HNHacker News
TopNewBestAskShowJobs

whatshisface

31,546 karma · joined September 12, 2012

submissionscomments
whatshisface··on Can we have the day off?
When you say technological era, do you mean anything after the introduction of potery, writing, or what? Because you can trace slavery back to the earliest recorded civilizations, but it won't be easy to trace it to before a written record.
whatshisface··on Can we have the day off?
The past had slaves too. The feudal arrangement of tenant farmers was only one system of labor. I don't think union bargaining was ever tied to specific advances in technology.
whatshisface··on Can we have the day off?
If it was utilization of fixed capital that motivated the maximum-length workweek of today and centuries past, they wouldn't mind who was on the shifts or how many so long as there were three of them.
whatshisface··on Can we have the day off?
The drivers behind the asymptotic scaling are the tasks that can't be compartmentalized into prompts, the same tasks that couldn't be compartmentalized into a request for another person to do.
whatshisface··on Can we have the day off?
The "flipping out" aspect is something that does not seem to have a lot to do with technology at this time or in the past.
whatshisface··on Can we have the day off?
Working out why the workweek is 5 days, non-negotiably, even if you'd be willing to be paid less in proportion, comes down to realizing that it's being maximized subject to the constraint that everybody would flip out if it was 6, and then working out why it's being maximized.

What it's telling you is that a company would rather have 4 people working 5 days a week than 5 people working 4 days a week. The reason for that is, productivity drops a lot when it's spread out over multiple people. The reason behind that is communication overhead - the more context an individual carries in their heads, the less likely their role will exist on an hourly basis in the industry.

So, if anyone wants AI to give us another day off, we need to think about how it can reduce the cost of "context switching" a whole person on and off a task, without simultaneously formalizing our roles so much that it gives us all five. ;-)

whatshisface··on I think Anthropic and OpenAI have found product-market fit
That is an estimate of the relative cost of one training step, but you have to multiply it by the number of training steps, an unknown quantity.
whatshisface··on I think Anthropic and OpenAI have found product-market fit
Inference has traditionally been far less expensive than training. One public example is the fact that hobbyists can run StableDiffusion ($600k training costs[1]) on their personal computers.

Speaking to your point, inference being dramatically less costly than training would not be seen as a delta from the norm. The model of providing inference for anything near the operational costs (like a utility would), would the delta from the norm if it were true.

[1] https://x.com/emostaque/status/1563870674111832066

whatshisface··on I think Anthropic and OpenAI have found product-market fit
Here are a few thoughts:

- The publicly available information about how inference costs compare to training costs is conflicted. EEs involved in datacenters talk about power usage spikes during training runs as if they were a major factor in the designs, but academic papers discussing cost-optimal scaling confidently treat inference-time compute as a major factor.

- On the side of the balance indicating that training is more compute-intensive after amortization than inference is that Chinese providers, constrained primarily by access to compute, have nearly unlimited token availability at a lower price than US providers (inference), but poorer model capabilities (training). That would make sense only if US providers are inflating inference costs by 20-30x due to amortized training costs that overseas providers were not able to take on (there are other factors too).

- If training >> inference, they're in a prisoner's dilemma that far exceeds the ordinary zero-marginals model of competition between firms (due to its huge discrete stepwise nature). On the other hand, if inference>>training, the high-level analysis popularized by certain thought leaders, that it's like a utility, would be true. You'd tend to count this as a vote for inference>>training, but the CEOs saying it at least have a huge incentive to agree because the alternative, the prisoner's dilemma, would stop investment very fast.

- The only voice in the story that I just told you to have anything to do with fact (as opposed to high-level analysis and ivory tower armchair management of a secretive business) were the rumors from facilities engineers. That shows you the state of our understanding...

- If we don't even know the ratio between amortized capital expenses and operational costs, outside investor analysis is impossible. It doesn't matter how finely they divide the accounting buckets for office ferns and indoor ferns if the single biggest part of their business is obscured for trade secret reasons.

whatshisface··on I think Anthropic and OpenAI have found product-market fit
There were no knowledge workers in the middle ages.
whatshisface··on Magnifica Humanitas
"Ye have heard that it hath been said, An eye for an eye, and a tooth for a tooth: But I say unto you, That ye resist not evil: but whosoever shall smite thee on thy right cheek, turn to him the other also. And if any man will sue thee at the law, and take away thy coat, let him have thy cloak also. And whosoever shall compel thee to go a mile, go with him twain."

"Now he that betrayed him gave them a sign, saying, Whomsoever I shall kiss, that same is he: hold him fast. And forthwith he came to Jesus, and said, Hail, master; and kissed him. And Jesus said unto him, Friend, wherefore art thou come? Then came they, and laid hands on Jesus and took him. And, behold, one of them which were with Jesus stretched out his hand, and drew his sword, and struck a servant of the high priest's, and smote off his ear. Then said Jesus unto him, Put up again thy sword into his place: for all they that take the sword shall perish with the sword. Thinkest thou that I cannot now pray to my Father, and he shall presently give me more than twelve legions of angels? But how then shall the scriptures be fulfilled, that thus it must be? In that same hour said Jesus to the multitudes, Are ye come out as against a thief with swords and staves for to take me? I sat daily with you teaching in the temple, and ye laid no hold on me."

whatshisface··on DeepSeek makes the V4 Pro price discount permanent
Producing great products is a game at which every player wins, because sellers must find willing buyers. It only fails if one participant panics and jumps out of the window, or if a significant number of people are not participating (this is always the case when wealth inequality is involved).
whatshisface··on How to convert between wealth and income tax
Yes, even if it never is. But you have to realize that these proposals to close loopholes are massive tax increases, not technicalities.
whatshisface··on How to convert between wealth and income tax
You can get a loan from a bank against your assets and spend only the loan. I think that's how it goes to zero.

>So it's in theory possible that someone could make millions tax free from selling a business, but that's a rare case and one the tax code explicitly allows for.

These guys have whole offices working to figure out how their day to day expenses can be converted into stuff like "selling a business." ;-)

whatshisface··on How to convert between wealth and income tax
A much more interesting formula would be how to convert between income and income tax - you'd think it worked according to the superficial bracket system, but in fact, it works along the lines of going to 0 at the top.

P.S. a wealth tax is a property tax. They have existed in the US since before the income tax (which was originally considered unconstitutional by its opponents).

whatshisface··on Seattle Shield, an intelligence-sharing network operated by the Seattle police
If not knowing who somebody is weighs into your consensual model of interactions with them, just wait until you hear about surveillance! :-)
whatshisface··on The Letter S, by Donald Knuth (1980) [pdf]
As Knuth points out, applying math to art is applying artistic sense to a continuum of forms all at once.
whatshisface··on Meta blocks human rights accounts from reaching audiences in Saudi Arabia, UAE
The idea that they would ban their propaganda networks, but not their alternatives, is really baffling...
whatshisface··on College students drown out AI-praising commencement speeches with boos
I'm sure VPs are sweating bullets over the instructions they will receive during the part of the shareholder's call where mutual fund managers dial their members and hold the phones up to each other.
whatshisface··on The last six months in LLMs in five minutes
Waterfall was famous for wasting developer time and extending delivery dates in exchange for simplifying management. If Claude time is comparatively inexpensive, but human oversight remains necessary, we will switch back to waterfall because the relative importance of the two resources will invert.
whatshisface··on At least 25 Flock cameras have been destroyed in five states since April 2025
To be fair to the loyalists, a lot of people were making this point at the time. Tally ho, gents.
whatshisface··on At least 25 Flock cameras have been destroyed in five states since April 2025
What's going on here is that out of 100,000 constituents, three know that Gary's Carpets is licensing the city reservoir as a PFOAs dump, and combined they have $1,000 of advertising reach if in-kind contribution is counted. Meanwhile, Gary's Carpets has a $60k advertising budget annually, donates to all five churches, subcontracts influence operations with bot farms, and attends weekly meetings of the grand lupus lodge.
whatshisface··on AI is a technology not a product
The whole point of AI is that if something different happens, it's not you doing it.
whatshisface··on The sigmoids won't save you
I'm not asking for anything close enough to the boundary for questions like that to be difficult. There are some ML systems like AlphaGo that have crossed the line in specific domains. It's just that making self-play and online learning work for huge LLMs is highly non obvious.

The idea is simply that the basic idea behind LLMs, that you're distilling the entropy out of the entire available world of text, is antithetical to creativity.

Further developing on the theme of self-play, humans have the ability to sense what we want (intellectually) and reach for it communally over thousands of years. It's an innate quality, and if AI starts participating (contrast to giving people psychosis) we will all be able to tell.

whatshisface··on The sigmoids won't save you
If you want a model, here's one: LLMs have never demonstrated the ability to go obviously beyond interpolating their training data. It takes an army of paid data producers solving homework problems to give ChatGPT the ability to do your homework. All vibecoded apps that turned out to be successful could put on a geological soil chart with other apps, probably on GitHub somewhere, on the corners. The prediction? They won't.

In this model, the exponential growth that everybody is freaking out about is only the realization of the modular software dream ("we'll only have to write an ORM once for all of human history!") and the sheer amount of knowledge in libraries.

It's at least falsifiable.

whatshisface··on The AI zombification of universities
This is one of those things that seems true in theory but doesn't have examples of success. There are people who self-study advanced mechanics at home, but they're child prodigies and enjoy (typically) going to universities as a way of meeting others who share some interests.
whatshisface··on Twin brothers wipe 96 government databases minutes after being fired
I know this is not a good year on the job market, but if you are traveling to work with a "go bag" and not leaving coffee mugs on your desk to prepare for being laid off maybe it is time to carry that go bag to some other buildings...
whatshisface··on UCLA discovers first stroke rehabilitation drug to repair brain damage (2025)
In this example, they're all different diseases.
whatshisface··on Students boo commencement speaker after she calls AI next industrial revolution
Quality standards benefit the buyers, who on the face of it are corporations replacing representative's constituents with capital investments not made in their districts, not the constituents themselves. I would expect there to be support for limiting quality somehow, maybe by enforcing copyright law (actually legal training datasets are a lot harder to obtain). Regulatory burdens do obstruct commercial investment, but the burden primarily falls on small participants who in the AI market are far from posing a threat to employment. Only the largest firms can threaten the wage and salary model of organization.
whatshisface··on UCLA discovers first stroke rehabilitation drug to repair brain damage (2025)
I think there's some kind of fallacy where you can look at five drugs, all of which came from a pool of 100 promising candidates, then look at the next 100 candidates and say for each one individually that it is not worth celebrating. I call it the, "rounding to zero" fallacy.

In reality, if you have 100 5% chances of a cure for a previously incurable illness, you can celebrate each chance a lot.

← PreviousPage 4 of 34Next →