zer00eyz

It's going to get hammered into oblivion long before it gets to where it can end up on sp500. The election, and what happens after may see them delaying the IPO (if I was one to bet)

But it also sort of doesn't matter. Anthropic, not in sp500, ends up a fail then the "bubble" pops and the sp500 gets ugly.

> As I grow in my career, on a day to day basis I feel like I'm failing or doing things badly.

This is one thing that we should lament about AI generating all the code. You never get to look back at your own "hand crafted" work with the distance of time. Progress or perspective, it does not matter, one would probably find something they could have done better.

I think I'm actually going to miss that feeling.

All I could think of when I looked at that is "Oh this is gonna be how Nvidia's chart looks some day soon(ish)"

> worse experience

This too is subjective.

No one designs for users any more. It's all gas no breaks on the path to enshitification. A lot of this is the motivation of the people building software.

Raise your hand if you ever saw a PM get promoted for removing a bunch of bad and/or unused features?

> I think it's a fundamental rule that the competence and effectiveness of an org must fall, over time

IBM, GE... how many former American giants failed to evolve.

I go back to this article on the failure of US steel: https://www.construction-physics.com/p/no-inventions-no-inno...

LOL

My first program was in 2nd grade on an Apple IIe in basic. It was a program to check if a word was worth a dollar (a=1, b=2, c=3...)

It became a family activity (every one taking turns at the computer) every night to just type in words and see what they came out to. It was all very boring.

I was sure I was going to win.

I lost (well 2nd place), to the kid with 6 older brothers, who had that many years of building out the family list.

> For any application at scale it'd be way more accurate to have a dedicated classifier

"at scale" or the above use case of "specialized domain"

Jev being cheaper than an a full LLM call creates places where other tools might work better later - when the volume or the available data warrants it.

System optimal, code optimal and cost optimal aren't always aligned. Pick the right tool for the job - and dont get married to it. On occasion worse is better, and moreso when your testing or it is a one off/disposable.

Caring about something also means you don't want the issue to be misrepresented.

This is the most effective altruist anti AI headline you could get out of this story.

250k trees seems like a lot from you looking out your window. From the perspective of trees we cut down every year (that we account for) it is close to a rounding error. This is a place where they take forest management seriously - and thats the only reason any one even noticed or had to get a permit.

> Maybe Amazon shouldn't have built anti-Dalek defenses into their warehouses.

https://www.gatech.edu/news/2026/04/01/why-iran-targeted-ama...

https://arstechnica.com/gadgets/2026/09/iran-strikes-on-amaz...

A lack of anti drone defenses turned out to be an actual problem - also interesting use case for your idea of durability to be judged against.

Depends on what you want to do with it:

https://featherless.ai/blog/abliterated-models-for-cybersecu...

Is a decent primer, then off to open router you go.

Anthropic: Touch too much security stuff and we ban you, and keep your money.

Kimi: We took out the worst of it so have at it for free.

Everyone else: Back to my ablated model on openrouter for 1/10th the price where I can brute force issues.

> Oof - it's not like goodfire - we have quantitative results and a clearly defined method!

Goodfire is a discovery from research, an ahaha moment.

"concept elements" would have been a better descriptor in both cases. The V for the cats and the cars, and for your idea.

> 16,068 LLM ... The Coherence Model is a 42.8M-parameter transformer trained

My question is this: was this the smallest model you could train on the 16k papers? Isn't there an optimal small size to identify everything that is common?

Wouldn't it make more sense to use such a model to strip OUT the commonality? Get rid of everything where the "existing distribution of expertise and collaboration networks" are already optimal - and what's left is the novelty?

Ultimately I would be curious to see if your existing "atoms" happen to follow Zipfs law...

> "idea atoms"

I made it to the first figure in the paper before I raised an eyebrow at this term.

Then I thought of "Already Clean": https://already-clean.goodfire.com

I cant help but think of Plato's forms -

> and stopped getting talked about a few days later? Yes.

No one was talking about PHP last week either. Till we all got a reminder that a massive chunk of the web is still happily running on it.

Being talked about is a terrible metric for anything other than how much people are talking about something.

If you're going to do this: start with: Dostoevsky, Tolstoy, and "Doctor Zhivago". These are the Poe/Twain/London for Russia (to some extent).

I highly reccomend Russian film as well: Stalker and Solaris (Not Cloonie, the Russian version) are also great. I have a bit of a love of Sci Fi, may not be your tastes.

You should also hunt around for anything and everything you can find on early Russian computing, Cybernetics, their efforts and thoughts on central planning, and the ternary computing projects they did.

> “Formally verified” does not mean “bug-free.”

Is formal verification the path forward? Meh... I dont think so. Does it have applications in some areas: yes. But this is just more robust "unit testing".

Anthropic keeps trying to convince you that "one agent" will do the job. It does so with the statement "We built 36 small verification tasks from real OSS-Fuzz code. Opus 5.5 proves half of them with about 40% fewer tokens than Opus 5. We expect this trend to continue as models are trained on long verification tasks."

But this is not the path forward. Robust E2E, durable (can your app go from python to rust with out loosing tests) Simulation/Emulation tests are the way. This is an "in bulk" and "brute force" activity. Go look at TigerBeetle for an example of this in practice. Go look at how oxide is doing testing.

Can a swarm of cheaper agents write these tests more effectively that Anthropics premium products. I assure you that it can.

If you look at OAI's actions (hugging face hack) and anthropic own publication on its version control AI is not a best of breed solution its a brute force hammer at it till you get what you need solution.

> What's the point of asking that in the first place

Not having read enough Russian literature.

> AI labs have not worked well with affected communities at all and often downright exploited them

The first reviews of OAI's dump on math are coming out. Quasi-Riemann Hypothesis Proof looks to be valid. The irony is that is uses and approach that the people who work in that area have disregarded for the most part. As well as mining some other parts of maths corpus. There really isnt "innovation" there.

For as much as OAI is guilty of leveraging work in progress, the complaints about some of the other work(s) are looking spurious (and I'm being very kind using that word).

Shit happens.

Years ago I lost $250K at work, well gave away, so real money out the door. I figured out why we were bleeding, that the mistake was mine (2008 vs 2009 in a function) and then changed the pattern to use explicit names rather than numbered events.

I figured at the end of the day I was fired. My boss laughed at me and told me good job for owning my mistake, fixing it, and then making sure it could not happen again.

Honestly don't just apologize - tell the whole story, in full. Tell us what you should have done differently at each step. Tell us where the process failed you and us. Those sorts of narratives and suggestions are how processes get improved. There is a humble and informative post mortem that you can write up and publish (beyond HN) for everyone's edification. But that should honestly be a Monday thing, when you have a clear head and some more perspective on the whole situation.

Honestly though, dont beat yourself up over this.

Anendophasia? Fair number of people have this.

> but anything of importance would have already had active maintenance and patching

Mostly because people either "did it" or yelled long enough and loud enough that it "got done". There is a saying about horses and water, but some people needed to be dragged, kicking and screaming and have their faces shoved in the well.

When headlines like "there wasn't an easy exploit" come out I have burned a lot of good will, on what people will perceive as "nothing burger". The security space has a knack for making a complete mockery of itself and then wondering why no one seems to give a fuck.

"I'm getting too old for this shit" really does apply to today.

Ya dont say.

Negative Free Cash Flow are the words every one on HN needs to learn.

Every thing before now, you will look back upon as the happy time. When every year storage, ram and compute grew and the prices went down. Your cloud provider made so much money they paid cash for hardware.

Now they borrow at 8% and net margins on ram/storage were around $0.10/GB ... today they sit at over a dollar. (SEE micron's s10's if you want to check the math on that).

Forget circular financing, its a circular firing squad because your going to have to figure out where the hell your going to run any of your gear.

4x to 10x for ram +8% for financing +30percent AWS margins. Good luck everyone its gonna get rough out there.

You're looking for comfort in the wrong place.

From microns S10: BITS shipped and net revenue.

Q1 '20 - $0.10 profit per gb shipped (ram and storage)

Q1 '23 - -$0.024 (cyclic ram market)

Q1 '25 - $0.131 this looks normal were on the upswing.

Q4 '26 - $1.136 Were all fucked.

If you look at a chart of S3 prices we have all been living in the "Happy Time" where the cost to run our ever growing software business has gone down. Githb says 2025 billion commits - 14 billion this year, and 2x the actions run. That means AI driven software growth is 2x to 14x.

But those ram/storage prices are going to hit the operational budget of EVERYONE soon.

Can your average sas survive a 4x price increase on its AWS/Azure bill? How about 10x (cause thats what the margins say the premium is on ram).

If the circular financing is a concern, the circular firing squad should be a bigger one. Your paying for tokens to build more software to pay more to run your software is not a sustainable cycle. People are starting to use the words CapEx again - and it's a good thing.

What I find amusing is that tech used to look outside itself for perspective. We did usability, ethnographic research was a big thing, I have taken paper prototypes (a long time ago) to the bar and chatted up strangers about them for a beer.

The fact that you're doing that with your wife, with her consent, is no different. The real value is in keeping you grounded by someone outside tech.

Cause it looks like it's mining one subfield for a pattern and applying it to another. (The take from the few math folks I know who have looked into items of their interest).

Computing has the same issue. If we swapped a PLC programer with your average SV software engineer for a week both of them would have a really bad time.

> They “used math for what they needed, which was to evaluate their models,”

OAI first paper on the prize was a pretty shitty thing to do. But the rest of this looks like they are just mining the corpus and taking a brute force approach to do it.

Im going to say this again, the "pure math" academic types need to get a grip. Plenty of your peers got PHD's and happily made a fortune in finance, while screwing the world (See 2008 housing bubble) and none of you said a peep about that.

> howunfortunate

I have never seen a user name better fit a hacker news comment.

This also perfectly sums up the whole emotional roller coaster of modern science communication.Oh that could be cool (title)... Hey this might be good work... Fuck thats depressing I will never get this time back.

I would have thought that too. But at the start of the year the non technical folks in my life (lawyers, writers) were excited about running containers and having things on "cron jobs".

Those same folks now use version control - and have saner backup methods than some companies that I have worked for. They are already using multi pass, multi agent workflows and have adopted some fairly sophisticated processes for validation. They are primed for "more agents" attacking a problem in parallel.

Find a worker out side the tech bubble making heavy use of AI and one will feel confident that their future employment is secure.

Someone will report back "It didn't one shot my odd ball task"

Did it do it in 2 tries? 10? It's still coming out cheaper. Brute force the work. This is the path that the labs are taking. It's the 1200 agents trying to solve Hacking Bench hacking hugging face. It's anthropic telling you that its 1000's of staff commits like 10k engineers.

Treat AI like you're deploying interns who don't need coffee, food or bathroom breaks. With the right metrics they will get to an optimal (enough) solution.

> "if I could do it, then it's allowed".

This is in my back yard: https://www.youtube.com/watch?v=wZjInmHsWdM

The thing is hams shenanigans tend to impact other hams and not the general public. Every dive bar, and friend group has one asshole, but typically not two.