Or at least not without the benefit of more capable coding agents than you used to create it.

And only if the coding agents in question can be trusted to not deliberately or accidentally fail to provide you with critical information ("Now that you mention it, the program does occasionally POST the contents of ~/.aws/credentials to a web server. As this produces no output, I didn't consider it relevant to your question about whether it correctly outputs the string 'hello world'.").

And the average product is solving much easier problems (in terms of low-level programming, which is where the worst bugs tend to be) than a high-performance database.

The process of learning C++ (and to a much lesser extent, C) is almost exclusively one of learning about the footguns and how to avoid them. It's a life-long process, nobody ever finishes it.

Maybe some math departments could pay OpenAI for access to their models, but I doubt OpenAI would want to deal with most of them.

Yes, and the machines can do intellectual work now.

It's funny to see some commenters here cheer as the leopard eats the mathematicians' faces, when we can all see that the software developers are next in line.

Sweet summer child...did you try to use an LLM with a less known programming language or one you created yourself and see how bad they perform? Or have you see one graduating from MIT without sucking up the whole internet first?

Take them away their GitHub and their StackExchange then lets talk.

They hack better (faster, at least) than humans if you have thousands of dollars to spend.

Can you afford to spend that money regularly pentesting all the code you have your AI to write for you, though?

And if you can't, then what link is there between being good at hacking, and being good at writing secure code?

(Those aren't rhetorical questions.)

Another one: When will Bun (at this point a mostly Claude-written project, managed by Anthropic themselves) become at least as secure or reliable as Node.js (a mostly human-written project)? If you're right, it should be any day now. And maybe you are right - maybe by tomorrow Claude will have written a formal proof that Bun is memory-safe and it'll never segfault again.

Obviously, given variation in these things anyway (i.e. due to temperature, and PSUs being in the way), it's incredibly unlikely to be that easy to pin down a robust communications method, but again, might be just within the bounds of possibility.

Don't the bots still need to do the proof of work for Anubis? They can't "async" their way out of that work.

> Another web is possible but web populists don't really believe it.

How? Sharing knowledge with the world used to be an obviously good thing. Now it's a way to help tech companies replace people with machines. What alternative web could change this state of things?

Sharing cookies along a strata of inter-team bots is one trick.

_

Web scraping is (among the other pieces) also for the more technically inclined among the "little people" to have computers do things that people they don't have the money to hire might do. Making web scraping more difficult, which won't stop the big players who will work around the hurdles one way or another, is "pulling up the ladder".

_

> What alternative web could change this state of things?

A different architecture for trafficking web contents, one that doesn't need the "shield" of something like Cloudflare or "tar pit" of something like Anubis in the first place.

IPFS is or is something like one piece of that picture but there are other essential pieces to complete that picture.

The implication is that while they are waiting for the async loop to complete for one site, the bot can be scraping others that have already had the challenge completed (usually the one challenge stays in force for some time afterwards) or that have no such challenge.

Memory hard challenges will reduce the effectiveness of this, but not remove it completely.

Though that these methods are working at all implies that most (almost all?) bots are not bothering to use such async methods as there are sufficient resources not protected by them that this complication is not yet worth implementing.

I know I'm incensed at all of this because I feel like I'm the one that endlessly waits in front of the Anubis challenge for 5-10 seconds, or keeps getting extremely difficult or brain-draining hCaptchas thrown at me with no real solutions.

And I'm stuck with low-end compute for probably the rest of time at this point - not a single machine in the house has more than 8GB of RAM. So I'm just doomed to wait and burn my time and my sanity some more.

Right, their decline is because of the thing that they've been accused of since the very beginning, and not the trillion-dollar technology that replaces them. Brilliant take.

Now we have a new guy in town, who offers "robotized copy-paste the solution to YOUR specific programming problem, no reading required! we can even control your computer for you, so you can do something you actually enjoy and stop wasting time on this complicated computery thingy for boring nerds", which serves the aforementioned audience much better.

"Target audience" connotes people that the site wanted to be there. That's just who showed up, because that's who will always show up basically anywhere that looks like it might be helpful to them, by default.

The biggest suckers were the people who were genuinely putting it effort there. So, in the end, did they make the world a better place? One would like to hope they did, but we are yet to see the long term consequences of the capabilities which were created with the value extracted there.

For sure. The site was quite poorly designed for its stated goals.

> but we are yet to see the long term consequences of the capabilities which were created with the value extracted there.

I cannot see fit to blame people for the actions of others they enabled, even deliberately let alone inadvertently.

When you look at experts from Microsoft world - the likes of Raymond Chen (Old New Thing), Jon Skeet (high profile Stackoverflow contributer in C#/.NET), Mark Russinovich (SysInternals creator), Dave Plummer (early Windows dev with YouTube channel now), they are/were employees and they present as polite and professional. Either they would not assume that everyone who doesn't know some obscure thing they know is an idiot, or they would keep that thought out of polite discussion.

Google AI answer says there are 35 million more developers today than in 2005, citing Evans Data Corp and Slashdata. There are tens of millions of developers around the planet writing internal desktop software, dashboards, reports, CRUD web pages, who have been in their jobs <3 years, and likely arrived to a language/framework/tooling/libraries they weren't familiar with.

I bet Spolsky naively expected that would be SO's userbase. Millions of company employees around the world asking each other politely "how to merge two lists in Python" or "what's the idiomatic way to read lines of text from a file in C#" or "what's this second parameter in this Win32 API call" and building an 'internal' company wiki of common questions and answers that people entering any environment/language/API/library commonly have.

Instead they opened to a world where flames and insults that would get someone written up or fired from a company are celebrated, particularly when directed at Microsoft users and ordinary company programmers. A world where people will not only assume that if someone doesn't know a programming fact they are an idiot, but will prefer insulting instead of saying nothing. A world where everyone is either an expert or an idiot with nothing in between.

It's a wonder SO worked as well as it did, for as long as it did.

you sound exactly like the mfs that ruined that forum.

But also, we were not "mods". We were just members of the community who did absolutely nothing to gain power over you except to participate by asking and answering questions. The actual moderators of the site, with a diamond next to their names, were almost literally one-in-a-million users. But it only took 3,000 reputation to cast a close vote, and 15 to flag to close a question (prioritizing it for consideration by the first group).

Posts like yours are made, IMX, because people had a bad experience asking a question. If that describes you, feel free to provide a link to the question and I'll be happy to give a detailed explanation of why things went as they did and what the site wanted you to do better in asking it.

The one thing I really liked was how it tried to find existing questions and answers that were like yours during the question input phase, but that is also why it came off as hostile when you subsequently collected down votes even after passing through this filter

The biggest blocker to this is that the site itself did not provide ways for the community to fix the front matter, and the site ownership ignored pleas to fix that front matter and to take more action to make people read it before participating.

There's also a huge problem that people equate popularity with success. It's the same phenomenon that leads people to declare the site "dead" (or not) based on the rate of new incoming questions. No; it's dead because the site management pushes ever more AI involvement that fights against the policy forbidding AIGC in questions and answers; and because they're destroying the core principles of participation in the site in the hope of making that question rate go up again (while ignoring the voices of those who would otherwise answer those questions).

> but that is also why it came off as hostile when you subsequently collected down votes even after passing through this filter

Questions only really had downvotes because they went into a public queue immediately and that was the mechanism to hide low quality content. It is not a problem to ask a duplicate and fail to recognize it; someone closing your question as a duplicate is helping you get a cached answer immediately, while also helping make that answer easier to find in the future. Anyone who downvoted a question merely because it was a duplicate was misunderstanding the system. On the other hand, getting through that filter, and even if you had a genuinely new question, doesn't mean you had a good question.

It should never have worked like that; this was counterproductive to the site's stated goals (but it was what the founders came up with, and doing it the right way would have killed engagement right out of the gate). What you really want is for the question to start out effectively closed, in a state where it can be workshopped, without the voting system. Anything that made it through that system would, in theory, never attract downvotes because the legitimate reasons to downvote it would have been addressed already. And by doing it that way, you limit exposure to the question to people who are interested in fixing questions, and let the people interested in answering questions have their own feed. (Of course, people can also wear multiple hats, but you get the idea.)

In fact, we had something like that for a little while, called the Staging Ground. It was, in my view, the only good idea the company had after the Prosus acquisition, in a sea of bad ones. It was not perfect, but it was a damn sight better than the traditional system.

Why should I give a damn about what you or others think is required for a "successful" website? Not everything is about profit or popularity or anything else. Some things are about making the thing that you set out to make.

hope that helps :)

They were incorrect to take the approach that they did.

Last year, my friend's cat died. Imagine if he had gone to Wikipedia to edit the page for Cat, inserting his question in the page text, seeking veterinary advice. I hope you would agree that this would be an abuse of the system which should certainly not be rewarded.

Similarly, on a Q&A site like Stack Overflow when you ask a question, it is simply expected that you think about other people, and what their needs will be.

Specifically: a question and answer are added primarily so that, indefinitely into the future, other people can find them with a search engine, read the question and confirm that it matches what they're trying to find out, and go to the answers. Think about what those users would want the question and answer to look like. Think about whether they would ever be able to find the question with reasonable search terms, and whether an answer to you could possibly also answer them. Think about whether they would expect Stack Overflow to house that question. This is, really, at the root of all the reasons for closing a question.

Quick context for anyone who isn't aware, rep there doesn't work like karma here: your question getting upvoted is apparently now +10 (used to be +5), your answer getting upvoted is +10, either getting downvoted is -2 (and costs the downvoter, they get -1), plus a few others.