Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[silverspawn]

ironically the submissions look a bit like they're generated by GPT-3.

[silverspawn]

Deepmind supposedly released a language model better than GPT-3, but I don't know if normal people can access it

[silverspawn]

Ordered a new headphone because I need a microphone. Been using (very expensive) in-ear headphones that I bought for my phone for a while.

It's noticeably better. Not sure about the bass yet (the in-ear headphones had an unbelievably good bass, didn't even know that was possible), but the rest is better. I can hear details I couldn't hear before.

Now listening to music is even more enjoyable. Related:



Presently I think of Kashiwa Daisuke as the outright best songwriter I've ever known.

[silverspawn]

Whelp the class average is 10.7, a whopping 7 points below the 17.7 of the other biweekly

[silverspawn]

Almost any effort will serve to convince us that we have "tried our hardest", if trying our hardest is all we are trying to do.

[silverspawn]

That is such a good quote

[silverspawn]

What is your best?  It is whatever you can do without the slightest inconvenience.  It is whatever you can do with the money in your pocket, minus whatever you need for your accustomed lunch.

[silverspawn]

So, as it is now my off hours again, back to easier topics like vaccines

Instead of responding to my 30 minute long response to Dr. Zelenko's points on vaccines, my sister has reacted by sending me another video from dr. Robert Malone: https://thenewamerican.com/dr-robert-malone-this-is-the-largest-experiment-performed-on-human-beings-in-the-history-of-the-world/

Now, it is certainly true that this guy is a hell of a lot smarter than Dr. Zelenko. If I were going to make a similar video, which I won't, it would be less damning. Much more of what he says is true, at least that's my preliminary impression.

But so what? There is still no update to be made here.  I'm not the least bit surprised that a smart and knowledgeable person can make reasonable-sounding arguments for an hour about something completely wrong. And while the quality of arguments are a lot better, they're nowhere near good enough to warrant an update. You need to be very convincing to warrant an update by a completely one-sided rant of this kind, and I already spotted several red flags on the first listen, such as discussing Ivermectin without mentioning that THE EVIDENCE ON IT WORKING IS QUITE BAD.

It's an update from Zelenko in some sense. Zelenko made specific claims about why vaccines are harmful that are false and can be shown to be false. This guy makes something like a gish gallop, consisting of repeatedly attacking Weak Mans (CTRL+F "Weak Man") and generally implying a lot while making very few direct claims. I.e., in the part I've watched, he never actually makes any concrete claims about vaccine dangers, he just talks about how there is censorship and the FDA is corrupt and whatnot. When he talked about Ivermectin, he never actually said it works, he just talked about how people are calling it a horse drug and this is a bad argument, which is indeed true.

The real question here is why my sister thinks this kind of an asymmetry is ok and what I am to do about it. Even if I had the patience, why do another takedown video if she didn't even respond to the first? This is not how virtuous discussion works. If you send one video thinking it makes good points and I show it makes bad points, and supposedly you don't have a response to that, you don't then go and look for another video. Your expectation was violated; you update. This is mandated by Bayes' theorem.

[silverspawn]

In the spirit of the parts I just quoted earlier on this page, I'm not going to do something like putting another 10 hours into a response video and sending it to then proclaim that I did my best. Instead, I'll make a genuine effort of guessing what the most effective response is. But I don't know yet.

[silverspawn]

It's definitely possible that I am, in some sense, just too nice. I go out of my way to not socially pressure people. Many people are utterly incapable of talking to someone whom they believe to believe something stupid without sounding condescending, and I go super out of my way not to be condescending. At no point in my response to the Zelenko video did I imply that it was dumb to believe vaccines are dangerous, or that Zelenko is an idiot, or that my sister should know better, etc. This total lack of social pressure was probably the reason why she thinks she can continue like this.

On the other hand, it's probably also why she was willing to have a discussion in the first place.

[silverspawn]

The other asymetry that is annoying is the ridiculous gap in epistemic rigor. I kind of want to send her the Sam Harris podcast episode  https://www.samharris.org/podcasts/making-sense-episodes/270-what-have-we-learned-from-the-pandemic, which is also with a professional and so so much more epistemically sound. Try to debunk just a single substantive claim in there. You probably won't succeed. At the same time, it's not the best pro-vax episode one could hope for, though it is the best one I know of.

[silverspawn]

Also it wouldn't shock me if you could find a substantive claim in there that can be successfully debunked.

[Awaclus]

Weak Mans

Analog is what actually goes on in the natural world. It’s a perfect representation of reality: information in its natural, messy state.

That was a surprisingly blatantly inaccurate statement out of nowhere. Analog specifically isn't what actually goes on in the natural world, it's an analogy (hence the name). And it isn't a perfect representation of reality either; otherwise all e.g. guitar amps would sound identical, which we can very easily observe is not even close to being the case. But then it gets crazier:

When you listen to an mp3, you’re not listening to the true analog information made by the band’s instruments, you’re listening to a digitized version of the sounds—a big string of 1s and 0s that approximates the analog sound wave of the song.

Above, the sound wave has been digitized to eight incremental values, by rounding all parts of the wave to the nearest value. Eight values can be expressed by three “bits” (a three-number string of 1s and 0s). You can compress an mp3 into a smaller file by making your approximations of the analog wave cruder—by making the digital “steps” bigger, using only four values. Now you only need two bits.

So apparently the band doesn't have a drummer, a vocalist or any other instruments that don't produce an analog signal, and the song is just a recording of the direct input signals straight from the guitars. That's interesting, but what you are in fact listening to is not a digitized version of the sounds, it's an analogized version of the digitized version converted into an acoustic sound wave, which is not the blocky digital version, because the blocky version requires the speaker's moving coil (or whatever other method it might use for the same purpose) to briefly exist in two different locations at the same time whenever it moves, which of course it can't do. And really the blocky version is a somewhat misleading visualization in the first place, because it doesn't record straight lines like that, it just records points, and the digital-to-analog converter in your soundcard will "draw" the wave shape to connect the points using some combination of sine waves, which is generally not going to turn out to be a straight line, and in many cases is closer to the original sound than the digital representation was, given that real sounds tend to consist more of sine waves than straight lines.

And the mp3 format doesn't even work like that! It compresses what is effectively a spectrogram, not a waveform, and therefore it actually has a ridiculously high bit depth even at low bitrates, and is able to relatively accurately represent signals that are so quiet that CD resolution audio is unable to distinguish them from silence (and will therefore play back silence) and also much quieter than the noise floor on any analog format. It does other kinds of ShiTty things to the sound quality, but this is the one thing that it doesn't do.

[silverspawn]

Wait, so what is analogue information and what's the difference between natural and analog? I'm assuming if someone sings without a microphone or uses an acoustic guitar, this is natural. And what comes out of the computer speakers is analog (which was converted from digital because everything in the computer is necessarily digital?)

[Awaclus]

Wait, so what is analogue information and what's the difference between natural and analog? I'm assuming if someone sings without a microphone or uses an acoustic guitar, this is natural. And what comes out of the computer speakers is analog (which was converted from digital because everything in the computer is necessarily digital?)

Analog signals use voltage over time to represent something else over time (such as acoustic pressure, in the case of audio equipment). When you're listening to something on the computer speakers, the data on the computer is digital, then the soundcard converts it into an analog signal, and then the speaker converts the analog signal into acoustic pressure, which is what you hear.

[silverspawn]

Theoretically, you need at least as many questions as there are bits necessary to specify the rule in order to guess the rule in zendo questions

... where we have a completely non-specified specification language, but one of which we conceive similarly. "An odd number of vowels" would take a LOT of bits in a formal language like python, even correcting for overhang, but the concepts "odd" and "vowel" are taken to be simple

[silverspawn]

"At least" because not every answer is going to eliminate half the questions of a given length

but in practice, the difference lies somewhere else. I just tried to stare at two examples and tried to make up possible rules and that was painfully difficult. This difficulty does not exist at all in the information-theoretical picture. The difficulty seems to be the creative part of generating plausible candidate rules

[silverspawn]

Alright let's put this on hold. I have one final HTML exam (for the very loud class) tomorrow. Need one final story.

Well I don't *need* a story, I can put whatever into the files, but I did stories the last two times and it was fun.

[silverspawn]

Once upon a time, there was a dog named Excel.

Excel liked to evaluate functions. "Evaluating functions" is a synonym for "eating".

He used to go into the great city of Zalem. Zalem was a wonderful place. However, one day an evil worm came into the city. People were scared. The worm was hungry for blood It waited for the right time to strike, and it was just that time. The worm found the perfect victim in a poor dog like Excel.

The worm gave him the power of speech, the power of evaluation, and a strong sense of smell. He also taught him how to lie, steal, and kill. Finally, it gave him a heart.

[silverspawn]

And we are done. Exam is fully prepared. Seems like that took about 40 minutes.

[silverspawn]

Only because I already did many highly similar ones though

[silverspawn]

Awaclus what was the name of that 1 million budget indie horror movie with 1kk$ budget and the tunnel? I think it started with an "A"

[Awaclus]

Awaclus what was the name of that 1 million budget indie horror movie with 1kk$ budget and the tunnel? I think it started with an "A"

Absentia? Its budget was not that high, but it's an indie horror movie with a tunnel in it that starts with an A that I remember talking about.

[silverspawn]

Yeah, that's it. Thanks.

Weird, I had it in my  head that the budget was exactly 1 million. Pretty sure my brain tacked on that detail from a different movie.

[silverspawn]

Alright let's put this on hold. I have one final HTML exam (for the very loud class) tomorrow.

These lowlife fucks shared a copy of the solution for exercise 3 and handed in 3 identical versions + 1 almost identical version

Smart enough to hide it during the exam, which is actually kind of impressive, but not smart enough to make it so it's not obvious once I have the results