Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Scaling long-running autonomous coding

simonwillison.net

Post reply on HN
Print view View original

101–110 of 112 posts

Re: Scaling long-running autonomous coding

#101

Post by theshrike79 »

Chipshuffle wrote: post #65
NiloCK wrote: post #53

Earlier quoted context omitted.

If you find yourself 50-first-dating your LLMs, it may be worth it to invest some energy into building up some better context indexing of both the codebase itself and of your roadmap.

Yeah, I admit I'm probably not doing that quite optimally. I'm still just letting the LLM generate ephemeral .md files that I delete after a certain task is done. The other day I found [beads]( https://github.com/steveyegge/beads ) and thought maybe that could be a good improvement over my current state. But I'm quite hesitant because I also have seen these AGENTS.md files become stale and then there is also the ques…

Beads is basically what github issues is, but local and built in a way that LLMs can easily use it. I had a self-made solution that was close, but moved to beads because it worked out of the box without disrupting my workflow that much.

I've used it quite a bit, but now that Gas Town is a thing Beads getting a bit bloated and they're adding new features left and right, dunno why.

Might have to steal the best bits of Beads (the averaged out cli experience and JSONL for storing issues in the repo + local sqlite cache) and build my one with none of the extra bells and whistles.

Re: Scaling long-running autonomous coding

#102

Post by theshrike79 »

ramraj07 wrote: post #16
halfcat wrote: post #10

So AI makes it cheaper to remix anything already-seen, or anything with a stable pattern, if you’re willing to throw enough resources at it. AI makes it cheap (eventually almost free) to traverse the already-discovered and reach the edge of uncharted territory. If we think of a sphere, where we start at the center, and the surface is the edge of uncharted territory, then AI lets you move instantly to the surface. If…

The fundamental idea that modern LLMs can only ever remix, even if its technically true (doubt), in my opinion only says to me that all knowledge is only ever a remix, perhaps even mathematically so. Anyone who still keeps implying these are statistical parrots or whatever is just going to regret these decisions in the future.

There are musicians who "remix" (sample) other artists music and make massive hits themselves.

Not every solution needs to be unique, in many cases "remixing" existing solutions in an unique way is better and faster.

Re: Scaling long-running autonomous coding

#103

Post by polyglotfacto »

sealeck wrote: post #4
simonw wrote: post #3

One of the big open questions for me right now concerns how library dependencies are used. Most of the big ones are things like skia, harfbuzz, wgpu - all totally reasonable IMO. The two that stand out for me as more notable are html5ever for parsing HTML and taffy for handling CSS grids and flexbox - that's vendored with an explanation of some minor changes here: https://github.com/wilsonzlin/fastrender/blob/19bf103…

I think the other question is how far away this is from a "working" browser. It isn't impossible to render a meaningful subset of HTML (especially when you use external libraries to handle a lot of this). The real difficulty is doing this (a) quickly, (b) correctly and (c) securely. All of those are very hard problems, and also quite tricky to verify. I think this kind of approach is interesting, but it's a bit sad t…

I think the current approach is simply not scalable to a working browser ever.

To leverage AI to build a working browser you would imo need the following:

- A team of humans with some good ideas on how to improve on existing web engines.

- A clear architectural story written not by agents but by humans. Architecture does not mean high-level diagrams only. At each level of abstraction, you need humans to decide what makes sense and only use the agent to bang out slight variations.

- A modular and human-overseen agentic loop approach: one agent can keep running to try to fix a specific CSS feature(like grid), with a human expert reviewing the work at some interval(not sure how fine-grained it should be). This is actually very similar to running an open-source project: you have code owners and a modular review process, not just an army of contributor committing whatever they want. And a "judge agent" is not the same thing as a human code owner as reviewer.

Example on how not to do it: https://github.com/wilsonzlin/fastrender/blob/19bf1036105d4e...

This rendering loop architecture makes zero sense, and it does not implement web standards.

> in the HTML Standard, requestAnimationFrame is part of the frame rendering steps (“update the rendering”), which occur after running a task and performing a microtask checkpoint

> requestAnimationFrame callbacks run on the frame schedule, not as normal tasks.

This is BS: "update the rendering" is specified as just another task, which means it needs to be followed by a microtask checkpoint. See https://html.spec.whatwg.org/multipage/#event-loop-processin...

Following the spec doesn't mean you cannot optimize rendering tasks in some way vs other tasks in your implementation, but the above is not that, it's classic AI bs.

Understanding Web standards and translating them into an implementation requires human judgement.

Don't use an agent to draft your architecture; an expert in web standards with a interest in agentic coding is what is required.

Message to Cursor CEO: next time, instead of lighting up those millions on fire, reach out to me first: https://github.com/gterzian

Re: Scaling long-running autonomous coding

#104

Post by polyglotfacto »

simianwords wrote: post #97
tabs_or_spaces wrote: post #93

> I think somebody will have built a full web browser mostly using AI assistance, and it won’t even be surprising > When I made my 2029 prediction this is more-or-less the quality of result I had in mind. There seems to be a lot of compensation and leniency made by the author here. So, it is seemingly impressive that someone was able to use agents to build a browser. But they used trillions of tokens? This equates to…

>But they used trillions of tokens? This equates to millions of dollars of spend. Are we really happy with this? Yes, arguably 5 million is a fair price and cheaper than what it would take to pay humans.

If you paid 5 cents for the code you would have been ripped off; it's throw-away stuff.

Re: Scaling long-running autonomous coding

#105

Post by polyglotfacto »

light_hue_1 wrote: post #31

Browsers are pretty much the best case scenario for autonomous coding agents. A totally unique situation that mostly doesn't occur in the real world. At a minimum: 1. You've got an incredibly clearly defined problem at the high level. 2. Extremely thorough tests for every part that build up in complexity. 3. Libraries, APIs, and tooling that are all compatible with one another because all of these technologies are bu…

So first of all, as per my other comments on this threads and coming from a browser engineer: the autonomous coding agents failed miserably.

Whether it is the best case scenario in terms of benchmark, I am not so sure.

The Web is indeed standardized and there are many open-source implementations out there. But how to implement the Web in a novel way by definition means you are trying to solve some perceived problem with existing implementations.

So I would rephrase your statement as such: rewriting an existing engine in another language without any novelty might be the best case scenario for autonomous coding agents.

As an example of approaching the problem in a novel way: the Fastrender code seems obsessed with metering of resources. Implementing the Web with that constraint in mind would be an interesting problem and not obvious at all. That's not what the project is doing so far by the way, since the code is quite frankly a bunch of spaghetti that does not follow Web standards at all(in a way that is unrelated to the metering story, so the divergence from specs is not novel, it's just wrong).

Re: Scaling long-running autonomous coding

#106

Post by m4nu3l »

simianwords wrote: post #97
tabs_or_spaces wrote: post #93

> I think somebody will have built a full web browser mostly using AI assistance, and it won’t even be surprising > When I made my 2029 prediction this is more-or-less the quality of result I had in mind. There seems to be a lot of compensation and leniency made by the author here. So, it is seemingly impressive that someone was able to use agents to build a browser. But they used trillions of tokens? This equates to…

>But they used trillions of tokens? This equates to millions of dollars of spend. Are we really happy with this? Yes, arguably 5 million is a fair price and cheaper than what it would take to pay humans.

There is a problem with this comparison. The agent had access to open-source browsers in its training set. So you'd need to compare the cost of creating an equivalent browser for a developer who has access to them, too. If all you need is standard browser functionality, you just use an existing browser. If you want to change some features or parts of the implementation, you fork it. A new browser written from scratch would be valuable if it had a novel implementation that resulted in a faster/more secure/robust/memory efficient or simply easier-to-use browser. So even if this had implemented the standard correctly, it wouldn't be worth more than the time it takes a developer to fork Chromium and change its name. Don't get me wrong, it's impressive, but not as impressive after you think that an LLM that regurgitates verbatim the code of Chromium when tasked to build a browser would have effectively succeeded at the task.

EDIT: About the rendering speed. It doesn't really make sense to compare it with a fully functioning browser, as you could potentially drop features or make bogus optimisations to go faster.

Re: Scaling long-running autonomous coding

#107

Post by amenhotep »

encyclopedism wrote: post #59
Zababa wrote: post #56

Earlier quoted context omitted.

Can you give examples of how that "LLM's do not think, understand, reason, reflect, comprehend and they never shall" or that "completely mechanical process" helps you understand better when LLM works and when they don't? Many people are throwing around that they don't "think", that they aren't "conscious", that they don't "reason", but I don't see those people sharing interesting heuristics to use LLMs well. The "the…

We know what an LLM is in fact you can build one from scratch if you like. e.g. https://www.manning.com/books/build-a-large-language-model-f... It's an algorithm and a completely mechanical process which you can quite literally copy time and time again. Unless of course you think 'physical' computers have magical powers that a pen and paper Turing machine doesn't? > Many people are throwing around that they don't "th…

A cursory read of basic philosophy would surely include the arguments against Searle's Chinese room, no? It's hardly settled.

Re: Scaling long-running autonomous coding

#108

Post by ontouchstart »

polyglotfacto wrote: post #103
sealeck wrote: post #4

Earlier quoted context omitted.

I think the other question is how far away this is from a "working" browser. It isn't impossible to render a meaningful subset of HTML (especially when you use external libraries to handle a lot of this). The real difficulty is doing this (a) quickly, (b) correctly and (c) securely. All of those are very hard problems, and also quite tricky to verify. I think this kind of approach is interesting, but it's a bit sad t…

I think the current approach is simply not scalable to a working browser ever. To leverage AI to build a working browser you would imo need the following: - A team of humans with some good ideas on how to improve on existing web engines. - A clear architectural story written not by agents but by humans. Architecture does not mean high-level diagrams only. At each level of abstraction, you need humans to decide what m…

How much effort would it take GenAI to write a browser/engine from scratch for GenAI to consume (and generate) all the web artifacts generated by human and GenAI? (This only needs to work in headless CI.)

How much effort would it take for a group of humans to do it?

Re: Scaling long-running autonomous coding

#109

Post by polyglotfacto »

ontouchstart wrote: post #108
polyglotfacto wrote: post #103

Earlier quoted context omitted.

I think the current approach is simply not scalable to a working browser ever. To leverage AI to build a working browser you would imo need the following: - A team of humans with some good ideas on how to improve on existing web engines. - A clear architectural story written not by agents but by humans. Architecture does not mean high-level diagrams only. At each level of abstraction, you need humans to decide what m…

How much effort would it take GenAI to write a browser/engine from scratch for GenAI to consume (and generate) all the web artifacts generated by human and GenAI? (This only needs to work in headless CI.) How much effort would it take for a group of humans to do it?

I'm not sure about what you mean with your first sentence in terms of product.

But in general, my guess at an answer(supported by the results of the experiment discussed on this thread), is that:

- GenAi left unsupervised cannot write a browser/engine, or any other complex software. What you end-up with is just chaos.

- A group of humans using GenAi and supervising it's output could write such an engine(or any other complex software), and in theory be more productive than a group of humans not using GenAi: the humans could focus on the conceptual bottlenecks, and the Ai could bang-out the features that require only the translation of already established architectural patterns.

When I write conceptual bottlenecks I don't mean standing in front of a whiteboard full of diagrams. What I mean is any work the gives proper meaning and functionality to the code: it can be at the level of an individual function, or the project as a whole. It can also be outside of the code itself, such as when you describe the desired behavior of (some part of) a program in TLA+.

For an example, see: https://medium.com/@polyglot_factotum/on-writing-with-ai-87c...

Re: Scaling long-running autonomous coding

#110

Post by ontouchstart »

polyglotfacto wrote: post #109
ontouchstart wrote: post #108

Earlier quoted context omitted.

How much effort would it take GenAI to write a browser/engine from scratch for GenAI to consume (and generate) all the web artifacts generated by human and GenAI? (This only needs to work in headless CI.) How much effort would it take for a group of humans to do it?

I'm not sure about what you mean with your first sentence in terms of product. But in general, my guess at an answer(supported by the results of the experiment discussed on this thread), is that: - GenAi left unsupervised cannot write a browser/engine, or any other complex software. What you end-up with is just chaos. - A group of humans using GenAi and supervising it's output could write such an engine(or any other…

That is a wonderful write up.

“This is a clear indication that while the AI can write the code, it cannot design software”

To clarify what I mean by a product. If we want to design a browser system (engine + chrome) from scratch to optimize the human computer symbiosis (Licklider), what would be the best approach? Who should take the roles of making design decisions, implementation decisions, engineering decisions and supervision?

We can imagine a whole system with human out of the loop, that would be a huge unit test and integration test with no real application.

Then human can study it and learn from it.

Or the other way around, we had already made a huge mess of engineering beasts and machine will learn to fix our mess or make it worse by order of magnitude.

I don’t have an answer.

I used to be a big fan of TDD and now I am not, the testing system is a big mess by itself.

Post reply on HN