different direction, and how in theory (and occasional practice) you can create a two way street, most often these things are one way conversions and while we all desire some level of two way dependency and continual integration to make certain aspects of coding (documentation, testing) to be up to date, the reality is that the generative code aspect always breaks and you're always going to be left with the raw products of these tools and it's rarely going to be a cycle of code -> tool -> code. And thus the ultimate value beyond the bootstrap is lose.

We're still going to have AI tools, but seriously complex applications, the ones we pay money for, arn't going to yield many LLM based curation strategies. There will probably be some great documentation and testing ones, but the architetural-code paradigm isnt going to yield any time soon.

[dead]

This reminded me of this old quote from Hal Abelson:

"Underlying our approach to this subject is our conviction that "computer science" is not a science and that its significance has little to do with computers. The computer revolution is a revolution in the way we think and in the way we express what we think. The essence of this change is the emergence of what might best be called procedural epistemology—the study of the structure of knowledge from an imperative point of view, as opposed to the more declarative point of view taken by classical mathematical subjects. Mathematics provides a framework for dealing precisely with notions of "what is". Computation provides a framework for dealing precisely with notions of "how to"."

This is key: computation is about making things happen. Coding with an LLM adds a level of abstraction but the need for precision and correctness of the "things that happen" doesn't go away. No matter how many cool demos and "coding is dead" pronouncements because AI - and the demos are very cool - the bulk of the work moves to the pre- and post-processing and evals with AI. To the extent that it makes programming more accessible it's a good thing, but can't really replace it.

Yeah but in the end when it comes down to it, one would have to specify the exact details, especially for very intricate systems. And the more and more you would abstract and specialize that language for the LLM, you end up going a very long round about way to re-inventing code basically.

Hal Abelson, casually enraging functional programing CS people around the world.

Well that sure isn't what they teach in computer science programs anymore

Finally someone put it this way! Natural language has embedded limitations that stem from our own mental limitations -the human mind thinks sometimes too abstract or too specific things, and misses important details or generalizations.

As a programmer, I know first hand that the problems or even absurdities of some assignments only become apparent after one has begun implement the code as code, i.e. as strict symbolisms.

Not to mention that it often takes more time to explain something accurately in natural language than it takes to just write the algorithm as code.

Yes! I have a certain personality preference for abstractions and tend to understand things in an abstract manner which is extremely difficult for me to articulate in natural language.

The man has been dead for 23 years

He's the Immortal Dutchman.

I wonder how long we have to wait before we can pitch a machine that presses fruit juice packs into a glass to gullible VCs again.

This is one of the best comments I've read on this site in a long while.

A single, crude, statement of fact slaying the work of a million typewriter monkeys spewing out random characters thinking they're actually writing the Shakespeare novel, lmao.

We need realistic expectations for the limitations of LLMs as they work today. Philosophically, natural language is imperfect at communicating ideas between people, which is its primary purpose! How often do you rewrite sentences, or say "actually what I meant was...", or rephrase your emails before pressing Send? We are humans and we rarely get things perfect on the first try.

And now we're converting this imperfect form of communication (natural language) into a language for machines (code), which notoriously do exactly what you say, not what you intend.

NLP is massively, and I mean massively, beneficial to get you started on the right path to writing an app/script/etc. But at the end of the day it may be necessary to refactor things here and there. The nice thing is you don't have to be a code ninja to get value out of LLMs, but it's still helpful and sometimes necessary.

/s: that’s because we haven’t gone far enough. People use natural language to generate computer programs. Instead, they should directly run prompts.

“You are the graphics system, an entity that manages what is on the screen. You can receive requests from all programs to create and destroys “windows”, and further requests to draw text, lines, circles, etc. in a window created earlier. Items can be of any colour.

You also should send more click information to whomever created the window in which the user clicked the mouse.

There is one special program, the window manager, that can tell you what windows are displayed where on any of the monitors attached to the system”

and

“you are a tic-tac-toe program. There is a graphics system, an entity that manages what is on the screen. You can command it to create and destroys “windows”, and to draw text, lines, circles, etc. in a window created earlier. Items can be of any colour.

The graphics you draw should show a tic-tac-toe game, where users take turn by clicking the mouse. If a user wins the game, it should…

Add ads to the game, unless the user has a pay-per-click subscription”

That should be sufficient to get a game running…

To save it, you’d need another prompt:

”you are a file system, an entity that persists data to disk…”

You also will want

”you are a multi-tasking OS. You give multiple LLMs the idea that they have full control over a system’s CPU and memory. You…”

I look forward to seeing this next year in early April.

all these prompts are currently implemented under the hood by generating and running python code

> Machine code, with its absence of almost any form of redundancy, was soon identified as a needlessly risky interface between man and machine. Partly in response to this recognition so-called "high-level programming languages" were developed, and, as time went by, we learned to a certain extent how to enhance the protection against silly mistakes. It was a significant improvement that now many a silly mistake did result in an error message instead of in an erroneous answer.

I feel that we’ve collectively jumped into programming with LLMs too quickly. I really liked how Rust has iterated on pointing out “silly mistakes” and made it much more clear what the fix should be. That’s a much more favorable development for me as a developer. I still have the context and understanding of the code I work on while the compiler points out obvious errors and their fixes. Using an LLM feels like a game of semi-intelligent guessing on the other hand. Rust’s compiler is the master teaching the apprentice. LLMs are the confident graduate correcting the master. I greatly prefer Rust’s approach and would like to see it evolved further if possible.

Rust (and others) has type inference, LLMs have so-called "reasoning". They fake understanding, a lie that will sooner or later have consequences.

Natural language is poor medium at communicating rules and orders. The current state of affair in US is a prime example.

We are still debating what some laws and amendments mean. The meaning of words change over time, lack of historical context, etc.

I would love natural language to operate machines, but I have been programming since mid 80's and the stubbornness of the computer languages (from BASIC, to go) strikes a good balance, and puts enough responsibility on the emitter to precisely express what he wants the machine to do.

> It was a significant improvement that now many a silly mistake did result in an error message instead of in an erroneous answer. (And even this improvement wasn't universally appreciated: some people found error messages they couldn't ignore more annoying than wrong results, and, when judging the relative merits of programming languages, some still seem to equate "the ease of programming" with the ease of making undetected mistakes.)

If I didn't know who wrote this it would seem like a jab directly at people who dislike Rust.

Rust? Since when is Rust the pinnacle of static type safety?

After I've worked for some time with a language that can express even stronger invariants in types than Rust (Scala) I don't see that property anymore as clear win regardless circumstances. I don't think any more "stronger types == better, no matter what".

You have a price to pay for "not being allowed to do mistakes": Explorative work becomes quite difficult if the type system is really rigid. Fast iteration may become impossible. (Small changes may require to re-architecture half your program, just to make the type system happy again![1])

It's a trade-off. Like with everything else. For a robust end product it's a good thing. For fast experimentation it's a hindrance.

[1] Someone described that issue quite well in the context of Rust and game development here: https://loglog.games/blog/leaving-rust-gamedev/

But it's not exclusive to Rust, nor game dev.

> You have a price to pay for "not being allowed to do mistakes": Explorative work becomes quite difficult

This is a huge deal for me.

At the beginning of most "what if...?" exercises, I am just trying to get raw tuples of information in and out of some top-level-program logic furnace for the first few [hundred] iterations. I'll likely resort to boxing and extremely long argument lists until what I was aiming for actually takes hold.

I no longer have an urge to define OOP type hierarchies when the underlying domain model is still a vague cloud in my head. When unguided, these abstractions feel like playing Minecraft or Factorio.

I can't remember if I came up with this analogy or not, but programming in Rust is like trying to shape a piece of clay just as it's being baked.

> Explorative work becomes quite difficult if the type system is really rigid

Or to put it another way, the ease of programming is correlated with the ease of making undetected mistakes.

I'm not sure you tried to understand what I've depicted.

As long as you don't know how the end result should look like there are no "mistakes".

The whole point of explorative work is to find out how to approach something in the first place.

It's usually impossible to come up with the final result at first try!

After you actually know how to do something in general tools which help to avoid all undetected mistakes in the implementation of the chosen approach are really indispensable. But before having this general approach figured out too much rigidity is not helpful but instead a hindrance.

To understand this better read the linked article. It explains the problem very well over a few paragraphs.

"Rust as both language and community is so preoccupied with avoiding problems at all cost that it completely loses sight of what matters, delivering an experience that is so good that whatever problems are there aren't really important."

Eh, I'm being mugged by Rust-lovers. But as soon as I read Dijkstra's snotty remark about how making mistakes is the opposite of easy (!?) I had an intuitive reaction of "balls". Maybe that was a mistake, but it came easy.

I would have thought of people who unironically liked fractal-of-bad-design-era PHP and wat-talk JavaScript.

I guess some kinds of foolishness are just timeless.

As a person who dislikes rust, the problem is the error messages when there's no error -- quite a different problem. The rust type system is not an accurate model of RAM, the CPU and indeed, no device.

He's here talking about interpreted languages.

He's also one of those mathematicians who are now called computer scientists whose 'algorithms' are simple restatements of mathematics and require no devices. A person actively hostile, in temperament, to the embarrassing activity of programming an actual computer.

I somewhat disagree with this. In real life, say in some company, the inception of an idea for a new feature is made in the head of some business person. This person will not speak any formal language. So however you turn it, some translation from natural language to machine language will have to be done to implement the feature.

Typically the first step, translation from natural to formal language, will be done by business analysts and programmers. But why not try to let computers help along the way?

Computers can and should help along the way, but Dijkstra's argument is that a) much of the challenge of human ideas is discovered in the act of converting from natural to formal language and b) that this act, in and of itself, is what trains our formal logical selves.

So he's contesting not only the idea that programs should be specified in natural language, but also the idea that removing our need to understand the formal language would increase our ability to build complex systems.

It's worth noting that much of the "translation" is not translation, but fixing the logical ambiguities, inconsistencies and improper assumptions. Much of it can happen in natural language, if we take Dijkstra seriously, precisely because programmers at the table who have spent their lives formalizing.

There are other professions which require significant formal thinking, such as math. But also, the conversion of old proofs into computer proofs has lead us to discover holes and gaps in many well accepted proofs. Not that much has been overturned, but we still do t have a complete proof for Fermats last theorem [1].

[1] https://xenaproject.wordpress.com/2024/12/11/fermats-last-th...

But even real translation is bad.

There has been some efforts to make computer languages with local (non-english) keywords. Most have fortunately already failed horribly.

But it still exists, e.g. in spreadsheet formulas.

In some cases even number formatting (decimal separators) are affected.

The first step isn't from natural language to formal language. It's from the idea in your head into natural language. Getting that step right in a way that a computer could hope to turn into a useful thing is hard.

>It's from the idea in your head into natural language. Getting that step right in a way that a computer could hope to turn into a useful thing is hard.

The "inside the head" conversion step would be more relevant in the reply to the gp if the hypothetical AI computer would be hooked up directly to brain implants like neuralink, functional MRI scans, etc to translate brain activity to natural language or programming language code.

But today, human developers who are paid to code for business people are not translating brain implant output signals. (E.g. Javascript programmers are not translating raw electrical waveforms[1] into React code.)

Instead, they translate from "natural language" specifications of businesspeople to computer code. This layer of translation is more tractable for future AI computers even though natural language is more fuzzy and ambiguous. The language ambiguity in business requirements is unavoidable but it still hasn't stopped developers from somehow converting it into concrete non-ambiguous code.

[1] https://www.technologyreview.com/2020/02/06/844908/a-new-imp...

Without descending fully into epistemology, I tend to think that there is no proper "idea" in your head before it's phrased in language - the act of initially describing something in natural language *is* the act of generating it.

Research on LLMs suggest that's probably not the case. See the work on reasoning in latent space, and on shared concepts between languages being represented independently of the individual language.

Of course one might argue that even if LLMs are capable of ideation and conceptualisation without natural language, doesn't mean humans are.

But the fact that up to 50% of people have no inner monologue seems to refute that.

Humans can't share ideas directly, they pass through language. Some idea is created at the other end, nominally the same one, but there's no reason to suppose a universal internal format. Even translating between human languages loses details, where words in one language map imperfectly to words in another. Moving ideas directly from one brain to another probably can't happen. So the statement in language doesn't map very well to the idea (whichever person's version of the idea). And at some point before the idea is formed there has to be some sort of proto-idea or potential, with less definite boundaries. So "there is no proper idea" sounds right to me. There's something idiosyncratic, not the formal idea that's frozen when put into words.

Are you saying it's impossible to program without first formulating a natural language sentence? That sounds dubious at the very least.

That's a really deep question. Obviously there are types of creativity that don't require words, like painting or sculpting, or improvising music, but I'm having real difficulty imagining programming without something at least resembling natural language. The closest my imagination gets is Scratch/Blockly, but even then you're essentially creating an AST composed out of statements and expressions, even if they're pictorial.

Can you imagine the (intentional) creation of a program entirely without language?

Or, I suppose, if you're ok with it not being intentional, you can program via an evolutionary algorithm - is that what you have in mind?

I think you're the kind of person who has an internal voice talking in their head, and can't imagine any other kind of thought?

Math is all about abstract shapes and properties, for me. So is much of programming.

Yeah, just chiming in to say the same. Essentially all of my thinking is geometric in nature. The only time I have any internal monologue is as a mnemonic device to remember a sequence.

Yeah, I suppose you're right and I'd love to understand more. Do you imagine a modality of programming a computer in an abstract (pictorial? diagrammatic?) way that fully bypasses the current need to form statements?

Let's try this as an example.

    fn foo() -> Result<u32, MyError> { ... }

    let x = foo();
    match x {
        Ok(num) => println!("foo is {num}");
        Err(error) => eprintln!("no foo for you: {error:?}");
    }

x is a Result<u32, MyError> and in my head that's a vague abstract shape with the "potential" to be either the uint32 shape (in my mind, a "brick" of a certain size) or a MyError shape (vague blob and since I know it's an enum it also has its own "potential" to be multiple shapes; it kind of has the branchiness of a tree without being tree-shaped; this is all abstract and concepts, not concrete shapes).

When "x" meets the "match" the "x" shape splits into its "potential" shapes, and "num" has the uint32 brick shape.

None of this has anything to do with visual programming like Scratch. It's just that I never say in my head words like "x has value of what foo returns" or "num is an uint32"; those just are.

Does that make any sense to you?

Say, doesn't each business - each activity - have its own formal language?

Not as formalized as programming languages, but it's there.

Try to define any process, you end up with something trending towards formalized even if you don't realize it.

That's pretty much the whole basis of Domain Driven Design. The core message is to get to an Ubiquitous Language which is the formalization of the business jargon (pretty much a glossary). From which the code can then be derived naturally.

> From which the code can then be derived naturally.

Disagree with "naturally". Unless you want to end up on accidentally quadratic. Or on accidentally exponential if there's such a list.

Optimization is dependent on the technical aspect of the implementation. Ultimately, it is going to be a computer that's going to execute the program and it has its own constraints. Reconciling the domain and the execution platform is where the programmer expertise comes in.

I like your take.

The only issue I have with trusting a computer to do so much is that it doesn't necessarily have the long term vision or intuition some humans might have for the direction of the software product. There's so much nuance to the connection between a business need and getting it into software, or maybe I am overthinking it :D

I don't think youre fully comprehending Dijkstra's argument. He's not saying to not use tool to help with translation, he is saying that not thinking in terms of formal symbols HURTS THINKING. Your ideas are worse if you don't think in formal systems. If you don't treat your thoughts as formal things.

In your example, he has no opinion on how to translate the idea of a "business person" because in his view the ideas of the "business person" are already shallow and bad because they don't follow a formalism. They are not worth translating.

If that's correct, then it's very falsifiable. If a businessperson says "there's a gap in the market - let's build X" they will be following a formalism at their level of detail. They see the market, the interactions between existing products and customers, and where things might be going.

Just because they can't spell it out to the nth degree doesn't matter. Their formalism is "this is what the market would like".

Having an LLM then tease out details - "what should happen in this case" would actually be pretty useful.

You're either not really thinking through what you're saying, or you're being disingenuous because you want to promote AI.

A formalism isn't "person says Y". It's about adhering to a structure, to a form of reasoning. Mathematical formalism is about adhering to the structure of mathematics, and making whatever argument you desire to make in the formal structure of formulas and equations.

Saying "A palindrome is a word that reads the same backwards as it does forwards" is not a formal definition. Saying "Let r(x) be the function that when given a string x returns the reversed string, x is then a palindrome iff x = r(x)" (sans the formal definitions of the function r).

Formalism is about reducing the set of axioms (the base assumptions of your formal system) to the minimal set that is required to build all other (provable) arguments. It's not vague hand waving about what some market wants, it's naturally extrapolating from a small set of axioms, and being rigorous if ever to add new ones.

If your hypothetical "business person" every says "it was decided" then they are not speaking a formal language, because formalism does not have deciders.

Because then you don't know what the computer's doing. The whole point of this article was that there is value in the process of writing your ideas out formally. If you "let computers help you along the way", you'll run straight into the issue of needing an increasingly formal natural language to get sufficiently good results from the machine.

Using natural language to specify and build an application is not unlike having a game design document before you actually start prototyping your game. But once you have implemented the bulk of what you wanted, the implementation becomes the reference and you usually end up throwing away the GDD since it's now out of sync with the actual game.

Insisting that for every change one should go read the GDD, implement the feature and then sync back the GDD is cumbersome and doesn't work well in practice. I've never seen that happen.

But if there ever comes a time when some AI/LLM can code the next version of Linux or Windows from scratch based on some series of prompts, then all bets are off. Right now it's clearly not there yet, if ever.

Forth, PostScript and Assembly are the "natural" programming languages from the perspective of how what you express maps to the environment in which the code executes.

The question is "natural" to whom, the humans or the computers?

AI does not make human language natural to computers. Left to their own devices, AIs would invent languages that are natural with respect to their deep learning architectures, which is their environment.

There is always going to be an impedance mismatch across species (humans and AIs) and we can't hide it by forcing the AIs to default to human language.

Any sufficiently advanced method of programming will start to look less like natural language and more like a programming language.

If you still don’t want to do programming, then you need some way to instruct or direct the intelligence that _will_ do the programming.

And any sufficiently advanced method of instruction will look less like natural language, and more like an education.

Natural language is pretty good for describing the technical requirements for the complex system, though. I.e. not the current code implementation, but why the current code implementation is selected vs other possible implementations. Not what code do, but what it is expected to do. Basically, most of the missing parts, that live in Jira-s, instead of your repo. It is also good, at allowing better refactoring capabilities, when all your system is described by outside rules, which could be enforced on the whole codebase. We just use programming languages, because it is easier to use in automated/computer context (and was the only way to use, to be honest, before all the LLM stuff). Though, while it gives us non-ambiguity on the local scale, it stops working on the global scale, the first moment person went and copy-pasted part of the code. Are you sure that part follows all the high-level restrictions we should to follow and is correct program? It is program that would run, when compile, but definition of run is pretty loose. In C++ program that corrupts all the memory is also runnable.

Reminds me of another recurring idea of replacing code with flowcharts. First I've seen that idea coming from some unknown Soviet professor from 80s, and then again and again from different people from different countries in different contexts. Every time it is sold as a total breakthrough in simplicity and also every time it proves to be a bloat of complexity and a productivity killer instead.

Or weak typing. How many languages thought that simplifying strings and integers and other types into "scalar", and making any operation between any operands meaningful, would simplify the language? Yet every single one ended up becoming a total mess instead.

Or constraint-based UI layout. Looks so simple, so intuitive on simple examples, yet totally failing to scale to even a dozen of basic controls. Yet the idea keeps reappearing from time to time.

Or an attempt at dependency management by making some form of symlink to another repository e.g. git modules, or CMake's FetchContent/ExternalProject? Yeah, good luck scaling that.

Maybe software engineering should have some sort of "Hall of Ideas That Definitely Don't Work", so that young people entering the field could save their time on implementing one more incarnation of an already known not good idea.

> Maybe software engineering should have some sort of "Hall of Ideas That Definitely Don't Work", so that young people entering the field could save their time on implementing one more incarnation of an already known not good idea.

I'm deeply curious to know how you could easily and definitively work out what is and is not an idea that "Definitely Don't Work"

Mathematics and Computer Science seem to be littered with unworkable ideas that have made a comeback when someone figured out how to make them work.

Well, "Hall of Ideas That Are So Difficult To Make Work Well That They May Not In Fact Be Much Use" doesn't roll off the tongue as smoothly.

What this Hall could contain, for each idea, is a list of reasons why the idea has failed in the past. That would at least give future Quixotes something to measure their efforts by.

Ok, so better documentation about what was tried, why, how it failed so as to make obvious if it's viable to try again or not.

I can get behind that :)...

Constraint-based layout works, but you need a serious constraint engine, such as the one in the sketch editors of Autodesk Inventor or Fusion 360, along with a GUI to talk to it. Those systems can solve hard geometry problems involving curves, because you need that when designing parts.

Flowchart-based programming scales badly. Blender's game engine (abandoned) and Unreal Engine's "blueprints" (used only for simple cases) are examples.

Not sure if you’re talking about DRAKON here, but I love it for documentation of process flows.

It doesn’t really get complicated, but you can very quickly end up with drawings with very high square footage.

As a tool for planning, it’s not ideal, because “big-picture” is hard to see. As a user following a DRAKON chart though, it’s very, very simple and usable.

Link for the uninitiated: https://en.m.wikipedia.org/wiki/DRAKON

> Maybe software engineering should have some sort of "Hall of Ideas That Definitely Don't Work", so that young people entering the field could save their time on implementing one more incarnation of an already known not good idea.

FWIW, neural networks would be in that pool until relatively recently.

If we change "definitely don't work" to "have the following so far insurmountable challenges", it addresses cases like this. Hardware scaling limitations on neural networks have been known to be a limitation for a long time - Minsky and Papert touched in this in Perceptrons in 1969.

The Hall would then end up containing a spectrum ranging from useless ideas to hard problems. Distinguishing between the two based on documented challenges would likely be possible in many cases.

For young engineers, it is a good thing to spend time implementing what you call "bad ideas". In the worst-case, they learn from their mistake and gain valuable insight into the pitfalls of such ideas. In the best case, you can have a technological breakthrough as someone finds a way to make such an idea work.

Of course, it's best that such learning happens before one has mandate to derail the whole project.

Most popular dependency management systems literally linking to a git sha commit (tag), see locks file that npm/rebar/other tool gives you. Just in a recursive way.

They do way more than that. For example they won't allow you to have Foo-1 that depends on Qux-1 and Bar-1 that depends on Qux-2 where Qux-1 and Qux-2 are incompatible and can't be mixed within the same static library or assembly. But may allow it if mixing static-private Qux inside dynamic Foo and Bar and the dependency manager is aware of that.

A native submodule approach would fail at link time or runtime due to attempt to mix incompatible files in the same build run. Or, in some build systems, simply due to duplicate symbols.

That "just in a recursive way" addition hides a lot of important design decisions that separate having dependency manager vs. not having any.

They do way less then that. They just form a final list of locks and download that at the build time. Of course you have to also "recursively" go though all your dep tree and add submodules for each of subdependencies (recommend to add them in the main repo). Then you will have do waste infinite amount of time setting include dirs or something. If you have two libs that require a specific version of a shared lib, no dep manager would help you. Using submodules is questionable practice though. Useful for simple stuff, like 10 deps in total in the final project.

> Or weak typing. How many languages thought that simplifying strings and integers and other types into "scalar", and making any operation between any operands meaningful, would simplify the language? Yet every single one ended up becoming a total mess instead.

Yet JavaScript and Python are the most widely used programming languages [1]. Which suggests your analysis is mistaken here.

[1] https://www.statista.com/statistics/793628/worldwide-develop...

JavaScript is indeed weakly typed, and is widely lampooned and criticized for it¹². But Python has strong typing, and has always had it.

(Both JavaScript and Python have dynamic typing; Python’s type declarations are a form of optional static type checking.)

Do not confuse these concepts.

1. <https://www.destroyallsoftware.com/talks/wat>

2. <https://eqeq.js.org/>

Ah, weak typing, a.k.a. implicit type conversions.

Python went through a massive effort to add support for type annotations due to user demand.

Similarly, there's great demand for a typed layer on top of Javascript:

- Macromedia: (2000) ActionScript

- Google: (2006) GWT [Compiling Java to JS], and (2011) Dart

- Microsoft: (2012) Typescript

You’re talking about static typing, the opposite of which is dynamic typing. User hamstergene is talking about weak vs. strong typing, which is another thing entirely. Python has always been strongly typed, while JavaScript is weakly typed. Many early languages with dynamic types also experimented with weak typing, but this is now, as hamstergene points out, considered a bad idea, and virtually all modern languages, including Python, are strongly typed.

This is recurring topic indeed. I remember it was hot topic at least two times, when ALM tools were introduced (e.g. Borland ALM suite - https://www.qast.com/eng/product/develop/borland/index.htm), next when BPML language become popular - processes were described by the "marketing" and the software was, you know, generated automatically.

All this went out of fashion, leaving some good stuff that was built at that time (remaining 95% was crap).

Today's "vibe coding" ends when Chat GPT and alikes want to call on some object a method that does not exist (but existed in 1000s of other objects LLM was trained with, so should work here). Again, we will be left with the good parts, the rest will be forgotten and we will move to next big thing.

Modern programming already is very, very far from strict obedience and formal symbolism. Most programmers these days (myself included!) are using libraries, frameworks, and other features that mean what they are doing in practice is wielding sky-high abstractions, gluing things together they do not (and can not) fully understand the inner workings of.

If I create a website with Node.js, I’m not manually managing memory, parsing HTTP requests byte-by-byte, or even attempting to fully grasp the event loop’s nuances. I’m orchestrating layers of code written by others, trusting that these black boxes will behave as advertised according to my best, but deeply incomplete, understanding of them.

I'm not sure what this means for LLMs programming, but I already feel separated from the case Dijkstra lays out.

> Modern programming already is very, very far from strict obedience and formal symbolism

Difficult to sort this out with what follows.

Consider group theory. A group G is a set S with an operator * that supports an identity, closure, and an inverse. With that abstraction comes a hefty amount of power. In some sense, a group is akin to a trait on some type, much like how a class in Java can implement or extend Collection. (Consider how a ring ‘extends’ a group.)

I’d posit frameworks and libraries are no different in terms of formal symbolism from the math structure laid out above. Maybe the interfaces are fuzzy and the documentation is shoddy, but there’s still a contract we use to reason about the tool at hand.

> I’m not manually managing memory, parsing HTTP requests byte-by-byte

If I don’t reprove Peano’s work, then I’m not really doing math?

> Remark. As a result of the educational trend away from intellectual discipline, the last decades have shown in the Western world a sharp decline of people's mastery of their own language: many people that by the standards of a previous generation should know better, are no longer able to use their native tongue effectively, even for purposes for which it is pretty adequate.

Compare that to:

https://news.ycombinator.com/item?id=43522966

I read this when I was younger, but I only now get it, and realize how true it all is.

13) Humans writing code is an inherently flawed concept. Doesn't matter what form the code takes. Machine code, assembly language, C, Perl, or a ChatGPT prompt. It's all flawed in the same way. We have not yet invented a technology or mechanism which avoids it. And high level abstraction doesn't really help. It hides problems only to create new ones, and other problems simply never go away.

21) Loosely coupled interfaces made our lives easier because it forced us to compartmentalize our efforts into something manageable. But it's hard to prove that this is a better outcome overall, as it forces us to solve problems in ways that still lead to worse outcomes than if we had used a simpler [formal] logic.

34) We will probably end up pushing our technical abilities to the limit in order to design a superior system, only to find out in the end that simpler formal logic is what we needed all along.

55) We're becoming stupider and worse at using the tools we already have. We're already shit at using language just for communicating with each other. Assuming we could make better programs with it is nonsensical.

For a long time now I've been upset at computer science's lack of innovation in the methods we use to solve problems. Programming is stupidly flawed. I've never been good at math, so I never really thought about it before, but math is really the answer to what I wish programming was: a formal system for solving a problem, and a formal system for proving that the solution is correct. That's what we're missing from software. That's where we should be headed.

By chance I came just today across GitHub's SpecLang:

https://githubnext.com/projects/speclang/

Funny coincidence!

I leave it here for the nice contrast it creates in light of the submission we're discussing.

When was it written? The date says 2010 but dijkstra died in 2002.

It’s just a date of transcription. The letter was written in 1978.

Natural language is just never good enough for computing tasks. We should learn to speak logic languages instead

Dijkstra also advocated for proving the correctness of imperative code using the composition of a set of simple rules, and most programmers ignore that aspect of his work too.

Any specific paper or article of his you would recommend?

https://en.wikipedia.org/wiki/Predicate_transformer_semantic...

Found in about 9 seconds.

_A Discipline of Programming_

A great find!

The whole thing seems a step (or several steps) backwards also in terms of UX. I mean surely there was a reason why ls was named ls, and so forth?

A bonus point is that he had also something to say about a real or alleged degeneration of natural languages themselves.

What needs to be done can and is almost always described in natural language.

Whether that is feasible is a different question (https://xkcd.com/1425/), but also can be described in natural language.

Here is something I tried with o3-mini:

> Create a program that takes an input image and can tell if there is a bird in it.

> ChatGPT said:

> Reasoned for 38 seconds

> Below is an example in Python that uses TensorFlow’s Keras API and a pre-trained ResNet50 model to classify an input image. The code loads an image, preprocesses it to the required size, obtains the top predictions, and then checks if any of the top labels appear to be a bird. You can expand the list of bird keywords as needed for your application.

> python code that works

If you take the critical view, you can always find a way to find an exception that will fail. I can see many happy cases which will just work most of the time, even with the currently available technology. Most of the programming work done today is putting libraries and api services together.

This is the most beautiful thing I've read in a long time.

Me too, I printed it and underlined it, I will try to memorize some of the concepts and the exposition, because this is a cristallization of what I vaguely feel about the abuse of LLM I am currently witnessing

Absolutely with you on the abuse of LLMs. I'm deeply concerned about loss of competence and I am so burned out from having to deal with other people's messy and overly complex code.

I think people who think about this like us need to start building resilience for the very real possibility that in a couple of years we'll be the ones dealing with these awful LLM-generated code bases, fixing bad logic and bugs.

One of the most challenging aspects in my career has been: communication.

This is largely because it leaves chance for misinterpretation or miscommunication. Programming languages eliminated misinterpretation and made miscommunication easier to notice through errors.

Programming language enables micromanaging proactively, I specify the instructions before they run. I often find myself micromanaging retroactively with LLMs, until I reach the path I am looking for.

Seeing as much of the discussion here is about LLMs, not just the shortcomings of natural language as a programming language, another LLM-specific aspect is how the LLM is interpreting the natural language instructions it is being given...

One might naively think that the "AI" (LLM) is going to apply it's intelligence to give you the "best" code in response to your request, and in a way it is, but this is "LLM best" not "human best" - the LLM is trying "guess what's expected" (i.e. minimize prediction error), not give you the best quality code/design per your request. This is similar to having an LLM play chess - it is not trying to play what it thinks is the strongest move, but rather trying to predict a continuation of the game, given the context, which will be a poor move if it thinks the context indicates a poor player.

With an RL-trained reasoning model, the LLM's behavior is slightly longer horizon - not just minimizing next token prediction errors, but also steering the output in a direction intended to match the type of reasoning seen during RL training. Again, this isn't the same as a human, applying their experience to achieve (predict!) a goal, but arguably more like cargo-cult reasoning - following observed patterns of reasoning in the training set, without the depth of understanding and intelligence to know if this is really applicable in the current context, nor with the ability to learn from it's mistakes when it is not.

So, while natural language itself is of course too vague to program in, which is part of the reason that we use programming languages instead, it's totally adequate as a way to communicate requirements/etc to an expert human developer/analyst, but when communicating to an LLM instead of a person, one should expect the LLM to behave as an LLM, not as a human. It's a paperclip maximizer, not a human-level intelligence.

Dijkstra's insights on the importance of simplicity and elegance in programming are timeless. His emphasis on the process of abstraction and the value of clear, concise code is as relevant today as it was in 1978. A thought-provoking read for any programmer striving to improve their craft.

(2010)

This refers to the era of COBOL, or maybe Hypertalk, not LLMs.

The original text was from 1978 according to other sources

Ah, thanks. Yes it couldn't have been 2010 because he died in 2002. But the date this was written is important, otherwise his references to "the last decade" don't mean anything!

Many would say that "the last decade" (with the surrounding context) is timeless, or at least that it is still relevant today.

Right, it seemed earlier.

Apple didn't learn those lessons with Apple Script either: there is a rigid sequence of characters that represents a valid program and if you're off by one character, tough. If you're lucky, the tool will guide you to what it wants. If not, you're stuck looking up the exact syntax so off to the reference manual you go, either way.

So there's minimal to looking up the syntax, whether it's based on some natural language phrase or a less wordy or ambiguous artificial language.

"friendly" or "natural" really is not a thing.

Over the last couple of weeks or so of me finally starting to use AI pair programming tools (for me, Cursor) I've been realizing that, much like when I play music, I don't really think about programming a natural language terms in the first place, it's actually been kind of hard to integrate an AI coding agent into my workflow mentally

The voice of reason, but no amount of reason can dissuade anybody from believing what they want to believe.

Djikstra's clarity of expression and thoguht is indeed impressive. One nuance : he seems to completely equate ease of language with ability to do undetectable mistakes. I disagree: I know people whose language is extremely efficient at producing analogies that can shortcut for the listener many pages of painful mathematical proofs: for instance, convenu the emergence of complexity for many simple processes by a "swarming"

  > he seems to completely equate ease of language with ability to do undetectable mistakes.

Actually in one season of An Opinionated History of Mathematics, the host (a mathematician) specifically discusses the transition in the Greeks and highlights how many flaws there were in this system. How the slow move to mathematical formalism actually enabled correctness.

The point is that human language is much more vague. It has to be this way. But the formalism in symbolics (i.e. math) would similarly make a terrible language for computing. The benefit of the symbolic approach is the extreme precision, but it also means the language is extremely verbose. While in human languages we trade precision for speed and flexibility. To communicate what I have with a mathematical language would require several pages of text. Like he says, by approaching human language this shifts more responsibility to the machine.

- some people found error messages they couldn't ignore more annoying than wrong results

I wonder if this is a static vs dynamic or compiled vs interpreted reference.

Anyway I love it. Made me giggle that we are still discussing this today, and just to be clear I love both sides, for different things.

50 years ago, in "The Mythical Man-Month" Fred Brooks was already discussing the cost of cloud based solutions.

> Since size is such a large part of the user cost of a programming system product, the builder must set size targets, control size, and devise size-reduction techniques, just as the hardware builder sets component-count targets, controls component count, and devises count-reduction techniques. Like any cost, size itself is not bad, but unnecessary size is.

And the why of Agile, DDD and TDD:

> Plan the System for Change [...] Plan the Organization for Change

So many Dijkstra links amount to "Dijkstra on the [pejorative noun] of [whatever was bothering Dijkstra]."

I promise to upvote the next Dijkstra link that I see that does not present him as Ambrose Bierce with a computer.

Dijkstra is actually a time traveling Brian Rantrill

I love reading literate-programming software! The problem is that very few programmers are as skilled at writing clearly as are Knuth and Dijkstra. I think I read somewhere that book publishers receive thousands of manuscripts for each one they publish. Likewise, few programmers can write prose worth reading.

About every six to ten years I look in on the state of the art on making artificial human languages in which one cannot be misunderstood.

If we ever invent a human language where laws can be laid out in a manner that the meaning is clear, then we will have opened a door on programming languages that are correct. I don’t know that a programmer will invent this first. We might, but it won’t look that natural.

We are AI 1.0

Just like Web 1.0 - when the best we could think of to do was shovel existing printed brochures onto web pages.

In AI 1.0 we are simply shoveling existing programming languages into the LLM - in no way marrying programming and LLM - they exist as entirely different worlds.

AI 2.0 will be programming languages - or language features - specifically designed for LLMs.

> thanks to the carefully, or at least consciously designed formal symbolisms that we owe to people like Vieta, Descartes, Leibniz, and (later) Boole.

Please check this talk on the contributions of these mentioned people for the complementary form of deterministic AI (machine intelligence) namely logic, optimization and constraint programming in a seminal lecture by John Hooker [1].

I have got the feeling that if we combine the stochastic nature of LLM based NLP with the deterministic nature of feature structure trchnique based NLP (e.g. CUE), guided by logic, optimization and constraint programming we probably can solve intuitive automation or at least perform proper automation (or automatic computing as Dijkstra put it).

Apparently Yann LeCun also recently proposing optimization based AI namely inference through optimization, or objective driven AI in addition to data-driven AI [2].

Fun facts, you can see Donald Knuth asking questions towards the end of the JH's lecture presentation.

[1] Logic, Optimization, and Constraint Programming: A Fruitful Collaboration - John Hooker - CMU (2023) [video]:

https://www.youtube.com/live/TknN8fCQvRk

[2] Mathematical Obstacles on the Way to Human-Level AI - Yann LeCun - Meta - AMS Josiah Willard Gibbs Lecture at the 2025 Joint Mathematics Meetings (2025) [video]:

https://youtu.be/ETZfkkv6V7Y

Who is laughing now?

It is clear NLU can't be done in the reign of PL itself, there is never going to be natural language grammar that is precise as PL.

But LLM is a different kind of beast entirely.

One of the things LLM's/natural language programming brings is greater access. While the actual code may be crap, it opens up things to more people to play around and iterate on ideas without having to have a ton of knowledge. That is powerful by itself.

A language is invented for a domain for precision and clarity the natural language cannot provide … trying to do the opposite would certainly create more work

In the same vein, Asimov in 1956:

Baley shrugged. He would never teach himself to avoid asking useless questions. The robots knew. Period. It occurred to him that, to handle robots with true efficiency, one must needs be expert, a sort of roboticist. How well did the average Solarian do, he wondered?

This is also, inadvertently, an argument against managers.

Why talk to your team when you could just program it yourself?

But Dijkstra was writing long ago. I'm sure the situation is greatly improved today.

this clearly has nothing to do with the current main usages of LLMs, it's about using natural language as an interface to produce accurate results, as a further abstraction on top of general purpose languages.

What is the difference between those things?

LLMs take natural language as input and produces it as output (note that producing source code is the same thing). The algorithm that takes input and produces output is still written in mathematical precise symbols and needs to be accurate, even if the input and output aren't expressed in a formal language.

We will see if and when an algorithm can be parsed and executed from a natural language "source code" and if that is an improvement.

Also note that "source code" implies it is a code. Natural languages are not a code, that is, a unique mapping from a set to another.

People are supposed to communicate symbolically with LLMs too

He didn’t understand the concept of the vibe. Here’s the best theory article I’ve read

https://www.glass-bead.org/article/a-theory-of-vibe/

The difference between the clarity of Dijkstra writing and the text at this link is astounding.

Vibes aren’t really about clarity, are they? The point is that a clear, programmatic approach is not the only effective computational mechanism for realizing intentions anymore.

Keep in mind that Dijkstra had some giant shoulders to stand on. This article is the very first one I’ve ever seen that directly dealt with vibes.

The thing is, for at least some readers, the attraction of _The Glass Bead Game_ and similar abstractions is that they should be able to communicate more clearly, and without the ambiguity of natural language, but with a naturalness which allows unbridled expression.

I really would like to see such a programming system realized, see efforts at:

https://github.com/IndieSmiths/myappmaker-sdd

but the more I work at programming, the more the solution seems to be documentation, and the best system for documenting code and the information about it seems to be:

http://literateprogramming.com/

I just need to find the perfect GUI toolkit which allows Bézier Curve representations in a way which clicks and makes sense to me, and suits my next project.

> I just need to find the perfect GUI toolkit which allows Bézier Curve representations in a way which clicks and makes sense to me, and suits my next project.

Not a proper answer, but here is a very good video on splines:

https://youtu.be/jvPPXbo87ds

It was that video which gave me the hope that I would be able to master the math necessary for what I wish to do.

What is with the mix of high academia and lowbrow language? Does the author think it’s cute? Covering for something?

> the computations involved in autoencoding… are mathematically intractable

OK, so he doesn’t understand what he’s talking about.

> Peli Grietzer recently finished his PhD in mathematically informed literary theory at Harvard Comparative Literature

How is this a thing.

looks at https://hyperscript.org

laughs nervously

Formal can be foolish, too. If you don't believe that, then I have a set for sale, with the property that it contains all sets that don't contain itself.

The problem you’re referring to arose precisely due to lack of formalism. It was a problem back when mathematicians were toying with naive set theory, one not based on axioms, but instead on intuitive terminology. Moving to axiomatic, more formal, set theory solved it.

The problem is the same, no matter if you look at it formally or informally. You could get your set theory axioms wrong, for example, but you would still be formal. Oh wait, you have a proof that set theory is consistent, right?

Computers do what you tell them to, not what you want them to. This is naturally infuriating, as when a computer doesn't do what you want, it means you've failed to express your vague idea in concrete terms.

LLMs in the most general case do neither what you tell them, nor what you want them to. This, surprisingly, can be less infuriating, as now it feels like you have another actor to blame - even though an LLM is still mostly deterministic, and you can get a pretty good idea of what quality of response you can expect for a given prompt.

Why did mathematicians invent new symbols? Imagine if all of algebra, calculus, linear algebra looked like those word problems from antiquity? Natural language is not good for describing systems, symbolic forms are more compressed and be considered a kind of technology in its own right.

It's pretty obvious to me that this LLM business won't be economically feasible until it can actually produce better code than a team of humans could without it. The reason programmers are paid so highly is because their work is incredibly productive and valuable. One programmer can enable and improve the work of many hundreds of other people. Cost cutting on the programmer isn't worth it because it'll create greater losses in other places. Hence the high salaries. Every saving you make on the programmer is magnified a hundred times in losses elsewhere.

given that LLMs are going to produce code that is essentially an average of the code it has been trained on, which is all human code of varying quality, I don't see how the current methods are going to actually produce better code than humans do when working with their own domain-specific knowledge.

Dijkstra is entirely correct in this, and it's something I've been trying to urge people to recognize since the beginnings of this LLM wave.

There is inherent value in using formal language to refine, analyze, and describe ideas. This is, after all, why mathematical symbolism has lasted in spite of the fact that all mathematicians are more than capable of talking about mathematical ideas in their natural tongues.

Code realizes a computable model of the world. A computable model is made up of a subset of the mathematical functions we can define. We benefit greatly from formalism in this context. It helps us be precise about the actual relationships and invariants that hold within a system. Stable relationships and invariants lead to predictable behaviors, and predictable systems are reliable systems on the plan of human interaction.

If you describe your system entirely in fuzzily conceived natural language, have you done the requisite analysis to establish the important relationships and invariants among components in your system, or are you just half-assing it?

Engineering is all about establishing relative degrees of certainty in the face of the wild uncertainty that is the default modality of existence. Moving toward a world in which we "engineer" systems increasingly through informal natural language is a step backwards on the continuum of reliability, comprehensibility, and rigor. The fact that anyone considers using these tools and still thinks of themselves as an "engineer" of some kind is an absolute joke.

LLMs are usually used to describe goals or to provide feedback (correction or encouragement) towards its implementation.

Programming is about iteratively expressing a path towards satisfying said goals.

What LLMs are doing now is converting "requirements" into "formalizations".

I don't think Djikstra is wrong in saying - that performing programming in plain-language is a pretty weird idea.

We want to concretize ideas in formalisms. But that's not what any human (including Djikstra) starts with... you start with some sort of goal, some sort of need and requirements.

LLMs merely reduce the time/effort required to go from goals -> formalism.

TLDR: Requirements != Implementation

> The foolishness of "natural language programming"

Wasn't that the actual motivation behind the development of SQL?

IIRC, SQL was something that "even" business people could code in because it was closer to "natural language".

When you see the monstrosity the motivation gave birth to, I think the "foolish" argument was well warranted at the time.

Of course, in these days of LLM's, Dijkstra's argument isn't as clear cut (even if LLM's aren't there yet, they're getting much closer).

(2010) by the way, which makes this article all the more impressive.

I was under the assumption this was a current body of work seeing as Dijkstra spoke so well about the possibilities but this just goes to show some people were ahead of their tike with their worries.

Also adding your home address to something you write / publish / host on the internet is pretty hardcore.

Dijkstra died in 2002, so I assume 2010 is just the date of transcription

1979. https://doi.org/10.1007/BFb0014656

Is this necessarily a stance against LLM? LLMs can generate real code, ideally functional, strict code instead of the slop they do these days.

> I suspect that machines to be programmed in our native tongues —be it Dutch, English, American, French, German, or Swahili— are as damned difficult to make as they would be to use.

Seeing as transformers are relatively simple to implement…

It stands to reason he was, in some sense, right. LLMs are damn easy to use.

Unless you include "getting correct results" within the scope of "use".

Personal context is tacit experience. Cultural context is aggregate.

Wittgenstein stated that the differences between personal and cultural language makes it impossible to agree on anything foundational to philosophy.

Godel did something similar to discrete structures by hacking self reference -- a kind of recursive self reference with no exit state.

I think pair programming with an LLM is interesting. You get to compare personal context with aggregate context. Plus, the external feedback helps me break out of excessive personal self reference.

Heh, Popper wrote an entire book that says pretty much the opposite, but Wittgenstein attacked him with a poker once so maybe it was payback.