kweingar

When Altman says "OpenAI is not for sale", is there any sense in which that is not a lie?

As part of the for-profit transition he is selling all of the non-profit's assets. Musk's group has offered to buy them, and Altman has declined, but he is still planning on selling them, no?

The transition involves a sale of the non-profit's assets. This is what Elon has offered to buy.

They can refuse Elon's offer, but they can't turn around and sell the assets 50% cheaper to someone else (as they had previously planned).

OAI's fiduciary duty is to their charitable mission. If selling to Elon for $97b jeopardizes the mission, then so would selling to preferred investors for $40b (as they would in turn face immense pressure by shareholders to realize an instant $57b gain by reselling to Musk!)

That's what this whole thing is about. Musk's offer will probably not be accepted, and he knows it. The purpose is to throw a wrench in OAI's plan to sell to insiders at a heavy discount, possibly making it impossible for the nonprofit to become a for-profit.

If this offer forces the preferred investments to cough up an extra $60b for the nonprofit, that's fantastic for the nonprofit mission.

Regardless of how you feel about the potential consequences of Musk being a significant shareholder of OpenAI, the alternative is not to let sama and other preferred investors buy the non-profit's stake at a >50% discount as they had planned to do before this offer.

Not respecting precedence is ok, as long as the user is aware of it.

When I use a calculator, I don't necessarily care if I can punch in a written expression verbatim from left to right. I just want to do a calculation.

Elon Musk promised to build this tech.

The MUMPS database is wild. When I was working in MUMPS, it was so easy and fun to whip up an internal tool to share with my coworkers. You don't have to give any special thought at all to persistence, so you're able to stay in the flow of thinking about your business logic.

But as you said, the language itself is almost unbearable to use.

I'd love to use HTMX at work. Sadly the security folks would probably balk at checking in JS code that uses eval(), even though you can disable eval at runtime in the HTMX config.

I thought about writing a script to remove all references to eval (and all tests relying on eval), but at that point it would probably be easier to just rewrite the library.

Taiwan has an official policy that it owns the mainland

> only OpenAI was able to productize

What do you mean? Anthropic and Google both have widely used products based on transformers.

This isn't really an argument though. It's an assertion that not honoring a commission agreement (or an employment contract) is equivalent to not paying for a license to an existing work. I tend to disagree. I could be persuaded otherwise, but I'd need to hear an argument other than "clearly these are the same thing."

The success of the Chrome browser on desktop proves otherwise, no?

It's interesting that the argument is "nobody can compete with defaults" when one of the proposed remedies is to break off the part of the company that was too successful at competing with defaults.

I like Go's error handling. The reader knows exactly which lines of code can encounter an error, and exactly how the error is handled.

I find that exception-based code is much harder to read. The happy path is clearer, but exceptional code paths are often completely obscured. It's harder to reason about what state the program is in when the exception is handled.

It's good to get an uneditorialized view so you can know what's going on without wasting your breath making the same complaints over and over.

But that doesn't mean the complaints are wrong. SCOTUS is an insane, dysfunctional institution.

The premise of judicial review is that if you have ideologically-aligned justices, then any petitioner who disagrees with Congress or the executive can file a lawsuit, hit the appeal button twice, and get SCOTUS to undo the law. The amount of law that is supposedly derived from the (tangentially related) terse sentences of the constitution is immense and, imo, completely unjustifiable.

The court's ability to strike down 100-year-old laws (whether acts of Congress or their own previous decisions), and their ability to effect these changes at times when the legislature is in a more or less favorable position to react, is destabilizing.

Even if this isn't convincing, just look at the farce that is lifetime appointments. Large swaths of our ways of living are dependent on the health, age, and egos of these people. We have had manifestly unfit justices cling on to the bench in hopes that they can retire under a friendly president. We can do better.

It seems incredibly convenient that a non-profit's leaders can say "I want equity in a for-profit company, so we will sell our assets to investors (who will hire me) and pass off the proceeds to some other non-profit org run by some other schmuck. This is in the public interest."

Can anybody explain how this actually works? What happens to all of the non-profit's assets? They can't just give it away for investors to own.

The non-profit could maybe sell its assets to investors, but then what would it do with the money?

I'm sure OpenAI has an explanation, but I really want to hear more details. In the most simple analysis of "non-profit becomes for-profit", there's really no way to square it other than non-profit assets (generated through donations) just being handed to somebody for private ownership.

My contract mentions nothing about time spent.

This doesn't seem relevant to a discussion about chargebacks, a feature that cash famously does not have.

> GPT-2 level

You must not have ever used GPT-2 if you think it is comparable to any commonly used chatbot today

The arguments in HN and elsewhere about monopoly seem to be missing the point imo. My problem isn't even with Apple's 30% take rate!

I use Patreon on the web on a non-Apple desktop computer. Apple is simply not part of the equation in my transaction with Patreon. But because of Apple's App Store policy, Patreon has to remove useful features from the web app. That seems like an abuse of power to me.

If you have Apple goggles on and disagree, I ask you to consider a reverse situation. Imagine that your favorite third-party iPhone apps delete features because some other company in charge of some other platform demands it. Technically the app developer could simply pull out of that other platform, but they do the math and conclude they will lose less revenue by degrading the iPhone app and pissing off iOS users. Doesn't that seem outrageous?

You're not objecting to public transportation, you're objecting to the public. The people on the bus or the train are not any different from the people at the supermarket or at the park.

Many Americans, like myself, are ok with being around people in a city. But you're right that there is a group of fearful suburbanites who would rather sit in isolation miles away from the cities they drive to than risk catching a whiff of B.O.

The states select the president, and in this case, the states agreed to adopt thr compact. Don't see a problem there.

There are zero product managers who read this Twitter thread about a random decade-old Chromium hack and will do anything about it.

That is a good point. I would happily let Trump go free if it means that Biden can delete all student debt without fear of prosecution.

Reading HN is caring about what other people do, no?

> It wasn't that long ago that we considered the brain and a computer to also be quite similar

Many people in tech still think this. It will take a while to dislodge the idea that if you squint enough, cells, brains, etc. are all basically computers, because it flatters the egos of computer programmers.

There are genuine connections between specific biological systems and certain ideas in computer science, but most of the casual analogies you hear from tech people only serve to inflate their own perceived authority ("I know a lot about computers, and cells/brains are like computers, so here is my take about cells/brains...")

This tweet is speculation, there is no evidence presented in the thread that it is A/B testing.

The message says "Sign in to confirm you're not a bot." Given that this message is being presented to Cobalt, an automated downloader tool, it seems likely that YouTube has detected bot activity and is throwing up barriers to stop botting.

That is very different than the implication that YouTube is testing out the idea of blocking human users from watching while logged out.

I'm talking about the copyright claims. If Bing and Google stand down when people request their sites to be removed, how would OpenAI have the resources to push against that?

If all you have to do to beat these claims is throw money at the problem, then why haven't the other (better funded) search engines done that?

Normally if I were recommending Linux to a friend, I would encourage them to use their existing computer rather than going out and buying new hardware just for Linux.

ChromeOS Flex might fit the bill though, and the installation process is probably the most straightforward and handhold-y out of the Linuxes. https://chromeenterprise.google/intl/en_uk/os/chromeosflex/