jdm2212

When your job pays you a dollar, you get an actual dollar. Your job never claws that money back. Unrealized capital gains are not money, and they routinely get clawed back. Unrealized capital gains just means someone in New York traded GOOG/GOOGL at a higher price than they did yesterday.

And, sure, super-rich people can in theory use the appreciated stock as collateral for loans and not pay taxes, but in practice Larry and most other centi-billionaires actually sell loads of stock and pay a lot in capital gains taxes because having your status as super rich dude who owns a huge yacht be totally dependent on Google's stock price is a dumb risk to take on, and it's worth paying some taxes to eliminate that risk.

You sort of can make more land -- by building tall buildings. Unfortunately the Bay Area has mostly outlawed that, too. And we can't blame that just on homeowners. SF "tenant advocates" are just as violently allergic to developers building new structures as homeowners are. But the worst offenders are definitely Peninsula and South Bay homeowners.

Then you'd have two (or three, or four) companies that Larry owns 5% each of that are collectively worth $4T.

You can't solve a shortage with demand subsidies. You need to expand supply. SF has too little housing relative to its population, and has perennially tried throwing money at the nonprofit industrial complex which has (obviously) perennially failed to solve the problem because it doesn't generate new housing units, it just bids up and reshuffles the existing ones (and steals a lot of money in the process).

Larry has sold tons of Alphabet stock and paid lots of capital gains taxes. This is easily available public information. The whole buy-borrow-die thing is sort of a stupid myth. Actual centibillionaires diversify because the risk of having a huge concentrated position is much greater than the liability of having to pay some capital gains taxes.

Google doesn't have much power. It can't arrest you or pass laws or vote. It just happens to produce a lot of profits for its shareholders (who are, overwhelmingly, average people with 401Ks) and a lot of profits means a big market cap.

If we need revenue to fund useful government programs, great, let's tax Larry. But I don't understand what problem is solved by expropriation qua expropriation.

Sure, yeah, but which of the two numbers I was multiplying together is, in your mind, too big, and should be made smaller, and by what mechanism?

Larry Page owns about 5% of Alphabet, which is worth $4T, so he has $200B give or take. Which part of that do you think reflects a "broken tax system"? Companies should get kneecapped if their market cap gets too high? Founders shouldn't be allowed to keep even a single digit percent of the company?

Their business model is based on making money from LLMs. That is some mix of (a) selling API access to businesses and (b) making money from ads / referral fees / whatever from consumer usage.

The businesses API usage will disappear overnight if OpenAI doesn't honor its contracts with businesses.

The actual money will be made because the driverless car manufacturer is self-insuring and cutting out the insurance company. Competition among manufacturers will then pass most of those savings on to you.

Similarly, many car makers offer heated seats without a subscription fee, because of the magic of competition.

I've worked for three different companies that all use some mix of enterprise agreements, self-hosting, and AWS Bedrock to ensure that proprietary data doesn't get trained on. It's not a difficult or complicated problem.

You are, again, flatly wrong about this. ZDR involves infrastructure separation that prevents any data flowing back.

Insurance will just be bundled with the self-driving car software. Insurance and car are separate today because driver and car are separate today. Once they're the same, there's lots of money to be saved by bundling the sale.

This depends on local defamation laws. It's an issue with all online reviews in countries like Germany where the onus is on Google to establish a review isn't defamatory.

This is flatly false. ZDR means the customer request is processed and nothing about its content remains on the model provider's servers.

Nothing is certain in life, but huge corporations very rarely engage in flagrant, intentional breach of contract at all, and particularly not with counterparties who have good alternatives available. If OpenAI turned out to be training on ZDR queries, they would as a practical matter have to retrain their models from the ground up on uncontaminated data (which would cost 9 or 10 figures), possibly pay considerable damages to counterparties, and offer a detailed accounting of what went wrong and probably fire a bunch of senior leaders.

By Buckmaster's own account, OpenAI (a) wanted him and Alpoge to publish their Euler result first and (b) wanted Buckmaster to author the Navier-Stokes paper, and (c) when he refused both of those, tried to get Alpoge to talk him into it.

Why would OpenAI go burn another $20 million trying to prove they didn't accidentally train on synthetic data derived from Buckmaster's codex sessions? They're desperately trying to give him credit anyway!

It produces "high level summaries" so it's perhaps legally akin to note-taking rather than audio recording?

The available information is essentially that story.

OpenAI's account: they heard a rumor that a Millennium Prize problem had been solved, so they tried to do it themselves and succeeded. Then they contacted the other researchers and were surprised to discover those guys hadn't actually cracked it, but offered the one of them who's not an Anthropic employee a co-authorship anyway. The conversations got testy.

Buckmaster's account: totally unsubstantiated accusations of plagiarizing from chat logs and plainly false accusations of OpenAI trying to get Alpoge removed as coauthor of a thing he was not an author of in the first place, and threats to ruin people's careers.

I think the synthesis is basically that Buckmaster and Alpoge had not quite solved the Navier-Stokes problem yet but thought they were really close, and had told friends as much, which is how the rumors got out. Now they're mad they got scooped. They aren't getting the money and recognition they thought they had locked down, and are engaging in a smear campaign.

The boardroom will absolutely veto a decision like "let's give all of our proprietary data way to a company that will use it to train a competing product". That's why zero data retention exists, and why it's corporate suicide to not do it correctly.

Seems like OpenAI did a boring normal corporate thing (find out your competitor made a breakthrough, try to replicate it) and then when the other mathematicians found out OpenAI had beat them to Navier-Stokes, they decided to lie about what happened because they were upset they didn't get to make the big breakthrough themselves.

They released a statement on how they solved one of the biggest open problems in math. That statement includes a narrative of how they decided to tackle the problem. Idk what seems retroactive and contrived about, like, "we thought our archrival had made a huge breakthrough and wanted to make sure we're not falling behind".

Why would it even be bad for OpenAI to try to beat Anthropic to cracking one of the 6 biggest open problems in math? Like who is getting punched?

Levent Alpoge is an Anthropic employee with access to Anthropic compute who routinely throws that at problems and then posts the results on Twitter. They thought they were competing with him! And they in fact were. It's just that in this case he supposedly was acting in a personal capacity, even though he was using internal Anthropic models for some of the work.

It'd be corporate suicide for them to be caught violating zero-data-retention commitments. But also if you're really paranoid you can just use ChatGPT on Azure or AWS, where nothing is flowing back to OpenAI at all.

RE item 2 -- in lab science, it is entirely normal to have competitive-verging-on-adversarial relationships between labs racing to get results first. Mathematicians apparently need to get used to the idea that math is a lab science now.

And that is a good thing! Competition moves us forward much faster than sitting on results to avoid hurting someone's feelings.

Why shouldn't they advance math research and test the limits of their models?

It is worth actually reading OpenAI's response, which is basically that they were trying to see if their models could do what Anthropic had already done, and were surprised to discover that (a) Anthropic had not done it at all, and (b) in fact no one else had done it yet. And then they didn't want to give an Anthropic employee coauthor credit for work that OpenAI had done, which idk seems pretty fair?

The most sympathetic interpretation is just what OpenAI actually claims: they thought the other guys already got there, wanted to see if OpenAI could do it too, and were surprised to discover the other guys hadn't gotten there yet.

No, the other guy is just wrong. Capital and labor are complements. If one is plentiful (capital) and the other scarce (labor) then the scarce one gets bid up.

You pay the one technician $200k instead of $80 because that technician has an outside option that will pay him $199k. He has that outside option because everyone is automating their warehouses and factories and stuff, and so in aggregate we basically wind up with 6 times as many warehouses and factories each employing 1 person.

It's a little more nuanced. There is not infinite demand for warehouses. So some of what happens is an $80k/year warehouse worker becomes a medical technician making $90k/year providing non-automatable services to the warehouse guy making $200k/year -- but the general story is everyone's wages go up. And history bears that out.