China has proxies that sell cheaper access to frontier models in exchange for permission to train on your data.

Right.

As an example: Oracle and PostgreSQL don't have dirty reads: READ UNCOMMITTED does nothing. MySQL's concurrency model depends on the engine.

What keeps users themselves from vibe-coding their own software?

If you don't own quality, why should I pay? You're just a middleman at that point.

If you squint hard enough, you'll notice your bank account serves as an IPC semaphore replenishing API credit balances.

Your wallet is now a real-time communication channel.

There was the commercial of a bank using mobile banking as a chat system by sending 1ct back and forth and using the description.

I recall some friends did that in colleage. Obviously with a bit of teenage humor in money transfer descriptions... which is not really funny, since there is no way to remove the transaction ever. Good times.

_RealShipOpts has a constructor.

With consent, yes.

Is there any indication Railway has consented to such disclosure?

But Railway has been blaming GCP for the outage. Shouldn't GCP be given an opportunity to defend itself?

Railway hasn't placed any blame that I've seen. They've posted a timeline of what happened, without any speculation on causes.

Is that what you'd want your vendors to do?

No, that’s what the money is for.

No, "virtual machine" alone doesn't make things safer.

Shrink your attack surface.

Use a completely locked down seccomp. Use nsjail or gVisor for containers. Use microvm or libkrun for full OS.

Lesser attack surface is what matters. Virtualization is only half of the story.

Avoiding problems outside your business problem domain, and can therefore guide the AI more effectively towards building the right thing.

Making things fit for purpose is not avoiding problems in my book.

If anything, quite often it's introducing more problems, because we know we'll run into them and they need to be addressed.

AI is sometimes quite lazy and refuses to solve the hard problems (sometimes making funny excuses like it would take weeks) until you make it explicit that it's important they are dealt with.

(Note that this still assumes that each biased-coin toss is i.i.d.)

They are orthogonal.

Suppose you have 100ms audio latency and no wait time. Then, natural pause will trigger response immediately but you won't notice it has started until after ~200ms (round-trip time). Twice as annoying.