robryk @robryk@qoto.org

#lazytoot

Do you know of anyone trying to make any kind of a MastodonAPI-to-{NNTP,Maildir} bridge?

@cjwatson @richardelwes

In that case you can make that world arbitrarily close to our own by making those truth-detectors arbitrarily computationally expensive.

@midwan What would you do differently if you didn't try to keep the look consistent?

(The reason I'm asking is that I think we've changed the way we design UIs in ways that I find hard to enumerate, and such examples would (a) make them easier to notice and enumerate (b) possibly find an interface that's actually better. An example of a difference, though IIUC not that much with Amiga, is the vanishing use of menus for anything other than switching between totally independent pieces of the UI.)

@cjwatson @richardelwes

> and a practical method to determine whether any statement about arithmetic is true or false.

I understood this as meaning "polynomial-time algorithm that determines whether a given statement expressed in arithmetic is true/provable (which would be the same in that world)". If so, NP=P, because you can ask your truth-determining system whether a witness exists.

Note that even if we relax the requirement to "proof-finding algorithm that finds ~shortest proof and is polynomial _in output size_", we still get P=NP by those means.

> but there could easily still be problems that take O(2^n) or worse time.

On the face of it, obviously EXPSPACE is still strictly larger than PSPACE (and EXPTIME is still strictly larger than P), so this should be true. OTOH, can't we describe some EXPSPACE-complete or EXPTIME-complete problem in arithmetic using a description sized polynomially in its input? I'd expect yes, and if so this provides yet another proof why determining truthfulness couldn't be polynomial in _input_ size. (Though this probably wouldn't be such a simple contradiction in the setup where we have an algorithm polynomial in proof length.)

@lebenalsviele@troet.cafe In welchem deutschsprachigen Land suchst du solche Bank?

@richardelwes

There would be ~no cryptography: no asymmetric signing/encryption schemes and no key extension schemes, so the only way to symmetrically encrypt something would be to use a key the size of the message.

Solving real-world problems by reducing them to a SAT instance (or something universal of a similar nature) would be even more popular. That would probably eat all the hard problems in operations research for example.

Approximate algorithms would be a lot less important and probably wouldn't develop as a field for some time: there'd be very little reason to use them (perhaps some polynomial speedup).

Finding "fastest algorithm for problem X" could be also done automatically once we describe X well enough, so a nontrivial amount of effort would move to making such descriptions easy to write in an obviously-correct fashion.

@dalias Which desirable properties? Do you want all percentages to be rounded with same precision? (Is it just as bad to round 0.5% to 0% as it is to round 50.5% to 50%?)

If the answer to the latter is no, then I posit that you want to minimize Kullback-Leibler divergence (i.e. "how many more bits we'd need to use to encode a value chosen from the actual distribution if we incorrectly believed it came from the rounded distribution"). This implies that your absolute precision should shrink as 1/sqrt(value) (so, your relative precision should grow as sqrt(value)).

After this choice is made the actual problem is something I'd want to know a nice answer to. When I was solving the ~same problem, I ended up with annoying heuristics that pick one entry (one with the largest value) as the one that will just be "100% minus all the others".

@brewsterkahle ActivityPub supports ~full html in posts. Mastodon is able to display that. There are forks of Mastodon (e.g. the one used by my instance) that allow users to create posts at the very least using markdown.

@galinash @mitchdenny @timnitGebru

> which also makes it quite likely he doesn't have industry experience (it's unusual to get hired for a tech job without at least a college degree).

While this might be correct inference, using such an inference reinforces it in the world. I'm not sure what to do about it in general other than be aware of it, and of how one uses the result :/

@anthropoid@101010.pl @ZayebistyMariusz@pol.social @AubreyDeLosDestinos

Poza tym rozróżnienie między defederowaniem a wyciszaniem jest zwykle pomijane.

@homelessjun @gregeganSF

I wonder if there's often some sort of game of telephone happening where "reacts badly to X" gets transformed into "is allergic to X".

@peterdrake @freemo @Gargron @stux @QOTO @trinsec

Re 5: I would expect that the reasonable approach to that is silencing, not suspending.

@kwf I was under the impression that solagg.com used to do this, got blocked by ~everyone and for some reason didn't try to domain-hop. I can't find the concrete statements I remember seeing about what it was doing precisely, though.

@rst @matthew_d_green

I was just surprised to learn that SpaceX has only ~2x the employee count of Blue Origin (according to wikipedia).

I expected to see employee count of BO to be similar to employee count of SpaceX at the time of first successful Falcon 1 launch. I can't easily find the latter value, but now I think it's unlikely.

@dalias It seems to me that all of these usecases would still work if they had cc semantics similar to replies or to boosts (i.e. by default/always cc the author of the quoted message/the cc list of the quoted message).

(Note that cc lists for replies are already weird: it's easy to accidentally send a reply that isn't actually sent to the author of the message you're replying to using some clients. Mastodon-the-service considers such replies to be valid enough to accept from users and federate outward.)

@agnieszka Wouldn't having it be obeyed be a kind of "giving consent via browser settings"?

@agnieszka

Would it be wrong to say that explicit values of `DNT` constitute user's statement of will?

I'd dearly hope we don't say that websites are required to prevent the user from using automation to give/refuse to give consent.

@k0rnelia_01 Ten podział nie obejmuje wszystkiego i też jest dość dziwny w niektórych okolicach.

Z rzeczy które są nim nie objęte/są nietypowe:
- "systemy wbudowane" (czytaj rzeczy typu Arduino, ale też typu pico8),
- rzeczy blisko systemu operacyjnego (imo dość ciekawym jest zrozumienie czego tak naprawdę potrzeba w systemie operacyjnym, żeby dojść do momentu gdy np. można wysłać e-mail),
- w ogólności aplikacje które nie używają/potrzebują sieci.

@rysiek @briar

Ah, I see, that does make sense.