HNHacker News
TopNewBestAskShowJobs

bwfan123

1,311 karma · joined December 10, 2024

submissionscomments
bwfan123··on The Kimi K3 Moment
> I never truly understood what the intended business model around LLMs was

What appeared initially to be a huge innovation was later easily duplicated by many. There are no platform-lockins or network effects. Switching costs for users are zero, and there are low barriers to entry, with vast numbers of models to choose from and more appearing every day. As a business a token will be a commodity like an electron. Doesnt matter who produces it, or how (solar, wind, coal, nuclear etc) as long as it powers my toaster.

bwfan123··on Short sellers notch $8.7B profit as SpaceX shares dip to IPO price
> who do you make money from?

When you own stock at a broker in a margin account, you may sign an agreement to allow the broker to lend out your stock to someone else. For lending your stock, you are entitled to a stock-borrow fee which usually is quite small say 0.25%, and paid by the borrower (short-seller). The borrower then sells the stock to someone else. At a later point, the short seller closes their position by buying it back, and returning it to you. This is roughly the mechanics of it. So, to answer your question, the short seller makes money from folks who buy high and sell low. In this specific example, the stock-borrow fee say was 5% because, the float is still low, and if the short seller borrowed at $165 after the IPO and sold it, and then bought it back at $135 and closed their position, they made money from folks who bought at $165 and sold at $135.

bwfan123··on Claude Code May–July 2026 weekly limits promotion
> but the whole gang of frontier labs, breastfed with a mix of VC money and pension funds from the average Joe, is being pressured by Chinese lab

Can you point to data that suggests rate of token revenue growth is slowing ? This announcement, and the price cuts from openai are hints in the direction of pressure from open weights models but not conclusive.

bwfan123··on AI boosts research careers but narrow the span of ideas explored: study
> AI is largely automating the most tractable parts of science rather than expanding its frontiers

By definition, creativity cannot be automated, and AI is a fantastic automation machine. It can explore thinking paths at a rate humans cannot match. But creativity is bringing the unthinkable into the thinkable, and that requires sensory experience [1]. Specifically, new definitions and symbols which never existed before. Imagine the concept vector space, and expanding that with new independent dimensions. Is that even possible ? When you look at history the answer is yes !. And each time there was an independent dimension added, it was an act of genius. It is an instructive exercise to name these moments in history where an independent dimension was added to human thought. Some examples in math would be the invention of a number, and in politics could be the idea of democracy. By contrast, LLMs are trapped in the vector space they are trained on, and they lack the feedback loop with sensory experience to be able to create and validate theories.

[1] https://philsci-archive.pitt.edu/28024/1/Scientific_Inventio...

bwfan123··on Nvidia, CoreWeave, and Nebius: Inside the Circular Financing of the GPU Boom
> For almost everything else, the answer is no. No one else would pay the real costs to run them. It'll require the whole industry to shrink down massively compared to what we're seeing now - down to a profitable (and much smaller) core.

If there is any data to support this, please share.

bwfan123··on Nvidia, CoreWeave, and Nebius: Inside the Circular Financing of the GPU Boom
Circular financing is a dead horse - dont beat it. Instead, what is more interesting could be: Is there a path to these builds becoming economically profitable ? Towards this, some metrics to watch are: 1) ROI per token per dollar 2) Enterprise token budgets. And at what point there is an overbuild relative to the token roi. Alternatively, pressure on token costs due to the open weights models etc.
bwfan123··on Muse Spark 1.1
> I'm below $1 writing code all day.

same, pi as a harness works best, better than claude code and open code.

bwfan123··on Postgres rewritten in Rust, now passing 100% of the Postgres regression tests
> All these "rewritten in rust" projects only reinforce the idea that a significant part of the rust community consists of software talibans and not of engineers who must deliver something that works and is reliable over time.

Nailed it ! There are some folks who behave holier-than-thou just because they happen to use some language. Language missionaries if you will, and they are insufferable.

bwfan123··on Postgres rewritten in Rust, now passing 100% of the Postgres regression tests
yea, I found LLMs poor at reasoning through system-wide behaviors. Good luck to the vibers when they try to rootcause a rare data-corruption problem. They will hand it off to fable which will attempt trial-and-error bug fixes which will eventually muck up the code. What happens then ?
bwfan123··on Benchmarking coding agents on Databricks' multi-million line codebase
Doesnt this prove that there is really no moat in proprietary models for coding usecases, or the gaps is narrowing ? Also, since GLM5.2 can be run equally on amd and nvda, I guess there is no hw moat either. Further, switching costs for users is minimal not only in agents but models too. So there is really no stickiness or user preference involved. For this usecase I think it is a race to the bottom for costs in a good way for developers.
bwfan123··on A global workspace in language models
> written in some kind of self-aggrandising, wooey wonderous

They are drunk on their own kool-aid. To the rest of us it is very annoying, and makes me want to say: it is just a freaking weights machine, stop.

bwfan123··on Zuckerberg 'Admits' Meta's Layoffs Were Ineffective
The bigger story here is that the AI agents were not effective. This is a huge admission and explains why Anthropic, OpenAI, and now Microsoft are creating giant teams of FDEs (forward deployed engineers). It ties in with Meta becoming a GPU cloud to sell "excess AI compute". xai and now meta renting excess AI compute and shovel-vendors are trying to keep the embers alive with free tokens in a last-ditch effort.
bwfan123··on Please stop the AI confidence theater
> fact that the mechanism is clearly weaponized for marketing

There is a lot of dumb money chasing the AI dream. That money is not asking hard questions about return on investment yet. But, the narrative is starting to shift as evidenced by this article. Even Meta is questioning the value of AI Agents.

In response, we see Anthropic, OpenAI, and Microsoft creating giant teams of FDEs (forward deployed engineers). The idea is to keep the AI dream alive one way or another because once that dies, the whole thing comes crashing down. So, dumb money is going to put up a strong fight in the marketing-sphere, and articles like these are needed to counter that.

bwfan123··on AI fake news complaining about how AI fake news is the death of real news
> It’s really hard to build a coherent pro-AI argument

agree, given that we have a natural tendency to believe what we read - or only comprehend what we believe [1].

[1] https://bear.warrington.ufl.edu/brenner/mar7588/Papers/gilbe...

bwfan123··on Why I Stopped Arguing with People
> I wanted to keep getting better. And the only door to that is the one ego keeps slamming shut.

Koan to the author: What was never lost can never be found.

bwfan123··on Most arguments are about ego, not ideas
> The value is in the feeling of euphoria you get when dominating the other person by being unequivocally right

Frequently accompanied by a feeling of despair when you are dominated by another.

> The fact that this thread even exists speaks to the fact of the extremely high survival benefit this behavior confers onto a human.

Dialog brings clarity. Clarity helps build tools. Tools help survival. So, if there is anything to seek, try clarity of understanding - not being right, and being right stands in the way of understanding.

bwfan123··on Godot will no longer accept AI-authored code contributions
Brandolini's law in action.

It takes 10x more effort to refute BS than to generate it. Reviewing code is refuting. So is verifying the correctness of propositions. Generating propositions is easy, Refuting it requires proving the truth value or finding contradiction.

For maintainers of open source whose time is scarce, their energies are wasted needlessly, and I am all for conserving and directing energies productively.

bwfan123··on Godot will no longer accept AI-authored code contributions
> if it runs, it runs, regardless of who or what made it.

If it fails to run who is responsible to fix ? Problem with AI is that it does not have causal-models of how something works, and the reason for that is that it doesnt interact with the world. I think of it as an armchair expert who spouts recommendations without having a grounded understanding in the real world.

bwfan123··on South Korea to spend $1T on more memory chip production and humanoid robots
> $585B on new fabs,

This is how overcapacity happens in a commodity. The is a rush to expand capacity to meet demand, and there will be overshoot. I hope these factories are built quickly so the memory crunch eases.

bwfan123··on Mag 7 starting to underperform [pdf]
I like that the invisible hand of market is slapping the Mag-7 for capex which is the only way to discipline them. Investors are waking up to say: hey, you are spending all your profits on data-centers, where is the return for me ? But, it surprises me that there are vast pools of capital which we collectively call the "market" that makes these calculations, or maybe a simpler causal explanation is the missing stock repurchase bid. At some point, one of the hyperscalers (msft ?) will break from the pack and announce reductions to capex and increase stock repurchases to stem the decline.
bwfan123··on Zuckerberg's Increasingly Bizarre War on Whistleblowers
> Nobody cares about the “how.”

The rich man is also perceived by the lizard brain to be wise, intelligent, witty, and handsome.

bwfan123··on AI in mathematics is forcing big questions
Loved Bessis's book and his writings because of his human-centric view which is missed by the ai maximalists. I just cant wait for the upcoming AI bust to skim the froth in the debate.
bwfan123··on Ford AI hiccups push carmaker to rehire ‘gray beard’ inspectors
Back in dot-com, there used to be a website called f'ed company that chronicled the dot-com dead pool. This time around there needs to be a similar website that records AI walk backs so it helps the mgmt class not make stupid decisions.
bwfan123··on Qualcomm to Acquire Modular
Mojo seemed like a passion project. The fundamental problem was never the lack of a great programming language, and inventing a python bastard child of a language is not a solution. But, respect and congrats to the Modular folks. HW companies have notoriously bad software teams and culture and hopefully this injects some good sw dna into the acquirer.
bwfan123··on Anthropic says Alibaba illicitly extracted Claude AI model capabilities
Model makers need to get off their high-horse, and face the reality that they are selling a commodity.
bwfan123··on Mark Zuckerberg directed Meta to create a prediction markets app
> There is no leadership here

They treat employees like shite, and the only folks left or willing to join are the mercenaries. It is a testament to their business model that it survives in spite of all this.

bwfan123··on OpenAI DayBreak – GPT-5.5-Cyber
Security is a great business model. You sell locks, and if thieves break in, you sell more and stronger locks. There are no consequences for the lock-maker. Similarly, AI tools can both find and fix vulnerabilities. Not only that, AI can create vulnerabilities in the code it generates. Now that is a perpetual money machine.
bwfan123··on GPT-5.5 hallucinates 3x more than MIT-licensed GLM-5.2
> I don't want a model that says "I don't know", because I will verify the answer anyway.

Few people actually review answers or code. Because they have been sold the myth that these models can do it all. The main problem is that LLMs dont have causal models, and as a result, their reasoning is a high probability word salad and not a logically sound argument. Particularly on tricky corner cases which it hasnt encountered. I would still agree with you that sometimes hallucinations are actually useful as it provides a strawman, and having even a hallucinated answer to spar with is better than a "dont know".

bwfan123··on Zen and the Art of Machine Learning Research
> To be done with doing, would appear to require passive income?

who is asking this question ?

bwfan123··on Sixty percent of US consumers say 'AI' in brand messaging is a turnoff
I immediately skip all youtube ads that are AI generated, and I can easily spot them. Not just that, I am planning to actively avoid those brands that are doing this. A small way to send a message that this form of AI sucks. In addition, I make it a point to compliment humans that assist me in customer support, and ask them to tell their management that they need to hire more humans.
← PreviousPage 4 of 18Next →