Their graph leaders are fusions of frontier models. Not of local models.
I saw one single fusion of local models.
If their claim and Opus 4.8 is a smaller, "older" model then again that's pretty absurd.
What is with all of these claims on HN recently where people want open models to win (which I understand and can support), but have absolutely bs claims supposedly showing it and ignore that the claims are entirely made up.
In this case they have only a single "fusion" of non latest models. They found a metric where Deekseek 4 beats out gpt 5.5 and Opus 4.8 (which is already pretty suspicious to me), and showed that adding two other models it still beats them out and doesn't pass Fable.
What in the world is the actual claim here? Fusion didn't drag it down??
If you found something where Deepseek 4 is already supposedly better than Opus and GPT 5.5 the rest is just smoke and mirrors.
Is HN now just dead internet theory? Nobody can look at this for 2 mins and conclude anything other than it's nonsense. And this is the third topic I've had this conversation on recently on HN. Highly upvoted blogspam, that claims "open models are great" and is heavily stretching the truth or out right intentionally misleading.
I'm not agreeing or disagreeing on open models, I'm saying why do people keep posting clear spam that is clearly misleading on the topic?
> The stat you read is flat out inaccurate. There are 60 minutes where the clock is running, and the vast majority of that is with the ball live and in play.
You are completely wrong. I know that's what your intuition feels like from watching games, but if you actually get a stopwatch out and clock it, you'll see it's much closer to op than what you posted. Very close in fact.
If you don't believe it, find a actual game and time it. I did this on a game.
For example from the 8min mark to the 2min warning (6 mins of clock time) there were only 2m15s of action. Only 37.5% of the clock time was action. And that was an extremely conservative time. It was the 4th quarter, had multiple scoring plays, and multiple timeouts - essentially everything possible to do to stop the clock and still it maxed out at < 38% of the clock was actual game time. If you instead do this for a full game (not the end of the 4th quarter), and especially games that aren't close and you'll see it's way less. Definitely not what you're expecting.
Football is a tiny amount of "action time" as compared to "clock time".
> Capitalism does not guarantee competition (quite the opposite, strong property rights are the nexus of anticompetitive opportunity) which does not remove all profit over the long term, it squeezes it onto assets, which is where that unearned income we were talking about originates from. If you have ever heard or given a business pitch, attended a class in business school, or listened to a VC for 30 seconds you have heard some heinously anticompetitive scheme and their plan to leverage it for personal gain by turning it into an asset they own. Network effects, platform effects, two sided markets, returns to scale, etc etc etc. Usually they don't work, but when they do and you get a stock or a deed or a title to a money fountain (exploitation fountain, seen from the other side) you get to stack trillions while the competition spends decades trying to cross your capitalism-created and capitalism-guaranteed moat.
I made this point many times a number of years back and gave up. It's incredible how an entire message board of HN that supposedly is extremely pro market competition, seems to entirely be unaware (or just collectively puts it's head in the sand) that the #1 strategy that most VC backed firms seem to target is "figure out out as quickly as possible how we can get out of having to compete with others". And they do so under the name of "a moat".
Building a moat is one of the most anti-market actions that can be taken. You hear commenters post non-stop about the ills of communism as it avoids market competition, but somehow every seems to just gloss over or ignore the fact that moats are designed to do the same thing and cause the same issue. Terrible allocation of capital.
> Obviously their statements are insincere, because they are building the bloody things. If they were sincere that AI is like nuclear weapons, then they would be devoting all their cash and energy into lobbying the government to nationalize them...
This comment makes no sense. Id you think this tech is dangerous and happening soon and clearly they think the safest way to have it releases is to do so first and model safe ways of doing things. Clearly we cab agree or disagree it's internally consistent what they are doing and aligns with their statements.
And you and OP think the best way to be first to release this is tie all of their funding for the exponentially growing expense is to they notoriously slow moving, bureaucratic government includinf funding process? And the best way to develop it is to directly tie their fate to this notoriously capricious administration?
These comments make no sense. Even if you're completely against Anthropic those comments make no sense.
> Your area again making the same mistake as before.
> You are making the most passionate defense of team openai
At no point did I mention Openai, referr to openai or imply anything about openai (just mentioned your reference). Nothing I'm saying weighs in on any form of discussion or debate between Deepseek & Open Models vs OpenAI.
The fact that you are unable to separate those two is your failing, not mine. Your argument is the equivalent of the following:
A: Deepseek ran into a burning building last week and saved 10,000 orphans from a fire.
Me: No Deepseek did not save 10,000 orphans from a burning building last week. Regardless of what you think of Deepseek it didn't save 10,000 orphans. It's an LLM in a computer, not a humanoid robot - if you look at that for 2 seconds you see that claim is nonsense.
You: By attacking those supporting Deekseek you have declared yourself for team OpenAI and are clearly an OpenAI supporter!
Me: Saying deepseek didn't save 10k orphans has nothing to do with OpenAI. It is a lie saying that deepseek saved 10k lives. It's an LLM chat bot. Regardless of how anyone feels about deepseek - discuss it on it's merits not on bs.
You: See! You keep defending OpenAI you open AI shill! Stop passionately defending OpenAI!
"You're on the team against me so I oppose everything you say".
Again it's the same problem - what you're doing. I'm not on "team OpenAI". I'm also not on "team deepseek". I'm commenting on how so much of the population is literally unable to see the world unless it is filtered through some "team" lens that they are for or against.
Judge the material based on what's in the material. Not as it boosting or hurting your "team".
The material in this article is crap judge it as crap and say so regardless of your team.
But here you look at my saying something negative about a post that is pro "team deepseek" so the only conclusion you're able to make is that I must be for the other team.
It's the inability to think critically that is astounding me here. So many opinion's people have now is now just "is it for team or against my team". They are unable to even think of anything else.
I wrote that entire post and you even said you couldn't understand it unless you put it through a lens of being for or against a team...
"Don't you understand? I'm on team deepseek! It doesn't matter what's written about it. Heck it doesn't even matter if it's all lies - it supports my team and here's why I love my team."
> So if a demographic group is more likely to apply to some jobs they are not qualified for, this paper would say they are being discriminated against.
Your understanding appears to be incorrect.
> Our research also found that this pattern does not appear to be the case in other circumstances. We analyzed data from the largest prior study of hiring decisions, which sent 83,000 applications to 108 Fortune 500 firms during the same time period as our study and did not focus on whether AI was used to make decisions. We found that the rate at which applicants were rejected from every firm they applied to in this data was no higher than what you’d expect if each company decided independently of the others.
If it were what you were asserting, then this behavior and results would persist even without AI being used. Instead when they remove the filter for AI decisions (and AI mono-culture in decisions) the effect is no longer present.
This seems to strongly support they argument that effectively a single AI makes a single decision for a candidate across "all" positions they apply for rather than independently assessing them for each position.
Essentially it's more or less saying they're is one hiring manager for the entire industry and if they have a random reason they don't like you, you won't be hired for any job in the industry.
There is a single evaluation function for the industry and if it puts you a negative for any reason in the model's distribution, every job that uses it will do so.
An AI generated article about single ai run test which in theory had many components and the AI judge declared deepseek "won"?
How many runs were there on each test to account for some temperature variance? Only one.
Did deepseek write better code? Did GPT's code have bugs when doing the regex? The AI "news" article doesn't actually say that. It says that grok thought that GPT's approach could have bugs so it declared deep seek the winner.
This is absolute worthless methodology. And barely measurable methodology - nothing more than a prompt. No definition of what the scoring approach actually is. No definition of what "precision" actually means in this context. This is absolutely worthless and has no business being in the site, forget about on the front page.
So why is it's on the front page? Because it aligns with the current "feels" of the community that deepseek will get better and it shows "bad things" about the en vogue to dislike closed models.
I happen to agree with both of the views, but this site is utterly worthless.
If you want HN to be astro-turfed to the max, just up vote content like this without any critical reading of the.
I mean the past 6 months of "here is my chat gpt blog post of how to use a coding agent" are 1000x better than this "news article".
Seriously the amount of respect I've lost recently for the HN community is incredible. A bit harsh, but very true.
Maybe it's generational thing, maybe it's due to the state of politics, maybe it's a side effect of me getting older, but recently online has turned into nothing but people explicitly (or implicitly) writing about their "team". Comments on this post are nothing but people who clearly see themselves as being on "team deepseek" or "team open models" or some similar variant writing posts in support even though this is probably one of the worst "articles" to make it to the front page on ages.
It clearly doesn't matter. It supports something on their "team" so they support it via comments.
If kills any form of intellectual discussion. It's all just "this is my team".
So "unlimited" for you is literally one week more than 5 weeks tech companies cap at?
Why not take 10 weeks or 12 weeks? If it's unlimited?
And how many other employees are only taking half of what they would be taking because they feel silently pressured to - and do all of those missing weeks make up for the single extra week you're taking?
Yah this was the part that really threw me as well.
"People would be pleased to have a rejection from us. They'd be proud to carry it sounds with them. Lucky them!"
It's funny, I see an article from Yegge and thought "I like that writer, I haven't read any of his stuff in a while, I'll see what he has to say." Then got to the end and see the links to gas town and gas city and remembered it was the same Yegge that while having accurate foresight about orchestration of agents also was a bit off the deep end in gas town.
But the biggest thing I see in this article is it really sounds like "here is the new company I landed at, and rather than make a post about its product, I'm going instead make a post about how terrible the problem it solves really is, and a post on a proposed solution. And the cues what I'll pop up in a few weeks and just coincidentally post about this new company that just happens to solve this problem in the way I've convinced everyone is the right solution."
While I don't have any evidence of this that's the feeling I left with. And if so, then "thought leaders" are a lot more interesting when not "talking their book."
Man, maybe it's time for me to give the verge a subscription. There the only ones actually doing any journalism here and a bunch of AI blogs skimming off the top.
You'd be shocked at how many people who work on ads really do delude themselves into thinking people find ads "useful".
Their usual justification is in the end somewhere tied to "people click on ads so they must find them useful". And yet somehow always ignores the fact that their platform often does all it can to hide that ads are ads and makes them look as much like content as possible.
> A search through GPT‑5.5’s SFT data found many datapoints containing “goblin” and “gremlin.” Further investigation revealed a whole family of other odd creatures: raccoons, trolls, ogres, and pigeons were identified as other tic words, while most uses of frog turned out to be legitimate.
My feedback was it seemed like it was interesting to make, but for me at least not really interesting to solve.
It felt more like solving two crosswords than a single combined one.
Places where lots letters are shared between the two make it interesting, but they're weren't many. Sharing a single vowel 'e' isn't too interesting. Otherwise the just share the length which I already know the length for a crossword clue.
It seems like there is a cool idea in there and keep working on it, it just isn't there for me yet.
Their graphs don't even support their claims.
Their graph leaders are fusions of frontier models. Not of local models.
I saw one single fusion of local models.
If their claim and Opus 4.8 is a smaller, "older" model then again that's pretty absurd.
What is with all of these claims on HN recently where people want open models to win (which I understand and can support), but have absolutely bs claims supposedly showing it and ignore that the claims are entirely made up.
In this case they have only a single "fusion" of non latest models. They found a metric where Deekseek 4 beats out gpt 5.5 and Opus 4.8 (which is already pretty suspicious to me), and showed that adding two other models it still beats them out and doesn't pass Fable.
What in the world is the actual claim here? Fusion didn't drag it down??
If you found something where Deepseek 4 is already supposedly better than Opus and GPT 5.5 the rest is just smoke and mirrors.
Is HN now just dead internet theory? Nobody can look at this for 2 mins and conclude anything other than it's nonsense. And this is the third topic I've had this conversation on recently on HN. Highly upvoted blogspam, that claims "open models are great" and is heavily stretching the truth or out right intentionally misleading.
I'm not agreeing or disagreeing on open models, I'm saying why do people keep posting clear spam that is clearly misleading on the topic?