(no subject)

Look this is spoilers for the movie oblivion, but it's a silly movie anyway...

the girl REPLACES HER HUSBAND WITH A CLONE TWICE

like her husband is just... replaceable! There's a lot of him! When something bad happens she gets a new one!

ok here's another one. In El Goonish Shive, Nanase and Elliot break up. Later, Nanase realizes she was a lesbian, and though she really liked Elliot was never really physically attracted to him. Then she gets together with Elliot's female copy, Ellen. Isn't that convenient!

So many problems you can solve by copying people

(no subject)

my own theory is that the discomfort of thinking is mostly a response to the danger of not attending to the environment, and it's reduced by quiet, being alone or with people you trust, not multitasking, not having anything urgent coming up, not feeling like there's someone who hates you, a lack of clutter and well-cleared paths for walking and moving (if you move your arm you wont bump your elbow etc), and anything else that gives you reason to believe its safe to ignore your environment

(no subject)

basically, if you have some evolutionary reason for expecting something, ask yourself:

• what do you know about the selection pressures during the time the trait was evolving? Do you know what would have been seelcted for, and how strongly? If it's something related to hunting, do you know how the sepcies hunted at that time? If it's something related to mating, do you know how the species mating rituals worked at that time? Not just a vague idea, but well enough to know what makes the difference in success vs failure, and what is negligible compared to other factors

• what do you know about the available genetic variation at that time? Are you sure alleles with the effect you're thinking of actually existed? Or were more common than alleles with different effect sthat responded to the same selection pressure?

Garbage in garbage out.

(no subject)

Francis Crick - What Mad Pursuit:

Biologists must constantly keep in mind that what they see was not designed, but rather evolved. It might be thought, therefore, that evolutionary arguments would play a large part in guiding biological research, but this is far from the case. It is difficult enough to study what is happening now. To try to figure out exactly what happened in evolution is even more difficult. Thus evolutionary arguments can usefully be used as hints to suggest possible lines of research, but it is highly dangerous to trust them too much. It is all too easy to make mistaken inferences unless the process involved is already very well understood.

(no subject)

fermat's last theorem and irrationality of ζ(3) were both important in the 90s... yes the ζ(3) proof was 70s but bear with me

they're both unsolved problems from the beginning of number theory. Not really comparable because fermats last theorem was pursued much harder, and because the proof was deeper and resulted in widely applicable theory.

BUT... here's where the 90s come in... the 90s is when the hard part of the ζ(3) proof was automated. Part (all?) of this was zeilbergers algorithm, which you may have used without realizing it in a computer algebra system. A different kind of generality.

and zeilberger pushed this politically, politicized judgments of what's "real math". You may have heard, its tragic what you learned in your math classes, just symbol manipulation and procedures, whereas real math is deep and about ideas and there's barely any numbers in the formulas. No, said zeilberger, the symbol manipulation is the real math, and its cool and shouldnt be disparaged.

i kind of wonder, how much more successful would this political move have been, had it not been for the fermats last theorem proof? Would we just take for granted, ah the increasing abstraction of 20th century math didnt connect with real problems, it turns out we should all obsess over symbol manipulation procedures and achieve generality with symbolic computation algorithms rather than with theorems? Idk probably this was just one of many things going on, i mean i feel like penrose recently getting the nobel prize is another big confirmation of the value of super abstract 20th century math although tbh idk what im talking about cause idk much about penroses work

(no subject)

scientism: metaphysical debates are stupid because as someone with an engineering degree i know the right answers

positivism: metaphysical debates are stupid because the questions are meaningless

if you translate all arguments into social conflicts, they can look the same

(no subject)

questionable content robots are a fantasy about a better life for humans, whether or not they were intended that way. Like, May having a shitty robot body and spending all her money on repairs is basically just someone trapped in poverty by a chronic health condition. And she has a gofundme. Except it permanently solved the problem--she buys a new body, which works (she's healthy) (and also she's like two feet taller and has huge boobs cause she wanted that too). Like this is a vision of how much better things could be, a reminder of how bad they are now

(no subject)

ok rereading the (second) transformation party chapter now that i have context.

Ashley met Elliot when she saw him transforming into a girl. So now I have some context for why this is important to Elliot... his ex Sarah had been a bit weirded out by that, and I think she said in the first transformation party that she didnt want their first kiss to be when theyre gender switched. But to Ashley this is part of the appeal, and their first kiss is when Elliot is gender switched.

Ok, cool, Elliot has a girlfriend who's really into transformation magic. So how come their third date is just like... normal? When they met it was exciting and supernatural, so how come now Ashley doesn't want to talk about magic?

so... ashley realizes that her fantasies are not things she'd want to happen in real life. Like her transformation fic includes like, sexy noncon transformation.

first of all, that means she feels she cant be open about the subject with elliot--she doesnt really want to share this side of herself with someone she really likes but doesnt know. So the most exciting side of her relationship to elliot to her, is also a subject she's avoiding.

Second, between date 2 and date 3, ashley found out she's gonna be a wizard. She's going to be able to do these transformation spells, while at the same time feeling like she's the last person who should be trusted with them.

Its complciated and elliot doesnt get it at all, he just knows ashley has magic on her mind but doesnt want to talk about it with him, and so he gets tedd to have the party so she'll have other people to talk to.

ok i think thats it... theres so much backstory to everything in this comic... a lot of effort to load context

(no subject)

i got a visceral lesson in the power of compression that you kids won't get, cause i remember when i first got a CD player that could decode mp3's, and started burning CDs with >100 songs on them, thinking, like, what the fuck.... like the correspondence between CD and album had been so firm for me, and I saw it shattered and never forgot

(no subject)

some things its weird to me to call it a "theorem", like the hellman-feynman theorem or the blackwell-rao theorem. Blackwell-rao's proof is just one application of jensen's inequality. It's not a mathematical achievement, a unit of math that should be remembered so we dont have to go through the proof every time. Rather its an expository achievement. After Blackwell chose to explain the concept of a sufficient statistic in this way, there's been no going back: that's how it's always explained. (not the only time Blackwell did this!)

generative models

a regression model is not a generative model

a regression model, plus an estimated probability distribution for the features, is a generative model

But that's a generative model of (feature vector, response) pairs. If the features are generated from some underlying object, it's not a generative model of that object...

... unless you can invert the featurization process and find an objet with those features

old fashioned science never stopped

• Evolution by mutation and natural selection
• Quantum mechanics
• Relativity

When you picture the origins of these theories it's Darwin making sketches of birds or Einstein scribbling equations.

We've done a lot since then, like the human genome project and the standard model, and the character of the research seems to have changed. Big projects where each person plays a small part, expensive equipment like particle accelerators and space telescopes, and computers.

I want to argue though that the "one person with a notebook" type research never really stopped, not just because people had to do that to explain particle accelerator results, but because people kept making scientific advances that in principle could have been made in the early 20th century.

• Density functional theory, originating in the 1960s, nobel prize 1998. It's the basis of the computer programs used to get wavefunctions for large systems. But also it's just a convenient mathematical reformulation of quantum mechanics. For example you can derive a theoretical basis for something like Pauling's electronegativity scale. I imagine Pauling would have loved that, though idk of any comment from him.
• The coalescent process, originating in the 1980s. It's what's used in computer programs for data-based inference of stuff like how long ago a SNP originated or whether it was selected for. But it's also just an extremely convenient way to derive some of the basic evolution math stuff. Fisher would have loved it had he lived to see it.

These are mathematical reformulations of existing theories, which turned out to be very useful for computational work, but which are also of independent value. It seems to me they could have been thought of at any time during the 20th century, but in fact were thought of pretty late. Basically, when Fisher, Einstein, Heisenberg etc were working out the foundations of the new 20th century theories, they didn't go as far as they could have with their pencils and paper. There was still plenty of worthwhile pencil-and-paper work left to do through out the 20th century. And, maybe even up to today, I don't know.

Oh, I'd add a third bullet point related to general relativity, if I knew anything about it. But there must be something. There's been an explosion of GR practice to explain all the weird stuff we see with modern telescopes, and it's hard for me to believe it has much resemblance to Einstein's GR practice.

I do think that these mathematical reformulations are events of comparable insight and impact to the original discoveries of the theories. They're why we can actually do stuff with the theories. I don't think we would care nearly so much about quantum mechanics if it didn't turn out that it could be applied.