nostalgebraist.tumblr.com →

slatestarscratchpad:

nostalgebraist:

robertskmiles:

nostalgebraist:

I know I my obsession with Big Yud / Less Wrong / “rationalists” / etc. must seem like it goes beyond the bounds of all sense at times, but really I’m just amazed at how much weird internet stuff I can manage to find by mining this particular vein. It just never ends

E.g. apparently there is this…

I think the multicoloured suns might actually be *post*rationalists, whatever that really means.

Their avatars are all recoloured images of the sun in different hues, and their names all take the format X of Y, for example Instance of Class, Element of Set, Member of Species.

They’re not that weird though, to be honest.

I’ve picked up on the multicolored suns and the X of Y thing, I just don’t know why there’s a set of people who do this, and what the theme is about.  Do they have a unifying feature (beyond “post-rationalism”)?  If I started a colored sun account with an X of Y name and posted anything I wanted, would I thus become a Weird Sun Twitter, or are there content requirements as well?

(I guess it’s a bit like various injokey username trends like [X]haver — jobhaver, sexhaver, etc. — but I can understand those because they’re funny, or intended to be, at least.  I don’t know what the suns are going for.  Maybe it’s just absurdism?)

I think Weird Sun Twitter started with one account with the relevant characteristics who was really interesting and spawned a lot of imitators. I don’t think the humor is absurdist in the sense of “literally nonsense”, but there are a

lot of complicated in-jokes, and I think a few of the imitators don’t understand them and do lapse into meaningless absurdism by mistake.

I also used to be confused about whether you could just start an account and become part of Weird Sun Twitter, but I’ve since met somebody who did just that, it worked fine, and apparently it’s how all of them started. The content requirements are to speak in noun phrases a lot and be really meta-level all the time.

I highly recommend @UnitOfSelection, who aggregates all the good weird sun tweets.

Well, now I know, I guess.  Thanks!

Although looking at @UnitOfSelection it’s pretty clear to me that I’m pretty far from actually understanding these people, per se … 

(via slatestarscratchpad)

nostalgebraist.tumblr.com →

I know I my obsession with Big Yud / Less Wrong / “rationalists” / etc. must seem like it goes beyond the bounds of all sense at times, but really I’m just amazed at how much weird internet stuff I can manage to find by mining this particular vein.  It just never ends

E.g. apparently there is this sub-subculture of rationalists (I think?) on twitter called “weird sun twitter.”  I don’t know what these people are about; I’ve never looked into them; I just occasionally see people make these baffling references to “weird sun twitter”

And, like, isn’t that name in itself just a gem?  "Weird sun twitter.“  We are blessed to live in a world that contains such things

I’ve been in this game for years and there are still things I don’t know about.  The well has no bottom

uncrediblehallq:

After nostalgebraist mentioned it, I went and picked up The Ayn Rand Cult.

Some of the things the book describes are uncomfortably familiar. But if everything it says is true, some of what went on in the heyday of Rand’s inner circle was a whole ‘nother level of bad compared to anything I’ve seen in the LessWrong community. Yudkowsky often does a terrible job of handing criticism, but if The Ayn Rand Cult is to be believed, Rand was basically constantly exploding at people for any hint of dissent. There were even regular “trials” that would lead to people being ostracized from the group. I’d previously been under the impression that when Rand broke with Nathaniel Branden, a lot of her followers left in disgust, but the book makes it out that many were forced out for hesitating to take Rand’s side in the affair.

The only thing I don’t know is to what extent the book may be exaggerating. Shlevy recently claimed that Rand was close to many people who had disagreements with her. On the other hand, he’s also said that “The situation when Rand was alive and especially during the heyday of the ‘inner circle’ was definitely qualitatively worse than what exists in the modern Objectivist movement and LW.”

Shlevy, want to shed more light on this?

I’m curious about this too – there’s a lot of interesting (if true!) info in that book, but it has such a ranting, hectoring, “Objectivism killed my puppy” tone that it’s kind of hard to trust the author sometimes.

how many worlds are in a BLT?

jadagul:

su3su2u1:

hot-gay-rationalist:

<snip>

And what MWI says, at any rate, is just that “you’re made of Q.M., you get superposed too,” which… like, I don’t know how anyone could possibly argue against that? The “and thus there are ‘copies’ of you” schitz appears to be a consequence of this, but it’s by no means the meat of the interpretation, and if we ever figure out that it’s not true in fact that “superposed mes” implies “multiple mes,” then people will be looking really embarrassed for having changed the name of the “relative-state interpretation” to “many-worlds interpretation.”

See, that is not all that MWI says.  There are lots of pieces to quantum mechanics, if I want to say “you’re made of QM,”

I can mean, 1. you are (part of) one of a set of classical histories of the universe, 2. you are (part of) a universal wave function 3. information about you is contained in a function that is a generalization of a Bayesian probability distribution 4. you are a collection of particles that interacts with a “pilot” wave function ,etc. 

Many worlds says I don’t need any postulates at all about how to connect quantum mechanics to a measurement.  The theory interprets itself. This would be great if anyone could make that happen.  No one can. So like every other physical theory, maybe we need more postulates to connect the mathematical structure to reality. 

I’m wondering if there’s a communications gap here between people who’ve studied physics and people who haven’t.  

The main thing I got out of Yudkowsky’s QM essays was the idea that 1) humans are in superposition as much as everything else, and 2) observed “waveform collapse” is about, roughly, our brains becoming entangled with whatever quantum state is “collapsing.”  

I read it as a refutation of the idea that humans are _not_ in quantum states and that “waveform collapse” is a result of our consciousness causing something to _happen_ to a quantum waveform.  Which is certainly an idea that has a lot of pop resonance.

Now, within the physics community this might not be what people mean when they ask whether they believe in many worlds.  But this is the question I read Yudkowsky as addressing.  Probably because I’ve mostly heard the pop-resonant idea of magical consciousness waveform collapse, and the alternate pop-resonant idea of alternate-timelines-where-dinosaurs-are-alive. 

I have no idea which of these arguments Yudkowsky was actually trying to advance.

Yudkowsky’s sequence does advance the ideas #1 and #2 you mentioned, and does a good job of presenting their intuitive appeal, IMO.

The reason the sequence raises hackles among physicists is that Yudkowsky presents the intuitive appeal (roughly/partially) as follows: “QM looks like a theory of amplitude flowing between configurations, and it’s a lot more straightforward if you just figure that all the configurations are real and the amplitude is really flowing between real things.”

Problem is, there is more than one way you can break a system down into configurations, and QM doesn’t allow you to pick a special one.

So, when Yudkowsky talks about configurations where photons do different things, and how QM is a theory of amplitude flowing between these configurations, and how you can’t reduce this to a theory where the photon does one single thing according to some rule set – that’s all good, as far as it goes.  But what he doesn’t mention is that you can always break down the same case into various different sets of configurations, and which of these ends up being useful depends on what you end up measuring (which means none of them can be uniquely “correct” before you measure, because the system doesn’t know what you’re going to do before you do it).

So the intuitive Yudkowsky picture is “gee, this sure looks like amplitude flowing between different configurations.  Why not just suppose that’s really happening, and the configurations are all real?”

And real QM is “this looks like a thing that could be broken down into amplitude flowing between configurations in various ways, and no such breakdown is more ‘true’ than any other, so if all the configurations are real, all the distinct configuration breakdowns have to be real.  My BLT simultaneously consists of five real ingredient components and two real half components (but not ten real half-ingredient components because noncommutivity).”

In other words, if you’ve already accepted the Yudkowsky picture of multiple coexisting real configurations, now imagine a bunch of sets of such things rather than just one set, all of which are equally valid, yet which cannot be combined into a single picture!  That is much less intuitive.  So the QM sequence makes MWI sound more intuitive than it is.

In this light, the post about the position basis being better seems (to physicists) like a sneaky (and unsatisfying) attempt to sidestep this issue, while seeming to non-physicists like an inexplicable detour.  It’s the moment when he kind-of-acknowledges the problems with his presentation, but only in the course of trying to sweep them under the rug.

(via jadagul)

hot-gay-rationalist:

I find it kinna funny that nostalgebraist and @su3su2u1 (who can’t be mentioned for some reason) say that HPMOR is bad HP fanfic when it’s the single most reviewed fanfic on FF.net.

Like, one can claim that it’s got its flaws, isn’t super good or whatever, but to say it’s bad in any objective sense, when so many people seem to like it? Seems to defy normal standards of the meaning of “bad.”

Although well when people say classics are good and then most people who read said classics think they’re boring shit, that’s also a weird definition of “good.”

I think we should make a difference between “pleasing-to-readers” and “well-written.” And then add a billion caveats to the latter because one man’s trash and all that.

(ftr I agree with a lot of criticism aforementioned tumblrites aim at the fanfic, I just don’t agree they mean the fanfic is bad :P)

(and like lots of stories, HPMOR is best read quickly and in one go so you don’t notice all the flaws)

There is a whole array of social conventions built up around these things, and I’m merely following those conventions.  Like, I agree that aesthetics is a complicated thing and it’s really hard to say what we actually mean by “good” or “bad,” and whether these things are objectively definable.  But we go on talking about “good” and “bad” art and when we do so, the convention is that popularity doesn’t really matter.

E.g. if I say “The Da Vinci Code is bad,” I am expressing a very commonly held and expressed opinion, even though The Da Vinci Code was bought and enjoyed by a great many people.  Likewise for Fifty Shades of Gray, or any one of a wide set of “bestsellers widely considered bad.”

Now, I agree that this is kind of weird – if a lot of people like this stuff, what do we mean by “it’s bad”?  It’s a hard question to answer because there is no good, complete theory of evaluative aesthetics to apply to it.  (I think the answer is a big mix of things like “easily replaceable enjoyment vs. unique, inventive enjoyment” and “has depths that reward re-reading vs. read it once and throw it in the trash” and so forth, with a bunch of elitism and arbitrary historical baggage that would need to be distinguished and scraped off to get a working theory.)

But people go on saying these things constantly, while a working theory of the kind I mentioned is far from sight and would take a very long time to produce.  Instead of pretending we have such a theory, I think we should just go with the resources natural language provides us with until something better comes along.

(Speaking differently in this particular case while everyone else goes on speaking the same way would feel a bit like responding to an “isolated demand for rigor,” though I’m not accusing you of making such a demand.)

(via hot-queer-rationalist-deactivat)

how many worlds are in a BLT?

hot-gay-rationalist:

nostalgebraist:

While I’m on the subject, I might as well talk about the silly analogy for the basis problem that’s I have floating around in my head —

So imagine you have a BLT.  (That’s a bacon, lettuce and tomato sandwich, if they don’t have those in your area.)  Say you want to describe a BLT as a “combination” of other things.

Well, there are many ways you can do this.  Here are two:

1. Break it down by ingredients: “piece of bread + bacon + lettuce + tomato + another piece of bread.”
2. Break it down into two pieces of a full BLT: “left half of a BLT + right half of a BLT.”

You could also do all sorts of other things (break into quarters, etc.) but let’s focus on those two.

Now, first of all, these are all equivalent descriptions.  If you have a BLT, it is not somehow “really” a stack of ingredients any more than it is “really” a left and a right half joined together, or vice versa.  You don’t look at a BLT and say, “ah, yes, that one is a ‘sum of ingredients’” and then look at some other BLT and say “oh, but that one is a ‘sum of two halves’.”  (If you like the map/territory distinction, this is in the map, not in the territory.)

Okay, now quantum mechanics gets involved.  The BLT here is analogous to the “wavefunction,” which describes a quantum system.

The twist here is that in quantum mechanics, “things you can measure about a system” (often called “observables”) correspond to “ways you can break down the wavefunction.”

So for our BLT, if we were to give it properties we could measure, they would be things like “which ingredient are we looking at” or “which half are we looking at.”

So the BLT’s sitting there, minding its own business, and we come along and say “let’s ask the question ‘which ingredient’?”  And we’ll get an answer like “lettuce” or “bacon.”  These happen randomly, but with predictable probabilities.

Or we can ask “which half” and we’ll get an answer like “left half.”

In fact, after the measurement, instead of seeing the full BLT that was there, we just see the piece corresponding to our measurement.  So if we go up to a BLT and ask “which ingredient?” and get the answer “bacon,” the BLT will change into just a bunch of bacon.  Or if we ask “which half?” and get “left half” it will turn into just the left half of a BLT.

This is what people often call “wavefunction collapse” and has led to weird ideas about physicists shaping the universe with their minds and stuff.

Okay, now for MWI and the basis problem.  The two ways of decomposing the BLT enumerated above could be called two “bases.”  Each has a question corresponding to it.  The whole “collapse with random results” thing is weird and people want to explain it.

One possible explanation, the MWI one, is that what we do when we ask, say, “which ingredient?”, is a bunch of copies of us get linked up to one ingredient each.  So if I see “lettuce” as a result there is a copy of me that sees bacon and so on.  It’s not that the BLT changes, because all those ingredients were already there; it’s just that I get linked up to a particular one of them (and various alterna-mes get linked to the others).

It’s not that the universe splits when I observe something — I think that is a misconception about MWI? — it’s that the BLT is made up of multiple things, and when I observe, one of me gets linked to one of the things.

Now here is the problem.  If you imagine the BLT as a sum of ingredients, and ask “which ingredient?”, this all works.  It’s like the “world with only bacon” and “world with only lettuce” and so forth already existed, waiting for me to get shunted into one or the other.

But what if I ask “which half?”  Now I either get shunted into the “world with only left half” or the “world with only right half.”  But that is a different set of worlds than we had with the ingredients!  If there really is a “bacon world” and a “lettuce world” and so forth before I do the measurement, which world do I end up in when I get “left half”?

And in fact, if you look at the laws of QM, you see that it treats all of these breakdowns identically.  It doesn’t seem to care about “ingredients” vs. “halves” until the measurement happens; all it sees is a full BLT, and it’s silly to say “well it’s really a combination of halves, all other descriptions are less real” (which is the gist of this post).

The motivation for the “worlds” thinking was the idea of “counting worlds” — that if you can say “there are five ingredients and two of them are bread” then you can make sense of a prediction like “I’ll see bread 2/5 of the time” by postulating five worlds, each corresponding to one ingredient.  (Actually, when people try to do the math for this it doesn’t work, but let’s sidestep that for now!)

But if we want to describe the whole universe in these terms, we’d want to be able to count worlds even before a measurement.  We’d want to be able to imagine a full BLT and say, “ah, yes, there are five worlds in there” (or whatever).  But the world-counting goes differently depending on which basis you use: there are five worlds ingredient-wise, and two halves-wise.  How many worlds are there in the BLT?  Five or two?  And what is in them?

The impossibility of straightforwardly answering questions like that is what makes an straightforward intuitive account of QM in terms of “worlds” impossible.

(If you want to connect this post to other popular presentation of QM, try thinking of “ingredients” vs. “halves” as “position” vs. “momentum,” or vice versa.)

Well I mean this all seems to assume that worlds are some sort of discrete indivisible thing, when quantum mechanics is very much continuous. I have no idea whence come the Born Probabilities (and pretty sure no one does), but any account of them that tries to make them into “world-counting” will predictably fail.

And what MWI says, at any rate, is just that “you’re made of Q.M., you get superposed too,” which… like, I don’t know how anyone could possibly argue against that? The “and thus there are ‘copies’ of you” schitz appears to be a consequence of this, but it’s by no means the meat of the interpretation, and if we ever figure out that it’s not true in fact that “superposed mes” implies “multiple mes,” then people will be looking really embarrassed for having changed the name of the “relative-state interpretation” to “many-worlds interpretation.”

“quantum mechanics is very much continuous” – not necessarily.  Remember where the name came from!  Sometimes your observable has a discrete spectrum.  Discrete energy states and stuff.  You can count those.

That cheap shot aside, even with a continuous spectrum, I’m not sure what your objection to “world-counting” is.  I’m using that phrase as a proxy for getting the Born probabilities out of some measure over your basis of eigenstates.

For a continuous spectrum, it won’t quite be “counting worlds” per se, but that’s the example that’s clearest to explain to people who haven’t studied math, which was the audience of that post (which is why I made this whole BLT analogy instead of just saying “vector space” and “basis”).

Getting the Born probabilities out of world-counting (i.e. making a measure over worlds) has always been a main goal of key MWI practitioners, including Everett, DeWitt, and Deutsch.  Here’s a randomly selected paper about it (which uses the term “counting branches,” incidentally).  Part of the appeal of having multiple worlds is that it gives you an ensemble, and an ensemble can let you explain randomness without recourse to ideas like “it’s just inherently random.”

But for this to work, the probabilities given by assuming “I’m a randomly selected member of the ensemble, chosen without bias” have to actually line up with the results seen in experiments.  This is what MWI has had trouble with.

“You’re made of Q.M., you get superposed too” isn’t the whole of MWI; the whole of MWI is, as you said, that there are “multiple mes” (and, historically, this is often attached to “we can thus explain the Born rule”).  This is basically a semantic issue – I don’t think you’re using the term “MWI” the way the physics community uses it.  When I talk about “MWI” I’m not just talking about “you get superposed too,” I’m talking about interpretations that attempt to define some specific, preferred breakdown into “worlds” which you can treat as an ensemble and derive probabilities from.

(via hot-queer-rationalist-deactivat)

how many worlds are in a BLT? →

how many worlds are in a BLT?

While I’m on the subject, I might as well talk about the silly analogy for the basis problem that’s I have floating around in my head –

So imagine you have a BLT.  (That’s a bacon, lettuce and tomato sandwich, if they don’t have those in your area.)  Say you want to describe a BLT as a “combination” of other things.

Well, there are many ways you can do this.  Here are two:

1. Break it down by ingredients: “piece of bread + bacon + lettuce + tomato + another piece of bread.”
2. Break it down into two pieces of a full BLT: “left half of a BLT + right half of a BLT.”

You could also do all sorts of other things (break into quarters, etc.) but let’s focus on those two.

Now, first of all, these are all equivalent descriptions.  If you have a BLT, it is not somehow “really” a stack of ingredients any more than it is “really” a left and a right half joined together, or vice versa.  You don’t look at a BLT and say, “ah, yes, that one is a ‘sum of ingredients’” and then look at some other BLT and say “oh, but that one is a 'sum of two halves’.”  (If you like the map/territory distinction, this is in the map, not in the territory.)

Okay, now quantum mechanics gets involved.  The BLT here is analogous to the “wavefunction,” which describes a quantum system.

The twist here is that in quantum mechanics, “things you can measure about a system” (often called “observables”) correspond to “ways you can break down the wavefunction.”

So for our BLT, if we were to give it properties we could measure, they would be things like “which ingredient are we looking at” or “which half are we looking at.”

So the BLT’s sitting there, minding its own business, and we come along and say “let’s ask the question 'which ingredient’?”  And we’ll get an answer like “lettuce” or “bacon.”  These happen randomly, but with predictable probabilities.

Or we can ask “which half” and we’ll get an answer like “left half.”

In fact, after the measurement, instead of seeing the full BLT that was there, we just see the piece corresponding to our measurement.  So if we go up to a BLT and ask “which ingredient?” and get the answer “bacon,” the BLT will change into just a bunch of bacon.  Or if we ask “which half?” and get “left half” it will turn into just the left half of a BLT.

This is what people often call “wavefunction collapse” and has led to weird ideas about physicists shaping the universe with their minds and stuff.

Okay, now for MWI and the basis problem.  The two ways of decomposing the BLT enumerated above could be called two “bases.”  Each has a question corresponding to it.  The whole “collapse with random results” thing is weird and people want to explain it.

One possible explanation, the MWI one, is that what we do when we ask, say, “which ingredient?”, is a bunch of copies of us get linked up to one ingredient each.  So if I see “lettuce” as a result there is a copy of me that sees bacon and so on.  It’s not that the BLT changes, because all those ingredients were already there; it’s just that I get linked up to a particular one of them (and various alterna-mes get linked to the others).

It’s not that the universe splits when I observe something – I think that is a misconception about MWI? – it’s that the BLT is made up of multiple things, and when I observe, one of me gets linked to one of the things.

Now here is the problem.  If you imagine the BLT as a sum of ingredients, and ask “which ingredient?”, this all works.  It’s like the “world with only bacon” and “world with only lettuce” and so forth already existed, waiting for me to get shunted into one or the other.

But what if I ask “which half?”  Now I either get shunted into the “world with only left half” or the “world with only right half.”  But that is a different set of worlds than we had with the ingredients!  If there really is a “bacon world” and a “lettuce world” and so forth before I do the measurement, which world do I end up in when I get “left half”?

And in fact, if you look at the laws of QM, you see that it treats all of these breakdowns identically.  It doesn’t seem to care about “ingredients” vs. “halves” until the measurement happens; all it sees is a full BLT, and it’s silly to say “well it’s really a combination of halves, all other descriptions are less real” (which is the gist of this post).

The motivation for the “worlds” thinking was the idea of “counting worlds” – that if you can say “there are five ingredients and two of them are bread” then you can make sense of a prediction like “I’ll see bread 2/5 of the time” by postulating five worlds, each corresponding to one ingredient.  (Actually, when people try to do the math for this it doesn’t work, but let’s sidestep that for now!)

But if we want to describe the whole universe in these terms, we’d want to be able to count worlds even before a measurement.  We’d want to be able to imagine a full BLT and say, “ah, yes, there are five worlds in there” (or whatever).  But the world-counting goes differently depending on which basis you use: there are five worlds ingredient-wise, and two halves-wise.  How many worlds are there in the BLT?  Five or two?  And what is in them?

The impossibility of straightforwardly answering questions like that is what makes an straightforward intuitive account of QM in terms of “worlds” impossible.

(If you want to connect this post to other popular presentation of QM, try thinking of “ingredients” vs. “halves” as “position” vs. “momentum,” or vice versa.)