IsTom

When you're picking most likely tokens, you get least surprising tokens, ones with least entropy and least information per token.

If you start with safety in mind and don't just try to bolt it on, you're in a much better place. With the kind of code you need in typical applications you could force vast majority of it in a shape that passes termination checks in theorem provers without much overhead, especially if you can just put gnarly things in standard libarary and validate (with proofs hopefully) once.

Though starting with C/C++ is a losing proposition in that regard. And I guess any kind of discipline loses to just throwing half-baked javascript at wall, because deadlines don't care about bugs.

A lot of software doing useful work halts pretty trivialy, consuming inputs and doing bounded computation on each of them. You're not going to recurse much in click handlers or keep making larger requests to handle the current one.

If you look from far enough, it becomes "Current world ⊨ I am typing right now" which becomes tautological again.

I wonder what's the EROI on building a tanker with 2% chance of being hit each time. They hold a lot fuel, but making them can't be light on energy.

Oil going through Hormuz is 20%, not 80% of global supply. It's true that demand is pretty inelastic, but it's not like it can't be cut at all.

Until courts start ruling that copyright applies to LLM scrubbing I think we're all sadly shit out of luck.

The most difficult thing about this was controls being unresponsive (at least on firefox).

He's very efficient with prose and I find it a joy to read (well, given what he's writing about it's not always joy, but still). I'm not sure he's following that rule 100% of the time, but it's close. Depending on the setting, you can often describe it through characters' actions or how it shapes them.

I've seen some juniors writing risoni code like that. They've heard that you shouldn't write big functions, so obviously they forcefully split things until they can't be split anymore.

Does Java not have some kind of helper function to join strings? Why'd anyone write loops like that?

Just look what kind of problems the easy task set is (hello world, echo line, count vowels, etc.). With best being ~10% of total in brainfuck this is 10 out of 20. You can google more solutions to these problems than that.

They are different, but on some fundamental level when you're writing code you're expressing an idea and it is still the same. The same way photograph and drawing of a cat are obviously different and they're made in vastly different ways, but still they're representations of a cat. It's all lambda calculus, turing machines, ram machines, combinator logic, posts' dominoes etc., etc. in the end.

> empirically demonstrate the code's behavior

That is completely insufficient for code of any real complexity. All this does is replacing known bugs with unknown bugs.

> Forget to set a timeout on a gen_server:call?

Default timeout is 5 seconds. You need to set explicit infinity timeout to not have one.

I get the other things, but believing Altmans's words is not high on the list of things to be considered evidence.

If you take AI image (that cannot be copyrighted) and adjust it in photo edition software of your choice then the changes are potentially copyrightable and the resulting image can be copyrighted (you need to ensure that your changes pass the low bar of creativity).

It's not clear to me how much code you would need to modify by hand to qualify for copyright this way, but that's not an impossible avenue.

Or just make time for more Very Important Meetings.

There's plenty of services that 3D print for you.

I'm not following the whole LLM space, but

> the compute needed to perform matrix multiplications goes up as the cube of their size,

are they really not using even Strassen multiplication?

https://en.wikipedia.org/wiki/Mannion_v._Coors_Brewing_Co.

I remember there being a somewhat similar case in (I think?) Netherlands a few years ago, but currently can't find it.

It's not copyrightable automatically, you have to argue that you did have an artistic input (e.g. composition). Typically nobody bothers to argue against copyrightability of a photo, but there's been a few cases.

They're waiting for claude to get up so they can use it to investigate why claude is down.

Sending message to a process has a cost (for purposes of preemption) relative to the current size of receiver's mailbox, so the sender will get preempted earlier. This isn't perfect, but it is something.

> actors still deadlock through circular messages

I've rarely seen naked sends/receives in Erlang, you mostly go through OTP behaviors. And if you happen to use them and get stuck (without "after" clause), the fact you can just attach a console to a running system and inspect processes' states makes it much easier to debug.

I don't think killing democratic representatives has as big of an effect as killing authoritarians. You can't have cult of the leader without the leader, but in parliamentary systems you'd have to off quite a few people.

> The DoW probably disagrees with that interpretation

Or perhaps, maybe, just a little maybe, DoW is getting absolutely excited about mass surveillance and kill-bots?

The only more fitting name currently would be Department of Peace

There's two issues with that. Sometimes people publishing sheet music alter it and then they can claim copyright. Second is that that old original sheet music can be illegible to modern musicians – it has evolved over time.

And staring directly at the sun is not recommended.