Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The impact of competition and DeepSeek on Nvidia

youtubetranscriptoptimizer.com

Post reply on HN
Print view View original

301–310 of 500 posts

Re: The impact of competition and DeepSeek on Nvidia

#301

Post by godelski »

lxgr wrote: post #275

The most important part for me is: > DeepSeek is a tiny Chinese company that reportedly has under 200 employees. The story goes that they started out as a quant trading hedge fund similar to TwoSigma or RenTec, but after Xi Jinping cracked down on that space, they used their math and engineering chops to pivot into AI research. I guess now we have the answer to the question that countless people have already asked: W…

Interestingly a lot of the math and physics people in the ML community are considered "grumpy researchers." A joke apparent with this starter pack[0].

From my personal experience (undergrad physics, worked as engineer, came to CS & ML because I liked the math), there's a lot of pushback.

  - I've been told that the math doesn't matter/you don't need math.
  - I've heard very prominent researchers say "fuck theorists" 
  - I've seen papers routinely rejected for improving training techniques with reviewers say "just tune a large model"
  - I see papers that show improvements when conditioning comparisons on compute restraints because "not enough datasets" or "but does it scale" (these questions can always be asked but require exponentially more work)
  - I've been told I'm gatekeeping for saying "you don't need math to make good models, but you need it to know why your models are wrong" (yes, this is a reference)
  - when pointing out math or statistical errors I'm told it doesn't matter
  - and much more.
I've heard this from my advisor, dissertation committee, bosses[1], peers, and others (of course, HN). If my experience is short of being rare, I think it explains the grumpy group[2]. But I'm also not too surprised with how common it is in CS for people to claim that everything is easy or that leet code is proof of competence (as opposed to evidence).

I think unfortunately the problem is a bit bigger, but it isn't unsolvable. Really, it is "easily" solvable since it just requires us to make different decisions. Meaning _each and every one of us_ has a direct impact on making this change. Maybe I'm grumpy because I want to see this better world. Maybe I'm grumpy because I know it is possible. Maybe I'm grumpy because it is my job to see problems and try to fix them lol

[0] https://bsky.app/starter-pack/roydanroy.bsky.social/3lba5lii... (not perfect, but there's a high correlation and I don't think that's a coincidence)

[1] Even after _demonstrating_ how my points directly improve the product, more than doubling performance on _customer_ data.

[2] not to mention the way experiments are done, since it is stressed in physicists that empirics is not enough. https://www.youtube.com/watch?v=hV41QEKiMlM

Re: The impact of competition and DeepSeek on Nvidia

#302

Post by visarga »

rayval wrote: post #284
fairity wrote: post #270

Earlier quoted context omitted.

I think it's worth double clicking here. Why did Google have significantly better search results for a long time? 1) There was a data flywheel effect, wherein Google was able to improve search results by analyzing the vast amount of user activity on its site. 2) There were real economies of scale in managing the cost of data centers and servers 3) Their advertising business model benefited from network effects, where…

In theory, the more people use the product, the more OpenAI knows what they are asking about and what they do after the first result, the better it can align its model to deliver better results. A similar dynamic occurred in the early days of search engines.

I call it the experience flywheel. Humans come with problems, AI asistant generates some ideas, human tries them out and comes back to iterate. The model gets feedback on prior ideas. So you could say AI tested an idea in the real world, using a human. This happens many times over for 300M users at OpenAI. They put a trillion tokens into human brains, and as many into their logs. The influence is bidirectional. People adapt to the model, and the model adapts to us.. But that is in theory.

In practice I never heard OpenAI mention how they use chat logs for improving the model. They are either afraid to say, for privacy reasons, or want to keep it secret for technical advantage. But just think about the billions of sessions per month. A large number of them contain extensive problem solving. So the LLMs can collect experience, and use it to improve problem solving. This makes them into a flywheel of human experience.

Re: The impact of competition and DeepSeek on Nvidia

#303

Post by coliveira »

rightbyte wrote: post #289
coliveira wrote: post #235

Earlier quoted context omitted.

Yes, for people in academia the trend is clear, but it seems that WallStreet didn't believe this was possible. They assume that spending more money is all you need to dominate technology. Wrong! Technology is about human potential. If you have less money but bigger investment in people you'll win the technological race.

I think Wall Street is in for surprise as they have been profiting from liquidating the inefficiency of worker trust and loyalty for quite some time now. It think they think American engineering excellence was due to neoliberal inginuenity visavi the USSR, not the engineers and the transfer of academic legacy from generation to generation.

This is even more apparent when large tech corporations are, supposedly, in a big competition but at the same time firing thousands of developers and scientists. Are they interested in making progress or just reducing costs?

Re: The impact of competition and DeepSeek on Nvidia

#304

Post by rightbyte »

lxgr wrote: post #293
rfoo wrote: post #285

Earlier quoted context omitted.

This is completely fake though. It was more like their founder decided to start a branch to do AI research. It was well planned, they bought significantly more GPUs than they can use for quant research even before they start to do anything AI. There was a crack down on algorithmic trading, but it didn't had much impact and IMO someone higher up definitely does not want to kill these trading firms.

The optimal amount of algorithmic trading is definitely more than none (I appreciate liquidity and price quality as much as the next guy), but arguably there's a case here that we've overshot a bit.

The price data I (we?) get is 15 minute delayed. I would guess most of the profiteering is from consumers not knowing the last transaction prices? I.e. an artificially created edge by the broker who then sells the API to clean their hands of the scam.

Re: The impact of competition and DeepSeek on Nvidia

#305

Post by ieee2 »

hn_throwaway_99 wrote: post #295

I'm curious if someone more informed than me can comment on this part: > Besides things like the rise of humanoid robots, which I suspect is going to take most people by surprise when they are rapidly able to perform a huge number of tasks that currently require an unskilled (or even skilled) human worker (e.g., doing laundry ... I've always said that the real test for humanoid AI is folding laundry, because it's an…

I saw such robot's demos doing exactly that on youtube/x - not very precisely yet, but almost sufficiently enough. And it is just a beginning. Considering that majority of the laundry is very similar (shirts, t-shirts, trousers, etc..) I think this will be solved soon with enough training.

Re: The impact of competition and DeepSeek on Nvidia

#306

Post by SirMaster »

adastra22 wrote: post #111
xuki wrote: post #99

Earlier quoted context omitted.

He was famous before the PS3 hack, he was the first person to unlock the original iPhone.

Yes, but it's worth mentioning that the break consisted of opening up the phone and soldering on a bypass for the carrier card locking logic. That certainly required some skills to do, but is not an attack Apple was defending against. This unlocking break didn't really lead to anything, and was unlike the later software unlocking methods that could be widely deployed.

Well he also found novel exploits in multiple later iPhone hardware/software models and implemented complete jailbreak applications.

Re: The impact of competition and DeepSeek on Nvidia

#307

Post by doctorpangloss »

rfoo wrote: post #285
lxgr wrote: post #275

The most important part for me is: > DeepSeek is a tiny Chinese company that reportedly has under 200 employees. The story goes that they started out as a quant trading hedge fund similar to TwoSigma or RenTec, but after Xi Jinping cracked down on that space, they used their math and engineering chops to pivot into AI research. I guess now we have the answer to the question that countless people have already asked: W…

This is completely fake though. It was more like their founder decided to start a branch to do AI research. It was well planned, they bought significantly more GPUs than they can use for quant research even before they start to do anything AI. There was a crack down on algorithmic trading, but it didn't had much impact and IMO someone higher up definitely does not want to kill these trading firms.

Who knows? That too is a bunch of mythmaking. One thing's for sure, there are no moats or secrets.

Re: The impact of competition and DeepSeek on Nvidia

#309

Post by UncleOxidant »

Even if DeepSeek has figured out how to do more (or at least as much) with less, doesn't the Jevons Paradox come into play? GPU sales would actually increase because even smaller companies would get the idea that they can compete in a space that only 6 months ago we assumed would be the realm of the large mega tech companies (the Metas, Googles, OpenAIs) since the small players couldn't afford to compete. Now that story is in question since DeepSeek only has ~200 employees and claims to be able to train a competitive model for about 20X less than the big boys spend.

Re: The impact of competition and DeepSeek on Nvidia

#310

Post by lxgr »

auntienomen wrote: post #297
lxgr wrote: post #275

The most important part for me is: > DeepSeek is a tiny Chinese company that reportedly has under 200 employees. The story goes that they started out as a quant trading hedge fund similar to TwoSigma or RenTec, but after Xi Jinping cracked down on that space, they used their math and engineering chops to pivot into AI research. I guess now we have the answer to the question that countless people have already asked: W…

DeepSeek is a subsidiary of a relatively successful Chinese quant trading firm. It was the boss' weird passion project, after he made a few billion yuan from his other passion, trading. The whole thing was funded by quant trading profits, which kind of undermines your argument. Maybe we should just let extremely smart people work on the things that catch their interest?

Interest of extremely smart people is often is strongly correlated with potential profits, and these are very much correlated with policy, which in the case of financial regulation shapes market structures.

Another way of saying this: It's a well-known fact that complicated puzzles with a potentially huge reward attached to them attract the brightest people, so I'm arguing that we should be very conscious of the types of puzzles we implicitly come up with, and consider this an externality to be accounted for.

HFT is, to a large extent, a product of policy, in particular Reg NMS, based on the idea that we need to have many competing exchanges to make our markets more efficient. This has worked well in breaking down some inefficiencies, but has created a whole set of new ones, which are the basis of HFT being possible in the first place.

There are various ideas on whether different ways of investing might be more efficient, but these largely focus on benefits to investors (i.e. less money being "drained away" by HFT). What I'm arguing is that the "draining" might not even be the biggest problem, but rather that the people doing it could instead contribute to equally exciting, non-zero sum games instead.

We definitely want to keep around the the part of HFT that contributes to more efficient resource allocation (an inherently hard problem), but wouldn't it be great if we could avoid the part that only works around the kinks of a particular market structure emergent from a particular piece of regulation?

Post reply on HN