Hacker Newsnew | past | comments | ask | show | jobs | submit | eab-'s commentslogin

The fact that this firm makes such defective environments is certainly worthy of attention, and most likely a completely irreversible reputational loss; however, I found the framing in this article of 'therefore all the P(doom) stuff is a psyop, specifically in order to defend this company' to be completely unjustified and frankly a little insane?

Why is it insane? It makes sense that Anthropic would want an insulating layer to do their dirty work and absolve themselves of culpability for distorting the facts.

This reminds me of Once Upon A Time in Shaolin.


> courts have found it to be legal

≠ ethical


I don't know if this falls under good ethical behavior but the one time I needed to mass scrape sites was to build an affordable housing directory for the Bay Area. Whether or not a unit of housing was under rent control was only available through private entities or apartment hosting sites and it was scattered and hard to find.

In my opinion, a directory of subsidized housing should have been provided by local governments and not through a plethora of real estate websites.


they're spending like 9 figures on a single marketing agency - this doesn't surprise me whatsoever


Honestly I expect it's just annoyed devs getting annoyed about the ratelimits on plans and post-hoc justifying. Now that Codex has far more capacity and their slot machine makes better outcomes (note: I am a heavy LLM-assisted coder) they feel like they have to justify their felt animosity towards these companies


Codex and Claude don't have meaningfully different rate limits since Claude doubled theirs.

Maybe you can get more headless use out of Codex but that's not gonna last. Investors are drying up and these companies need to get to profitability.


is there examples of the sort of things that's been built with these systems? it often feels like complication and abstraction for its own sake


He's very transparent about his addiction to agentic coding. He's doing it for the giggles as far as I can tell. Trolling and being a personality. I don't mean that in a bad kind of assessment. He's said as much.


> his addiction to agentic coding

Why would it be an addiction versus obsession? Getting obsessed with something and building things around it, even if it's nonsense, is artistically genuine.


It could be either. The bigger critique that people have is that the entire thing is an obvious LARP. None of this is "built for enterprise" or deployed in common ops. This is a zero-risk, zero-judgement project where vanity PRs are encouraged so that another agent can take credit for the vanity fix.


> the entire thing is an obvious LARP

Eh, plenty of people play with Sims. Having them do something that feels even tangentially useful has an understandable appeal.


That's fine, I also have side projects that do absolutely nothing useful. Most people are taking issue with the deluded "it's all so obvious" rhetoric used in Yegge's writing.


> Most people are taking issue with the deluded "it's all so obvious" rhetoric used in Yegge's writing

Thanks. This is helpful.


I don’t know. I have coding so much I got into technology management a decade ago. But I can’t read myself away from personal projects, it’s addictive in a way.


yeah that's true. i used the word addiction because i remember reading him using that exact word. and including tokens and llms alongside the rise of gambling and such.

but i could be wrong.


Dolelite is the only thing I could find.


I used to have an emoji password for my Android phone, and had the exact same issue after a reset! It's an odd but pretty terrible failure mode for locking oneself out...


You say locking oneself out, but I decline to consider any situation where a password can be set but not later entered as one where the user bears even a modicum of fault.


I remember a website that silently removed everything but the first 8 characters from the "password" field upon registration but somehow didn't do the same on the login page. It took me several hours and several password resets to actually log in after registration, since for some reason the trimming happened client-side and only when typing the password manually (and I was pasting my password from a password manager).


In a similar vein, I remember encountering a site where the frontend enforced basic complexity requirements ala “use at least one number and one symbol” but the system would silently drop all non-alphanumerics when it saved (presumably in some kind of failed conversion on the way into the backend DB). So setting a password like “foo_bar4!” would become “foobar4” which was surprising. What blew my mind though was when I figured out the stripped password worked to log in, which was how I eventually figured out what was happening, escaped the reset flow, and generated a compliant password.


We're so far down this path the language around the problem is distorted. Ownership has been perverted and the only thing you control is the bill.


Do we even control the bill? You could buy a annual-sub-paid-monthly, be unable to cancel it because you're locked out of your account, and then get taken to collections when you terminate it on the payment side.


Relevant xkcd: https://xkcd.com/2700/


> We have proofs that are gigabytes (I believe even terabytes in some cases) in size, but we know they are correct because they check in Lean.

I'm not aware of any of these. There's some SAT-like results that were not verified in Lean at that sort of scale, but Lean proofs of individual problems are nowhere near that. For example, Mathlib (think a Lean4 math stdlib) is 6GB including compilation artifacts, and iirc <100MB text.


> Haha. How do you reconcile a proof with actual code?

Languages like Lean allow you to write programs and proofs under the same umbrella.


As if Lean does not allow to circumvent it's proof system (the "sorry" keyword).

Also, consider adding code to the bigger system, written in C++. How would you use Lean to prove correctness of your code as part of the bigger system?


I mean, it's somewhat moot, as even the formal hypothesis ("what is this proof proving") can be more complex than the code that implements it in nontrivial cases. So verifying that the proof is saying the thing that you actually want it to prove can be near impossible for non-experts, and that's just the hypothesis; I'm assuming the proof itself is fully AI-generated and not reviewed beyond running it through the checker.

And at least in backend engineering, for anything beyond low-level algorithms you almost always want some workarounds: for your customer service department, for engineering during incident response, for your VIP clients, etc. If you're relying on formal proof of some functionality, you've got to create all those allowances in your proof algorithm (and hypothesis) too. And additionally nobody has really come up with a platform for distributed proofs, durable proof keys (kinda), or how to deal with "proven" functionality changes over time.


I find it surprising that GPT managed to complete (with indeed a harness that doesn't seem very powerful!) and Claude still cannot.


Don’t they give got a minimap?

What are they actual differences?


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: