Hillel

Like there will be a "load file" button and there's no event or javascript wired to it, it just sits there. Stuff like that

I would be slightly less mad about the slopification of SPA tooling if any of them actually worked. But none of the buttons *do* anything

I did not! My bad

it's an open question! we don't even know if P is a strict subset of # P, or # P of PSPACE (both conjectured)

This is actually... not the worst thing so far

(PSPACE is the set of all problems that can be done with efficient use of *memory*, even if they are inefficient in *time*. You'd expect that set to be larger, as you can reuse memory but not reuse time.)

Also the weirdest one is that BQP is conjectured to be a strict subset of PSPACE, but it's not proven.

The weird thing is we haven't proven that P is a strict subset of PSPACE, either.

That's just weird, man.

And BQP is conjectured to not be in NP, but not be a superset. So there are some problems that quantum computers possibly do exponentially faster than classical ones, but also many problems they're no matter at. Which is pretty neat!

To me the interesting ones are the relationships between the categories. We suspect, but do not know, that P is larger than NC, meaning there are some efficient algorithms that cannot be made 10x more efficient with ten computers instead of one. It's a formalization of sorts of Amdahl's law

NC is DPs where a REALLY efficient *parallel* algorithm exists. VERY roughly speaking, if a problem is of size N, and you get N² processors to solve it, can you do it in logarithmic time?

(actually polylogarithmic, like log(n)³ and such, but that's a technical detail imo)

So how does this related to NC and BQP?

BQP is the "easy" one: it's BPP, except instead of a deterministic computer, you get a quantum computer. This opens up considerably more useful algorithms, including some conjectured to not be in NP (efficient "lucky guesser" algorithm)

BPP: DPs where we know an efficient algorithm, but the algorithm sometimes get the wrong answer, but it gets the right answer more often than not.

That's tricky! Historical examples are primality testing, but nowadays we know deterministic efficient primality tests

I'm going to blatantly ignore this prompt and instead talk about P, DLOGTIME, and BPP.

P: Decision problems (yes/no answers) where we know an efficient algorithm (grows polynomially with size of input)
DLOGTIME: DPs where we know a REALLY efficient algorithm (grows logarithmically with input)

This would require me to know anything at all about the topic D:

Fuckit gonna try StickyKeys for a while and report back

Bugs that literally cost money How much does a software bug cost a company? Well that's a messy question. It depends on the type of bug, the broken behavior, the type of company, etc. And...

One of the less cool things is when @leanpub.bsky.social has a bug that makes the coupons activate at the wrong date

Another one to the Bugs That Literally Cost Money List buttondown.com/hillelwayne/...

This sounds great! Get your @leaflet.pub drafts ready!

> The idea is pretty simple: on April Fools' Day…a participant produces genuine content that's very different from their normal produced content. It could be a different format, a different topic, a different style, anything.

I'll do it tomorrow

I have somehow been bestowed a SECOND jar of cream of tartar

@arachno.capital @mat4nier.bsky.social halp what recipes use, like, a *lot* of cream of tartar

Which is how we know Prolog influenced Erlang :D

I wondered why so few languages adopted the Prolog style of writing functions as f/arity, like add/2 or exit/1. Just now I realized the reason: most languages don't wildly change the behavior of a function based on arity. Prolog's partition/4 is quite different from partition/5!

Might not skeet about all of them today, gotta space that content out

I gotta lot of tasks this week I need to procrastinate, Gimme software topics to skeet about

Five more go up later today! Maybe next time I should give the first batch an hour delay so people don't get them all while the newsletter is still in flgiht

One of the cool things about being self published is I can give out free copies of the book whenever I want

All data from here: www.eia.gov/electricity/... As far as I can tell (but haven't looked to deeply), this far exceeds the increased electricity demand from AI

Here's a tiny bit of hopium:

In 2024 we were planning on adding 160 GW of renewable energy to the grid by 2030.

As of the start of this year, we had added 100 GW and are planning on adding 220 GW more by 2030.

(In 2022 we planned for 60 GW in 2030. We hit that by the end of 2023)

Trump administration drops suits against law firms with ties to Democrats and other Trump foes | CNN Politics The Trump administration has decided to drop its prolonged court fights against four law firms with ties to Democrats, after it had sought and failed to cut out the firms’ access to the federal govern...

Take our wins where we can www.cnn.com/2026/03/02/p...

I feel so scared and hopeless these days. I'm not going to sugarcoat my feelings with irony or jokes or caveats about a better future. I'm just getting out the fear I carry in my marrow. I'm just scared.

April Cools' Club

It's March, which means it's one month to April Cools, which means time to start writing *something* www.aprilcools.club