Can you reverse engineer our neural network? • Hacker News Board

Re: Can you reverse engineer our neural network?

#181

Post by ses1984 » Fri, Feb 27, 2026, 9:24 PM UTC

smithoc wrote: post #166

ses1984 wrote: post #150

Earlier quoted context omitted.

You could just read the article. This is outside my domain, and I don’t know the details, but in many cases Jane street functions as a market maker, market makers have access to information they can exploit to skim from anyone that trades through them, especially retail investors who place market orders. Pump and dump is a strategy that whales can use to bully smaller traders, not unlike how in poker the smaller your…

> Please correct me if I’m wrong. You are, about pretty much all of this. Being a market maker doesn't provide any special information. I'm guessing someone misunderstood something like Level II quotes ( https://www.investopedia.com/articles/trading/06/level2quote... ) as being information that hedge funds / investment banks / pros have that retail traders don't... but it's just semi-public information that anyone ca…

If it’s so ridiculous then why is this practice specifically forbidden on US regulated exchanges?

Re: Can you reverse engineer our neural network?

#182

Post by ericd » Fri, Feb 27, 2026, 9:44 PM UTC

gloryjulio wrote: post #149

lupire wrote: post #128

Earlier quoted context omitted.

Money is an IOU. It's not a secret that US central bank policy is to devalue those IOUs at a rate of 2% per year in order to artificially stimulate demand for production.

We can have different opinion on how much inflation is too much. But the universal consensus is that a little bit of inflation is good. It's ok if you think 2% too high. Maybe 1.25% is better. But we cannot just dispute this basic economic model and thinking that 0 or negative inflation (which would cause the stop of investment), or no consensus(that would just cause more chaos) is better. That's just absurd

The value of the USD wasn’t monotonically inflating away for most of its history, people were mostly fine? Were those periods disastrous in some way?

Re: Can you reverse engineer our neural network?

#183

Post by aizk » Fri, Feb 27, 2026, 9:45 PM UTC

expensive_news wrote: post #139

stingraycharles wrote: post #3

This is pretty cool, I wasn’t aware of these types of challenges. How does one even approach this? Feels to me like it’s similar to dumping a binary with an image, the format being entirely custom. And/or trying to decode a language or cipher, trying to recognize patterns.

I was one of the solvers. It took me about a week to figure out. This is what I wrote out in my submission with the answer: > After looking at the final two layers I was somewhat quick to intuit that this was some sort of password check, but wasn’t entirely sure where to go from there. I tried to reverse it, but it was proving to be difficult, and the model was far too deep. I started evaluating the structure and saw…

Did you get an interview with Jane Street?

Re: Can you reverse engineer our neural network?

#184

Post by ericd » Fri, Feb 27, 2026, 9:47 PM UTC

b112 wrote: post #107

ericd wrote: post #76

Earlier quoted context omitted.

It’s an important function, but the guys making these bets frequently pull down $10-$100M per year, each. That’s a huge toll to extract from the productive economy for playing this game. And then there are the guys managing things like pensions, skimming a percentage every year, just because they happen to be locked into that position, meanwhile underperforming a basket of index funds. Just happily eating away at the…

There's no such thing as money being squandered. Money isn't a resource. It's not food, or energy, or even an allotment of work. If you try to seize or reduce, what you perceive as excess, wasted money supply, that money supply will simply cease to exist.

It’s squandered from the perspective of the laborers and savers, because they don’t have it anymore.

I didn’t say anything about excess money supply, though, were you responding to a different comment?

Re: Can you reverse engineer our neural network?

#185

Post by aizk » Fri, Feb 27, 2026, 9:52 PM UTC

I worked on a puzzle like this roughly 2 years ago from Anthropic. I did the first half, the easier part of the CTF, and my friend did the second half, the more technical ML stuff. We both got interviews at Anthropic, which was cool - I wasn't anywhere close to nailing an interview at Anthropic but it gave me a lot of confidence to end up going all in on tech, which paid off greatly. My friend's short write up: https://x.com/samlakig/status/1797464904703910084

Re: Can you reverse engineer our neural network?

#186

Post by 0x3f » Fri, Feb 27, 2026, 9:55 PM UTC

dgacmu wrote: post #37

0x3f wrote: post #7

Earlier quoted context omitted.

We don't have any reliable and scalable way of doing this allocation, though, so it's a bit like saying that all the resources are wasted being locked up in asteroids.

We do have a lot of policy levers that are tilted in favor of making money in finance, though, and we could change those levers. - The carried interest loophole - We could add small transaction taxes - We could raise capital gains taxes - We could be a little more focused on enforcing antitrust - We could raise higher end marginal tax rates to reduce the relative attractiveness of off-scale payrate jobs - We could pr…

I'm not sure you could alter these without significant negative effects on the other things you're trying to encourage.

It's not like capital is uninvolved in the provision of biotech, or that medical startup founders aren't also motivated by massive tax-efficient future payouts, for example.

If anything I'd think you'd want to encourage the movement of investment into riskier bets, which would generally mean _decreasing_ capital gains taxes.

Re: Can you reverse engineer our neural network?

#187

Post by 0x3f » Fri, Feb 27, 2026, 10:01 PM UTC

dgacmu wrote: post #142

logicchains wrote: post #43

Earlier quoted context omitted.

>long-term societal importance without bringing in any form of centralized resource allocation. The onus is on the biomedicine industry to demonstrate it's capable of producing anything of societal importance because so far it's largely failed to deliver. There's nothing noble or scientific about throwing good money after bad into an industry that's continuously failed to deliver.

You'll note I didn't try to specify what was of import. We can create mechanisms that enable more people to follow their own idea of what is important instead of merely what is lucrative. Not everyone will agree with the choices other people make, and it wouldn't eliminate money as a motivating factor, it would just slightly reduce the strength of that signal relative to other potential signals such as "I feel like I…

Honestly I think you'd be better off just making _existing_ cheaper than trying to discourage wealth seeking. The cost of living, especially housing, is a huge motivating factor to try and 'get rich first', before doing the 'meaningful work' part.

Re: Can you reverse engineer our neural network?

#188

Post by bumby » Fri, Feb 27, 2026, 10:24 PM UTC

wat10000 wrote: post #69

bumby wrote: post #25

Earlier quoted context omitted.

It’s been a few years since I looked deeply into it, but I think we produce enough to have everyone survive but not necessarily thrive. At the time, it came out to something like 1700 kcal per person. Even if we did have enough, the next problem is logistics of allocating that food to everyone who needs it.

According to this, it’s more like 3,000kcal/person/day: https://ourworldindata.org/agricultural-production#explore-d...

Where are you seeing that value in the link?

The average per capita is closer to 2,600 kcal/day. Not sure how that breaks down when normalized by the individual country population. It also doesn't include waste. In the US at least, waste is near 40%.

Re: Can you reverse engineer our neural network?

#189

Post by robotpepi » Fri, Feb 27, 2026, 10:25 PM UTC

charcircuit wrote: post #158

robotpepi wrote: post #155

Earlier quoted context omitted.

it seems to me that the problem is quite the opposite. people believe that the "importance of allocation of capital" (good euphemism by the way) is WAY more important that it really is. do we need extra personalized ads in each of our machines? do we need instant financial trades and people optimizing instant transactions? We don't need a sophisticated AI to "inform" the customers. there is a ton of things that are t…

>do we need extra personalized ads in each of our machines? If we want local, small, or niche businesses to exist. >do we need instant financial trades and people optimizing instant transactions? If we want people to easily get a fair price when they want to buy or sell something.

we had local, small and noche business before. and even today.

the point again is that nobody says that certain mechanism of the market can have positive effects. the point is that way overestimated. we have extremely complex procedures that cost insanely amounts of money for stuff like ads. we could have a fraction of that power and people would still know about the products they need, etc.

Re: Can you reverse engineer our neural network?

#190

Post by wat10000 » Fri, Feb 27, 2026, 10:33 PM UTC

bumby wrote: post #188

wat10000 wrote: post #69

Earlier quoted context omitted.

According to this, it’s more like 3,000kcal/person/day: https://ourworldindata.org/agricultural-production#explore-d...

Where are you seeing that value in the link? The average per capita is closer to 2,600 kcal/day. Not sure how that breaks down when normalized by the individual country population. It also doesn't include waste. In the US at least, waste is near 40%.

At the top, under FOOD, select "All food." In the country list, select "World." Above the map, click "Line." Finally, mouse over the graph to see the values. The latest value they have is for 2023 and shows 3,005.52kcal/day. Awful lot of significant figures there....