robryk @robryk@qoto.org
I enjoy things around information theory (and data compression), complexity theory (and cryptography), read hard scifi, currently work on weird ML (we'll see how it goes), am somewhat literal minded and have approximate knowledge of random things. I like when statements have truth values, and when things can be described simply (which is not exactly the same as shortly) and yet have interesting properties.
I live in the largest city of Switzerland (and yet have cow and sheep pastures and a swimmable lake within a few hundred meters of my place :)). I speak Polish, English, German, and can understand simple Swiss German and French.
If in doubt, please err on the side of being direct with me. I very much appreciate when people tell me that I'm being inaccurate. I think that satisfying people's curiosity is the most important thing I could be doing (and usually enjoy doing it). I am normally terse in my writing and would appreciate requests to verbosify.
I appreciate it if my grammar or style is corrected (in any of the languages I use here).
Joined Nov 2020
In a similar discipline, we did lots of hypothesis testing in XIX century physics without having anything resembling a complete model of the objects under observation (e.g. the whole investigation of "cathode rays": someone noticed a glow next to the cathode and then started trying to formulate hypotheses on how the effect that causes stuff to glow can travel from the cathode). Hypothesis testing is very useful during model building: you have a candidate part of a model, and so you pick something to observe that would behave in a very specific way in that model. If it doesn't, you know the partial model's wrong.
It's not the hypothesis for what the problem is, but a sequence of hypotheses that you try to quickly refute. For example, first hypothesis that people usually try to confirm/refute is "there is an actual problem". There are many times afterwards when you do indeed try to look for leads, but:
- they are interspersed with hypothesis testing (it's not that hypothesis testing is concentrated at the end),
- IMO doing prep work to make them shorter/fewer is one of the more useful kinds of prep work, because they rely in some way on luck.
So I agree that it makes sense to concentrate on non-hypothesis-testing parts in any advice disproportionately to the time they should usually take, because they can be helped more by working out patterns to follow compared to the hypothesis-testing parts.
Yes. It's a tool that does that much faster than groups of humans.
@vicgrinberg TIL about https://en.wikipedia.org/wiki/S-process. I thought all nucleosynthesis past iron happened in supernovae.
Why doesn't the larger problem also imply that it's impossible to "make" non-racist humans in this environment?
@rossgrady You don't even need guns: if you can ram the gate (or if there's no gate), you can punch a hole from up close (though hot oil would immediately start coming out, so this might be easiest to do with a gun) or you can pump salty water in via the vent piping (unless those transformers use a membrane to allow for oil's expansion instead being open to air).
media description discourse
@moonbolt The standard social weirdnesses around correcting someone are reinforced here, because a reply can notify other people.
@niconiconi What was on the wheel? Capacitors, or something more interesting?
@TRyanGregory@mstdn.science How large a fraction of all pediatric hospitalizations are influenza-related? (Sadly, one of the graphs has per-capita Y axis while the other one has absolute Y axis, so I can't tell from them alone.)
@b0rk If these are "one-shot" tools (i.e. not really interactive ones, the tool gets a request and spits out some output, you usually amend the request a few times, but nothing depends on previous requests) then I'd try them when I suspect I'll want to have a very good in-depth understanding of what's wrong (and how the parts that were broken work normally).
This is a good opportunity to learn how what that tool says related to real world, because you anyway intend to become very sure about the state of the real world, so will be in a position to notice tool's quirks. If you start using a tool in cases where you will stop at shallow understanding of the world around the problem, you'll have no way to realize that you're using the tool incorrectly.
Why specifically for one-shot tools? Because with interactive tools this approach of slightly distrusting the tool is sadly too impractical.
I'm not sure if this is the right analogy. AP ID is a URL under which you can find metadata and public signing keys for the account. It's an in-envelope concept (to borrow e-mail terminology) and has ~nothing to do with routing (in fact the metadata returned points at your inbox, outbox, and other collections, which can be hosted wherever the handler of that URL desires).
I do agree that it's bad that we have two levels of IDs and different pieces of software consider different levels as persistent.
> I already knew you can't have multiple identity domains (the ones in the username) hosted at the same instance.
This is an implementation choice, not enforced by protocol. Nothing prevents someone from making instance server software that simply handles multiple domains.
@rysiek From some POV, everything we do is pattern-matching. Starting from image understanding, going through imagining things described in text, even up to (a large part of) finding out new things.
@rysiek I'm curious whether you think that it's potentially impossible, or whether existence of such a simulation matters for the question assuming that it's possible, or something else (that we'd learn something interesting while developing such a simulation that could provide evidence/arguments here?).
At which point along the line of "human brain", "simulation of human brain", "simulation of human brain executed by hand by humans" would you place the line?
What I was mostly disappointed by is silent assumptions, not the opinion.
@rysiek I'm somewhat disappointed by that article: it doesn't mention that one can think of a human brain as a Chinese room, and asserts that the whole system of a Chinese room doesn't "understand" Chinese without argument.
I enjoy things around information theory (and data compression), complexity theory (and cryptography), read hard scifi, currently work on weird ML (we'll see how it goes), am somewhat literal minded and have approximate knowledge of random things. I like when statements have truth values, and when things can be described simply (which is not exactly the same as shortly) and yet have interesting properties.
I live in the largest city of Switzerland (and yet have cow and sheep pastures and a swimmable lake within a few hundred meters of my place :)). I speak Polish, English, German, and can understand simple Swiss German and French.
If in doubt, please err on the side of being direct with me. I very much appreciate when people tell me that I'm being inaccurate. I think that satisfying people's curiosity is the most important thing I could be doing (and usually enjoy doing it). I am normally terse in my writing and would appreciate requests to verbosify.
I appreciate it if my grammar or style is corrected (in any of the languages I use here).
Joined Nov 2020
