Re: Outside The Lines

1

Contact your lawyer or seek emergency judicial attention if your recount is painful or lasts longer than 4 weeks.


Posted by: JP Stormcrow | Link to this comment | 11-28-08 11:51 PM
horizontal rule
2

Booooring.

Basically both sides are gaming the count so that the ultimate result reported in newspapers is not unfavorable to their side. Probably this is because the first two days of recounting reduced Coleman's already-small advantage, so Coleman sterted making frivolous challenges, and after a lag so did Franken. It's continuing to escalate.

Coleman has already declared victory two or three times, and his national supporters have been talking about a stolen election for weeks. When the final recount shows Coleman with an advantage, he'll claim victory again. Then the challenged ballots will be counted, which will probably (not certainly) reduce Coleman's margin, and probably the talk about a stolen election will amplify. The loser will then go to court -- Franken wants rejected absentee ballots considered. If it's Franken who goes to court, Coleman will scream bloody murder. Stolen election again.

Ultimately this will be decided by the courts, Governor Pawlenty, or the Senate. The Coleman strategy is to put maximum public pressure on them, at best getting them to cave in, and at worst sending Franken to the Senate under a cloud.

The recount is all kabuki now. The election is really undecidable -- in 9 out of 10 trials (YMMV), 2.9 million coinflips would show a bigger gap. Any factor affecting 200+ votes could change the result.

Coleman is going balls to the wall and Franken should too, but no one should take it too seriously. I'd say that it's 50-50 right now if everything is done honestly (according to the various ways of doing it honestly), or maybe 60-40 Coleman.

I won't be additionally upset if the recount sends Coleman to the Senate. I'm already upset at Minnesota's stupid voters and the local media (and Senator Klobuchar), and I expect the Republicans to stink up the place. If the Senate arbitrarily decides to seat Franken in the end, that will be as fair as most of the other possible outcomes, and they should do it if they can.


Posted by: John Emerson | Link to this comment | 11-29-08 8:26 AM
horizontal rule
3

The man-on-the-street sense I get is that pretty much everyone is discounting for a probable Coleman win, eventually. Nobody outside the parties seems that excercised about it though. After all, if they'll kill Wellstone, what won't they do?


Posted by: minneapolitan | Link to this comment | 11-29-08 12:02 PM
horizontal rule
4

Intrade is currently giving Franken a 33% chance, but prices are volatile.


Posted by: pdf23ds | Link to this comment | 11-29-08 12:15 PM
horizontal rule
5

Was Al Franken literally the best Minnesota could do? When you're running against a total douchebag in the midst of a major Democratic wave and you're still in a photo finish, that's prima facie evidence to me that you suck as a candidate and as a person.


Posted by: Adam Kotsko | Link to this comment | 11-29-08 12:24 PM
horizontal rule
6

Intrade is currently giving Franken a 33% chance, but prices are volatile. shadow markets are full of shit.

Fixed.

Do any econ journals determine what to publish based on shadow markets?


Posted by: rob helpy-chalk | Link to this comment | 11-29-08 12:26 PM
horizontal rule
7

Franken had a serious problem with a lot of old, off-color comedy routines. Coleman ran a vigorous campaign which was, as I remember, 90% character assassination. Not criticism of Franken on the issues, but personal criticism which was in some cases completely ungrounded. It was a perfect Rovian campaign, because most voters ended up concluding "it's a dirty campaign, but both sides are equally to blame".

Wobegon has a more liberal than average Democratic Party and a more conservative than average Republican Party, but there are a fair number of old New Deal Democrat voters who are socially conservative, and the level of cynicism in the state is deplorably low.


Posted by: John Emerson | Link to this comment | 11-29-08 12:41 PM
horizontal rule
8

Why is it that when people disagree with me, they always do so with such animosity towards my position? Wait, don't answer that.


Posted by: pdf23ds | Link to this comment | 11-29-08 12:56 PM
horizontal rule
9

Think about the different remedies for ignorance we employ. Hypothesis testing and model building attempt to remove doubt by putting our questions in a form that can be checked against outside information and then bringing in new information. Meta-analysis and peer review are essentially methods of checking the reasoning that has been used on existing information.

Shadow markets don't check reasoning or produce data, the gauge people's confidence, largely a product of individual personalities. Rather than overcoming bias with reason and data, they merely aggregate bias arbitrarily.


Posted by: rob helpy-chalk | Link to this comment | 11-29-08 1:10 PM
horizontal rule
10

They don't aggregate bias, they aggregate biased knowledge, and all knowledge is biased, no matter what the source. Intrade prices will be susceptible to systemic biases, but so will any other aggregation method you care to name.


Posted by: pdf23ds | Link to this comment | 11-29-08 1:32 PM
horizontal rule
11

BTW, where are you getting "shadow market"?


Posted by: | Link to this comment | 11-29-08 1:33 PM
horizontal rule
12

Is that not the right term? I just meant these toy markets people set up to make predictions of empirical phenomena.


Posted by: rob helpy-chalk | Link to this comment | 11-29-08 2:31 PM
horizontal rule
13

Knowledge isn't biased; knowers are biased.

Granted, any individual is going to have both knowledge and bias. More precisely, their beliefs will be derived both from reliable mechanisms and unreliable mechanisms that reflect non-epistemic idiosyncrasies. When we aggregate people's beliefs, we can either do so using methods that try to filter out beliefs from unreliable sources, or we can just smoosh all beliefs together as is. Markets do the latter.

I think it is weird that markets are considered to have epistemic utility when nothing in a market involves justification or reliability.


Posted by: rob helpy-chalk | Link to this comment | 11-29-08 2:35 PM
horizontal rule
14

all knowledge is biased.

Thanks, Foucault. So the solution is simply to say "fuck it" and just average everything, rather than trying to see which pieces of knowledge are less biased?


Posted by: F | Link to this comment | 11-29-08 2:38 PM
horizontal rule
15

Thank god we don't rely on prediction markets for, say, gravity. All kinds of things would be floating around.


Posted by: F | Link to this comment | 11-29-08 2:39 PM
horizontal rule
16

The term I usually use, and which Robin Hanson (the guy who came up with prediction markets) uses is "prediction markets". Another one is "ideas futures". I don't know precisely what a "shadow market" is, but it seems refer to markets that are of shady legality or illegal, which would probably currently include Intrade in the US.

we can either do so using methods that try to filter out beliefs from unreliable sources, or we can just smoosh all beliefs together as is. Markets do the latter.

Well, to some extent, yes. But they do have one mechanism to filter out the less reliable sources--if you're not good at predicting you lose money. So to the extent that people learn from their mistakes in the market, they either improve or leave the market. Advocates of prediction markets (of which I'm a lukewarm one) say that this mechanism is strong enough to make the markets quite useful.

And markets aren't terribly rational. And that's because they're made up of ... people. Like you. You think you're smarter than the market? Then feel free to ignore it, or better yet, make money off of it.

It seems like the phenomenon of bubbles are a counterexample. Have bubbles happened in prediction markets?

It seems there ought be some good way of betting on a bubble popping and thereby stealing its momentum. If there were a good way to do that, could the real estate bubble have been averted? Is there some sort of instrument we're missing, or is it a truly systemic problem?

Knowledge isn't biased; knowers are biased.

All knowledge has an intimate and inextricable connection with bias. How's that?

I'm taking a subjectivist, Bayesian view of knowledge here, if that helps any. In my use of the world "knowledge", it's is somewhat interchangable with "belief". I use "knowledge" to mean "fairly certain belief", and "belief" either as "uncertain belief", or "certain or uncertain belief". In fact, I think a view of knowledge in which a thing is only knowledge if it's true is a bit naive. We don't have unmediated access to the external world, so we can never judge a belief to be true, except probabilistically. And since beliefs are nothing more than probabilities, there's really no subjective bright line you can draw between belief and knowledge.

Would you say that a belief can be biased?


Posted by: | Link to this comment | 11-29-08 4:18 PM
horizontal rule
17

While I'm resetting my name cookie, might as well change it.


Posted by: paranoid android | Link to this comment | 11-29-08 4:20 PM
horizontal rule
18

Most of the challenges are completely stupid and make you want to punch one of the candidates in the nuts for trying to subvert democracy or make this harder than it needs to be.

You know, the candidates themselves are not the ones issuing the challenges, so I'm not sure violence towards them would be productive, or even necessarily warranted.

As I understand it, challenges are issued by when two people, one from each campaign, are hand counting as fast as they can for hours, and the "challenge" pile tends to be the odd or weird. Again, from what I understand, "challenge" typically isn't a big deal involving lawyers, courts, judges and all that. It is, for the most part, a pile one comes back to later for a second look.

In other words, democracy is proceeding just fine.


Posted by: michael | Link to this comment | 11-29-08 4:37 PM
horizontal rule
19

We don't have unmediated access to the external world, so we can never judge a belief to be true, except probabilistically.

Eek!

And since beliefs are nothing more than probabilities, there's really no subjective bright line you can draw between belief and knowledge.

Eek!!


Posted by: parsimon | Link to this comment | 11-29-08 5:20 PM
horizontal rule
20

Off-topic!

So I have a faculty-job interview this week, which I was doing fine prepping for. My best friend from high-school era (who is still a very good friend, and lives close to the uni) has chosen this moment to throw an unprecedented royal snit about something my new girlfriend (who he just met for the first time) did to viciously slight him, which is entirely in his head, but he's spitting mad and is refusing to see me.

I'm very upset and can't focus on my interview prep! Two big lectures! How can I calm down and focus?


Posted by: Lyndon Baines Johnson | Link to this comment | 11-29-08 5:25 PM
horizontal rule
21

19: those two statements appear to me to be sort of obvious, hard to take issue with, but I realize milennia of philosophers have probably disagreed with me.

Is there some sort of instrument we're missing, or is it a truly systemic problem?

truly systemic problem.

OT, but does anybody know of any good foreign policy / strategy type sites to follow what's going on in India? I'd say after this there is a strong chance of an Indian strike back, possibly leading to escalation that could even reach the nuclear level.


Posted by: PGD | Link to this comment | 11-29-08 5:31 PM
horizontal rule
22



Posted by: | Link to this comment | 11-29-08 5:31 PM
horizontal rule
23

20: depends entirely on your psychology. If it were me, I'd just tell the friend I couldn't deal with this now and don't want to hear about until after the interview, then put it out of my mind. But you're not me. And perhaps related to that, you probably have more friends too -- my ability to ignore others when necessary is pretty highly developed.


Posted by: PGD | Link to this comment | 11-29-08 5:34 PM
horizontal rule
24

20: Deep breathing/deep relaxation, yoga-style. Lie down on the floor (on a warm rug!) in a quiet room, low lights, no music or anything. The idea is to empty your mind, accept what is at the moment. Focus on your body -- notice whether one of your legs is extending longer than the other. Whether you have one shoulder hitched up higher than the other; whether you're arching your neck. Accept these things, and casually correct them (not in a fidgety manner), then notice the correction.

Shall I go on? Maybe not. That's what I'd do.


Posted by: parsimon | Link to this comment | 11-29-08 5:41 PM
horizontal rule
25

21: those two statements appear to me to be sort of obvious, hard to take issue with, but I realize milennia of philosophers have probably disagreed with me.

Oh, you don't need the millennia of philosophers. Truth isn't probabilistic except in a certain very limited sense. Of course, you're talking economics, about which I know nothing, so my comment was insolent if nothing else.


Posted by: parsimon | Link to this comment | 11-29-08 5:59 PM
horizontal rule
26

The statement "beliefs are probabilities" needs some unpacking.


Posted by: ben w-lfs-n | Link to this comment | 11-29-08 6:03 PM
horizontal rule
27

20: Role play a conversation in your head with your best friend. In this role play you are calm, you don't over-explain your position, you validate his feelings but don't give ground on your own.

Once you've done that, do some of 24, then get back to work. Don't have the conversation with your best friend until you need to, and under no circumstances before your interview.

Alternatively: Tell your best friend to come over and have sex with you and your new girlfriend. I think that's how they worked it out in Chasing Amy.


Posted by: Wrongshore | Link to this comment | 11-29-08 6:06 PM
horizontal rule
28

26: A belief is a proposition plus a strength, or probability. You believe the proposition, given your other beliefs, including the evidence for the proposition, will be true with some probability. Every belief has a strength. And in one sense, the word "probability" can include the subject of the probability. In that sense, beliefs are probabilities.


Posted by: pdf23ds | Link to this comment | 11-29-08 6:35 PM
horizontal rule
29

(There is a person over there, 0.6) is a proposition plus a probability.


Posted by: ben w-lfs-n | Link to this comment | 11-29-08 6:38 PM
horizontal rule
30

26: The whole thing needs unpacking.


Posted by: parsimon | Link to this comment | 11-29-08 6:41 PM
horizontal rule
31

Right, and that's a valid belief. You might come across that pair if you heard a noise across the room that you thought might be a person but haven't turned around to look yet.


Posted by: pdf23ds | Link to this comment | 11-29-08 6:41 PM
horizontal rule
32

Cookies, yum.


Posted by: paranoid android | Link to this comment | 11-29-08 6:43 PM
horizontal rule
33

They say now that Somali pirates are being recruited in Wobegon.


Posted by: John Emerson | Link to this comment | 11-29-08 7:00 PM
horizontal rule
34

And since beliefs are nothing more than probabilities, there's really no subjective bright line you can draw between belief and knowledge.

Knowledge is the intersection between belief and reality.


Posted by: Gay Org Wilhelm F. Hegel | Link to this comment | 11-29-08 7:03 PM
horizontal rule
35

The prediction markets don't just aggregate biases, they offset biases, where sellers cancel out buyers,
unless a great swath of participants are using the market to hedge (eg, bet on McCain not because you expect McCain to win, but because wining would pay you relaction-to-France costs).

Question for Bay Areans: How did the BART vote conclude ?



Posted by: Econolicious | Link to this comment | 11-29-08 7:09 PM
horizontal rule
36
Posted by: | Link to this comment | 11-29-08 7:10 PM
horizontal rule
37

Oh, bleah. It wasn't clear to me who wrote 16, actually, but no matter: the elision between knowledge and belief, the suggestion that knowledge would only be possible given unmediated access to the external world, the consequent declaration that knowledge/belief are at best probabilistic: Really, now.

This defines knowledge as an impossibility from the outset; it's a form of skepticism, an outgrowth of mind/body dualism. So there. Bleah.


Posted by: parsimon | Link to this comment | 11-29-08 7:17 PM
horizontal rule
38

37 to 31, 29, 28.


Posted by: parsimon | Link to this comment | 11-29-08 7:19 PM
horizontal rule
39

This defines knowledge as an impossibility from the outset; it's a form of skepticism, an outgrowth of mind/body dualism.

So?


Posted by: teofilo | Link to this comment | 11-29-08 7:29 PM
horizontal rule
40

Because skepticism is stupid.


Posted by: John Emerson | Link to this comment | 11-29-08 7:41 PM
horizontal rule
41

So, if you say something like (say) "Our breakup, between you and me, was very difficult" you are making at best a probabilistic statement.

The problem is that this divorces the way in which we operate with truth and knowledge from the way in which we actually do operate with truth and knowledge statements.

Of course we adhere to certain standards of evidence, and we have what we call reasons for believing and knowing the things we do. The point is that somewhere along the line, those reasons are sufficient to claim knowledge. To insist that we're still dealing with nothing better than probabilities is essentially to say that we're fundamentally divorced from the world. This is a radical claim. It's also wrong.


Posted by: parsimon | Link to this comment | 11-29-08 7:46 PM
horizontal rule
42

It will be obvious that this subject matter engages me. But it's a Saturday night, and I worked today, so I think I will listen to some old-timey music, chill time.


Posted by: parsimon | Link to this comment | 11-29-08 8:11 PM
horizontal rule
43

||

Also, my oh-most-friend, a member of my true family, is awaiting the birth of his and his loved one's child. As we speak. I am worried. I know this man like my own heart, and I .. hope it all goes well. My heart leaps for him (them), then fear, worry.

|>


Posted by: parsimon | Link to this comment | 11-29-08 8:25 PM
horizontal rule
44

the suggestion that knowledge would only be possible given unmediated access to the external world

Perhaps we need to consider what knowledge really is, and not worry so much about whether something is knowledge or not. If we believe something with probability .99999 (as we do many trivial statements), is it not certain enough to consider knowledge for any reasonable purpose? I don't mean to be a skeptic at all, I just think that there's no such thing as justified absolute certainty. Not even for mathematical truths.

And WTH does that have to do with mind/body dualism? :P

So, if you say something like (say) "Our breakup, between you and me, was very difficult" you are making at best a probabilistic statement.

One with a very high probability. And more importantly, in this case, one that no one else (unless they could read your mind directly) can make with any more authority about you, since you have privileged access (to some degree) to your own mental states. Yes, that's knowledge, but you could conceivably be wrong, as you can about any fact. (If you define truth about someone's mental state to be self-report then it can't, but that's a dubious definition). The example gets into cogsci questions that we can't even begin to answer right now. Might I suggest a less fraught example?

The problem is that this divorces the way in which we operate with truth and knowledge from the way in which we actually do operate with truth and knowledge statements.

More like, "from the way in which usually consciously model our own internal processes regarding truth and knowledge statements". I'd assert our internal processes themselves more closely resemble my description than the position I'm arguing against. (Whose?) Folk psychology is not real psychology.

To insist that we're still dealing with nothing better than probabilities is essentially to say that we're fundamentally divorced from the world. This is a radical claim. It's also wrong.

Daniel Dennett must be radical and wrong, then. Poor guy.

It only claims that we don't have unmediated, non-error-prone access to the world. Photons can do funny things, you know, and all eyesight is is a photon sensor and a whole huge heap of algorithms to processes the signal.

My position might require a naturalistic view of the world, I'm not sure. But that's not terribly radical either.


Posted by: paranoid android | Link to this comment | 11-29-08 8:48 PM
horizontal rule
45

There's way too much material here, pdf. Yes, my example was fraught. And Dennett I'm not up to date with, but no sneeze, he.

We're probably talking past each other. Certainly we don't have unmediated, non-error-prone access to the world. My question is why we would think there could be such a thing in the first place.


Posted by: parsimon | Link to this comment | 11-29-08 8:59 PM
horizontal rule
46

Oh boy, it's the argument from illusion!


Posted by: ben w-lfs-n | Link to this comment | 11-29-08 9:23 PM
horizontal rule
47
Posted by: | Link to this comment | 11-29-08 9:29 PM
horizontal rule
48

I just think that there's no such thing as justified absolute certainty.

See, this is just silly. I went to the movie Australia tonight. That's not a statement of 80% probability, of 90% probabliity, or of 99.9% probability.

The wife thinks Ms. Kidman is too perfect, too cold. That's a statement of belief, but about opinion, not fact. Again, the probability model has nothing to tell us here.


Posted by: CharleyCarp | Link to this comment | 11-29-08 9:48 PM
horizontal rule
49

46: Yeah. I was beginning to have a conniption earlier, but 's'okay.


Posted by: parsimon | Link to this comment | 11-29-08 9:50 PM
horizontal rule
50

Charley, you and your wife could be brains in a vat.


Posted by: parsimon | Link to this comment | 11-29-08 10:12 PM
horizontal rule
51

50 -- I know a couple of very skilled tailors who can make you a great magical suit of clothes.


Posted by: CharleyCarp | Link to this comment | 11-29-08 10:22 PM
horizontal rule
52

See, this is just silly.

Oh, I dunno. Depends on what is is, I guess. For the purposes of living together in relative comity in a shared universe of agreed-upon meaning: yeah, just silly, sure. But if you wished to step outside those parameters: well, maybe not "just silly," after all? but then again, who wants to go there, and why?

Nicole Kidman is one cool customer, to be sure. But multiple botox treatments do not necessarily enhance one's ability to effectively emote, maybe.


Posted by: | Link to this comment | 11-29-08 10:22 PM
horizontal rule
53

48: I'm really not interested in arguing that point.

46: Sophist.


Posted by: paranoid android | Link to this comment | 11-29-08 10:22 PM
horizontal rule
54

Oops! 52 = me. And 50 = ha!


Posted by: Mary Catherine | Link to this comment | 11-29-08 10:24 PM
horizontal rule
55
Posted by: | Link to this comment | 11-29-08 10:25 PM
horizontal rule
56

51: ??

Do I need one? I'd love one, but I think I'm missing your point.


Posted by: parsimon | Link to this comment | 11-29-08 10:31 PM
horizontal rule
57

This cloth, they tell me, is invisible to anyone who was either stupid or unfit for his/her position.


Posted by: CharleyCarp | Link to this comment | 11-29-08 10:38 PM
horizontal rule
58

Men han har jo ikke noget paa!


Posted by: et lille Barn | Link to this comment | 11-29-08 10:39 PM
horizontal rule
59

In fact, I think a view of knowledge in which a thing is only knowledge if it's true is a bit naive. We don't have unmediated access to the external world, so we can never judge a belief to be true, except probabilistically. And since beliefs are nothing more than probabilities, there's really no subjective bright line you can draw between belief and knowledge.

I think this is mixing up some true things and some false things. Suppose that Parsimon believes that Obama is president-elect, and believes it on the same evidence that most of us have. And suppose for the sake of argument that this means that she has good reason for her belief, even though her evidence doesn't make it impossible that her belief is false, but just makes its falsity extremely unlikely. It's just possible that the election-coverage was a giant scam organized byThe Onion, but it's very unlikely. So she's making a justified probabilistic judgement that Obama is president-elect.

None of this makes the idea that knowledge requires truth naive. All it means (Gettier issues aside) is that if Obama is in fact president-elect, then Parsimon knows it (because she has a justified true belief), and that if he in fact isn't, she doesn't (because she has a justified but false belief). The bright line between belief and knowledge isn't subjective, its objective.


Posted by: Amit | Link to this comment | 11-29-08 10:40 PM
horizontal rule
60

48: and here is my other hand, thus proving the existence of multiple external objects.


Posted by: Keir | Link to this comment | 11-29-08 10:50 PM
horizontal rule
61

58 to 60.


Posted by: CharleyCarp | Link to this comment | 11-29-08 10:54 PM
horizontal rule
62

It's just possible that the election-coverage was a giant scam organized by the Onion, but it's very unlikely.

I suppose there are constructions where pretending that this might be non-gibberish have some use. Sort of like how in math it might be useful to think of multiple dimensions. In the world in which humans actually live, though, it is gibberish. The probability that the Onion created the general election is zero.


Posted by: CharleyCarp | Link to this comment | 11-29-08 11:04 PM
horizontal rule
63

59: You see, I would much prefer to talk about "true belief" in those circumstances instead of "knowledge". Defining knowledge is such a can of worms; do we really need to open it?


Posted by: paranoid android | Link to this comment | 11-29-08 11:08 PM
horizontal rule
64

There's a difference between true belief and justified true belief (+blahblah gettier ho ho). so if you mean "knowledge", best to eat some worms.


Posted by: Cala | Link to this comment | 11-29-08 11:10 PM
horizontal rule
65

57: I'm intrigued.

You all know I'm just talking about philosophical idealism versus realism. pdf is on about a form of idealism (which has nothing to do with ideals) according to which mental constructs, be they ideas, or sense-data, percepts, concepts, are the only things to which we have unmediated access, and they are our world. Since we can't get outside our heads in order to actually touch the external world, there is no absolute (no true) knowledge of that world, only belief, justified to a high degree of probability, that there's a correspondence between our ideas/percepts/concepts and the external world those things purport to represent.

There's always that vanishingly small possibility that we're just brains in a vat, though. You never know.

I have to shrug. Whatever version of this you adopt, it ignores the ways in which we actually do judge things true and false.


Posted by: parsimon | Link to this comment | 11-29-08 11:12 PM
horizontal rule
66

"it ignores the ways in which we actually do judge things true and false."

Just because we completely ignore things with a very small probability doesn't mean my model is inaccurate. In fact, we do process things with very small probabilities, but they just never reach consciousness before being discarded for incoherency. When we dream this mechanism (and some others) are disabled, and we can contemplate self-evidently absurd things without noticing the absurdity.

W-lfs-n, should that "are disabled" in the last sentence be "is disabled"? I never know what to do when a parenthetical expression changes the number of a verb.


Posted by: paranoid android | Link to this comment | 11-29-08 11:18 PM
horizontal rule
67

Also, I take offense at being identified with idealism. As far as I understand the term, I'm no idealist.


Posted by: paranoid android | Link to this comment | 11-29-08 11:19 PM
horizontal rule
68

I scribbled the above, 65, thinking that some context and background was needed, but on reflection, well, no more or less helpful than anything else. There is no straightforward way through this issue: there are not two clear positions, one or the other of which is right. Both are right. The question/issue is ill-formed. This is not an original thought.


Posted by: parsimon | Link to this comment | 11-29-08 11:20 PM
horizontal rule
69

This is not an original thought.

Oh, to think an original thought! How sweet a reward!

(Supply is dwindling.)


Posted by: paranoid android | Link to this comment | 11-29-08 11:27 PM
horizontal rule
70

There are all sorts of original thoughts. On this particular topic, though, not so much: it's been done to death. Hence my impatience and sloppiness in addressing it.

As for your grammar question, though I am not W-lfs-n.

should that "are disabled" in the last sentence be "is disabled"? I never know what to do when a parenthetical expression changes the number of a verb.

The confusion over how it should read means you should rewrite it. Instead of

When we dream this mechanism (and some others) are disabled, write:

"When we dream, this (and some other) mechanisms are disabled ..."

This makes it clear that the parenthesis could be replaced by commas: "this, and other, mechanisms" -- the subject is plural. The "some" in "some other(s)" is extraneous, unless you have a point to make about the mechanisms which are not disabled in dreaming.


Posted by: parsimon | Link to this comment | 11-29-08 11:46 PM
horizontal rule
71

"When we dream, this (and some other) mechanisms are disabled ..."

This doesn't actually change anything; you just shifted the plurality from inside the parentheses to outside. Take out the parenthetical expression entirely and you'll see that it's still ungrammatical.

This is a frequent problem in using parenthetical expressions in English, and there isn't really a good solution to it short of abandoning any attempt to include the parenthesis within the phrase. So, "this mechanism is disabled (as are some others)" would work, but it's not very elegant.

In other news, I just finished the first draft of my personal statement. Go me.


Posted by: teofilo | Link to this comment | 11-30-08 12:05 AM
horizontal rule
72

So, "this mechanism is disabled (as are some others)" would work, but it's not very elegant.

Heh. Yes, "this mechanism is disabled. Others are as well, but that's beside the point, so I don't know why I'm mentioning it, but that's why I put it in parentheses."

Snort.


Posted by: parsimon | Link to this comment | 11-30-08 12:17 AM
horizontal rule
73

I just finished the first draft of my personal statement.

Woo! Congrats! I hope it includes original thoughts.


Posted by: Stanley | Link to this comment | 11-30-08 12:25 AM
horizontal rule
74

pdf is on about a form of idealism ... according to which mental constructs, be they ideas, or sense-data, percepts, concepts, are the only things to which we have unmediated access, and they are our world. Since we can't get outside our heads in order to actually touch the external world, there is no absolute (no true) knowledge of that world, only belief, justified to a high degree of probability

Not if the external world is just ideas, too.


Posted by: Bishop George Berkeley | Link to this comment | 11-30-08 2:47 AM
horizontal rule
75

What is the advantage gained by saying that some x is 99.99999 [...] 9 % likely, rather than certain? Just honoring a taboo?

At my URL is my take on the idea that the laws of thermodynamics are merely statistical probabilities. Nut:

The reason why we ignore this chance may be seen by a rather classical illustration.... If an army of monkeys were strumming on typewriters they might write all the books in the British Museum. The chance of their doing so is decidedly more favorable than the chance of the molecules returning to one half of the vessel.... (Eddington, Arthur, The Nature of the Physical World, Gifford Lectures, 1928, p. 72.).....
A physics professor at Yale, William R. Bennett, has calculated that if a trillion monkeys typed ten random characters a second, it would still take a trillion times longer than the universe has been in existence just to produce the sentence, "To be or not to be, that is the question."
Moving from calculation to experiment, The Monkey Shakespeare Simulator, in existence since 2003 with a hundred monkeys typing at a vastly accelerated speed, has produced just nineteen letters from The Two Gentlemen of Verona after 42,162,500,000 billion monkey years: "Valentine. Cease to 1dor:eFLPoFRjWK78aXz."

Posted by: John Emerson | Link to this comment | 11-30-08 5:31 AM
horizontal rule
76

But if you did the simulation on the iPhone where there are multiple possibilities that would be autocorrected to the right word, I bet the time would be much shorter.


Posted by: SP | Link to this comment | 11-30-08 6:31 AM
horizontal rule
77

iPhone: reducing possibilities for the world since 2007.


Posted by: JP Stormcrow | Link to this comment | 11-30-08 6:43 AM
horizontal rule
78

It seems very likely that I just finished the first draft of my personal statement.

Corrected for the benefit of the brains in vats contingent. Best of luck with it. If indeed you've done it. And if indeed there are grad schools.


Posted by: CharleyCarp | Link to this comment | 11-30-08 7:07 AM
horizontal rule
79

i just saw her photo and thought that Britney Spears looks more like Marylin Monroe, not Lyndsay Lohan
b/c she is growing more mature perhaps


Posted by: read | Link to this comment | 11-30-08 7:22 AM
horizontal rule
80

Now that's idealism.


Posted by: CharleyCarp | Link to this comment | 11-30-08 7:24 AM
horizontal rule
81

Parsley is describing the behavior of human beings. Paranoid PDF is describing the behavior of spam filters.

No doubt, spam filters model some aspects of human cognition. In fact, they perform better than humans at some tasks, like filtering spam. But that doesn't mean that spam filters model everything humans do when we walk around knowing things.


Posted by: rob helpy-chalk | Link to this comment | 11-30-08 7:28 AM
horizontal rule
82

Parsley is describing the behavior of human beings disabled by education.

Years ago, my daughter had a social studies teacher, a haughty European, who proclaimed that every place in America was just like every other place. Any squirrel can tell a difference between Atlanta and Albany, especially this time of year, and we ended up call the teacher, around here, Ms. Squirrel Brain.

I'd be interested to hear, from any member of the brains in vats contingent, some human purpose for pretending not to know what any squirrel knows (that it exists, and that the nut it is carrying is external).


Posted by: CharleyCarp | Link to this comment | 11-30-08 7:39 AM
horizontal rule
83

What you should be wondering is why we pretend that you exist, Charley.


Posted by: Brains in Vats | Link to this comment | 11-30-08 7:47 AM
horizontal rule
84

82: I didn't think our leafy, book-selling friend was denying ordinary squirrel-knowledge. That seemed more what pdf was doing, by insisting that we assign some probability to the existence of our nuts.


Posted by: rob helpy-chalk | Link to this comment | 11-30-08 7:51 AM
horizontal rule
85

Everymall in America is pretty much the same as any other mall.


Posted by: John Emerson | Link to this comment | 11-30-08 7:57 AM
horizontal rule
86

"There is a difference between Albany and Atlanta" is currently running at 99.999999741 at Intrade. Good money to be had for those with the courage of their convictions.


Posted by: JP Stormcrow | Link to this comment | 11-30-08 8:00 AM
horizontal rule
87

"Everymall": A trope in early 21st century American aspirational economic literature, in which a relatively unexceptional shopping mall serves as the unlikely locus of the recovery of the economy. First used in a series of articles written by "Everybeard", later revealed to be a pseudonym for Thomas Friedman.


Posted by: JP Stormcrow | Link to this comment | 11-30-08 8:12 AM
horizontal rule
88

The irony is that it would actually be running at about 97%, because with fees it doesn't make sense for anyone to buy above that level. So even a metaphysical certainty would never be more than 97% likely according to the prediction markets.


Posted by: SP | Link to this comment | 11-30-08 8:13 AM
horizontal rule
89

85 -- Any junior high school kid can tell the difference between Georgetown Park and Potomac Mills.


Posted by: CharleyCarp | Link to this comment | 11-30-08 8:16 AM
horizontal rule
90

88: Ironically A = A is trading at 95%, because of fees and a persistent, Hegelian spambot persuading people to buy A != A.


Posted by: rob helpy-chalk | Link to this comment | 11-30-08 8:24 AM
horizontal rule
91

Everymall in America is pretty much the same as any other mall.

Given Emerson's location, he may be forgiven for forgetting the two mall rule that obtains elsewhere.


Posted by: apostropher | Link to this comment | 11-30-08 8:35 AM
horizontal rule
92

Plaxico Burress missed a game because of his injured leg, so he pulled out his gun and shot it.

In his own defense he was quoted as saying "Plaxico Burress knows how to handle guns safely. Plaxico Burress does not have firearms accidents. That fucking leg had it coming. It's damn lucky I didn't shoot it twice."


Posted by: John Emerson | Link to this comment | 11-30-08 9:04 AM
horizontal rule
93

Like a mantis. The world of the mantis is not like the world of the unmantis.


Posted by: John Emerson | Link to this comment | 11-30-08 9:17 AM
horizontal rule
94

20: As suggested by others, can you tell your friend that you can't even think about this until your current professional crisis is passed?

Once that's happened, I'd think about looking into whatever he's mad about under the assumption that he isn't necessarily being an unjustified jerk. In an old-friend v. new-girlfriend conflict, I'd guess that the odds that the old friend is acting out of character is lower than the chance that the new girlfriend is behaving badly, just because you have a longer baseline of experience with the first than with the second.


Posted by: LizardBreath | Link to this comment | 11-30-08 9:29 AM
horizontal rule
95

some human purpose for pretending not to know what any squirrel knows

I know, and every squirrel knows, that my dogs cannot catch a squirrel. Yet I pretend that my dogs can catch a squirrel, out of compassion and empathy.

But I am not a a brain in a vat. Or at least my dogs pretend I am not a brain in a vat.


Posted by: bob mcmanus | Link to this comment | 11-30-08 9:42 AM
horizontal rule
96

What the Squirrel Knew could be the name of a philosophical murder mystery.


Posted by: rob helpy-chalk | Link to this comment | 11-30-08 9:44 AM
horizontal rule
97

A friend's dog caught and killed a squirrel, and then was sad because he'd only wanted to play.


Posted by: John Emerson | Link to this comment | 11-30-08 9:48 AM
horizontal rule
98

I'm really surprised how silly people think it is to say that there are no absolute certainties. I'm not interested in arguing about it because it's been done. Someone read that and come back and argue with me, if you will.


Posted by: paranoid android | Link to this comment | 11-30-08 10:04 AM
horizontal rule
99

What I was doubting is the usefulness and meaningfulness of saying that something is 99.99...9% certain, especially when the statistic does not report any measurable quantity, and the 0.000...1% remainder is nothing very specific.

It's like a peculiarity of notation, or a concessio n to a taboo. It reminds me of a primitive hand calculator that reported 2 X 2 = 3.99999999.........

As I said, I think that the use of thermodynamic probability terminology in natural language leads to unhelpful statement. Something that happens once in a trillion years in one of a trillion universes can be called "merely very unlikely", but in natural language that's the same as impossible.


Posted by: John Emerson | Link to this comment | 11-30-08 10:13 AM
horizontal rule
100

I think you misunderstand us, PA. We all agree that there are no absolute certainties in the sense that fool proof arguments can't be developed against Cartesian skeptics and brain in a vat scenarios. The difference between us is that we respond to these arguments by lowering the bar for what counts as knowledge. You don't have to rule out the brain in the vat scenario to know things.

Your view seems to work like this:

"No absolute certainties" --> "every belief is a probability assigned to a statement in a Bayesian machine"

But the more mainstream philosophical answer is:

"No absolute certainties" --> knowledge need neither be certain nor absolute.

Notice how much more conservative the latter implication is.


Posted by: rob helpy-chalk | Link to this comment | 11-30-08 10:13 AM
horizontal rule
101

What about upthread when I said I use the word knowledge differently, to mean "beliefs with high probability"? Like I said repeatedly, I have a slightly different conception of knowledge, lowering the bar for what counts. I think I agree with both of your statements, and don't see a contradiction between them.


Posted by: paranoid android | Link to this comment | 11-30-08 10:17 AM
horizontal rule
102

101: I use natural language the same way most people do. Things that have a chance less than .0001 (or in some circumstances, .05) are impossible, things more certain than .9999 are simply "certain". It's just when you ask me what exactly I mean by those words that I might give different answers.


Posted by: paranoid android | Link to this comment | 11-30-08 10:19 AM
horizontal rule
103

fool proof arguments can't be developed

It's not the lack of arguments but the surfeit of fools.


Posted by: CharleyCarp | Link to this comment | 11-30-08 10:22 AM
horizontal rule
104

really what is at stake are not the two implications I listed, but their consequents. The two consequents are compatible, because the latter is much weaker than the former.

"every belief is a probability assigned to a statement in a Bayesian machine" --> "knowledge need neither be certain nor absolute."

but not vice versa by a long shot.

I am merely advocating sticking to the weaker conclusion.


Posted by: rob helpy-chalk | Link to this comment | 11-30-08 10:24 AM
horizontal rule
105

OK, well I came to the first conclusion using a much wider range of reasoning I haven't even touched on in this thread. In this thread I've been taking it as a given.


Posted by: paranoid android | Link to this comment | 11-30-08 10:27 AM
horizontal rule
106

I can imagine cases when your assignments of percentages might have specific meanings (e.g. a certain asteroid impact), but they're really few. The notation strikes me as useless.


Posted by: John Emerson | Link to this comment | 11-30-08 10:30 AM
horizontal rule
107

108: It usually is. It's useful when programming Bayesian nets of various sorts, or AIs.


Posted by: paranoid android | Link to this comment | 11-30-08 10:35 AM
horizontal rule
108

Even outside the .99999 type of probability where the issue is whether you're absolutely certain or not, I think assigning mathematical probabilities is often a mistake. I see people making arguments of roughly the form: "I'm pretty sure X is going to happen. Call it a probability of .8. And then say that if X happens, Y is probably going to happen (say, p=.8 for that too), and if Y happens, Z will likely happen as well (with p=.8 there too.) Counterintuitively, while I'm pretty sure all of X, Y, and Z will happen, the mathematics demonstrates that the chance that all three will actually happen is less than 1/2. Isn't math wonderful, how it proves things you wouldn't expect?"

And of course an argument like that is nonsense. While you can pull a numerical probability equivalent for "pretty likely" out of your ass and not be importantly wrong if you stop there, once you're doing math with it, your initial errors compound. And simpleminded probability calcuations that apply to events that aren't correlated in any way with each other are not likely to be a good way to approach real-world situations -- if you're thinking about X, Y, and Z as a series of real world events, there's a good shot their outcomes are correlated.

I'd like people to quit oversimplified use of probability math in situations where they haven't really thought through the underlying assumptions.


Posted by: LizardBreath | Link to this comment | 11-30-08 10:40 AM
horizontal rule
109

So if you say 99.999999% in one case and 99.99999999% in the other, you're expressing a precise distinction? I can easily imagine cases in which that would be true, but they'd be fairly artificial cases involving large urns with one white ball, or chemical reactions in extremely dilute solutions.


Posted by: John Emerson | Link to this comment | 11-30-08 10:42 AM
horizontal rule
110

84: I didn't think our leafy, book-selling friend was denying ordinary squirrel-knowledge.

No, I wasn't. Charley seems to have misunderstood something about my (increasingly shouty, sorry about that) comments last night.

106 and 102 are good -- thanks, rob.


Posted by: parsimon | Link to this comment | 11-30-08 11:41 AM
horizontal rule
111

It usually is. It's useful when programming Bayesian nets of various sorts, or AIs.

But this (and Bayesianism in general) is a really flawed approach to a general model of knowledge. It is a very specific statistical view with some implications that don't work well in all cases. In many practical cases, it's a good approach to parametric modeling, and has the advantage of giving a systematic view. This view doesn't generalize so well though (hence the continued usefulness and popularity of non-Bayesian approaches).


Posted by: | Link to this comment | 11-30-08 12:04 PM
horizontal rule
112

It's kind of striking how personality predisposed you to believe certain things. For example, I would have assigned a probability of 0.9999 that PA would think knowledge could be generally modeled by probabilities, and 0.95 that he liked Overcoming Bias. I would put the actual probability that knowledge can be described by probabilities around 0.05, and I think the person who writes Overcoming Bias is a crazy person.

I notice this frequently in arguments -- I'm arguing with somebody, and the argument gets bogged down at the point where it becomes clear that they're just the kind of person who thinks X, and I'm just the kind of person who thinks not-X. It's as if most ideas are sorta true, and what gets us across the gap to belief is purely a function of personality.

What's really surprising to me is how much ideas seem to come in packages (which is why I would have guessed PA's opinion). Sometimes these packages are purely tribal, as in Berube's "Since 9/11 I'm outraged by Chappaquiddick", but frequently there is some sort of greater commonality.


Posted by: Walt Someguy | Link to this comment | 11-30-08 12:53 PM
horizontal rule
113

114: I'm not sure it's so much personality as it is the collection of the other beliefs that you've already seen me advocate. I mean, sure, personality helps a bit, but I'd say the greater factor is that different sets of beliefs tend to come in big huge chunks. It's why it's so notable when someone seems to hold an unusual mix of them.


Posted by: paranoid android | Link to this comment | 11-30-08 1:13 PM
horizontal rule
114

BTW, I've already linked to OB here in the past on a few occasions. I'd be surprised if this is the first time you've seen me link there.


Posted by: paranoid android | Link to this comment | 11-30-08 1:14 PM
horizontal rule
115

To use the Cold War formulation, Pars, I found you soft on reality-denial. Apologies for any misunderstanding.


Posted by: CharleyCarp | Link to this comment | 11-30-08 1:24 PM
horizontal rule
116

Parsimon is a commie!


Posted by: paranoid android | Link to this comment | 11-30-08 1:26 PM
horizontal rule
117

I think the idea that knowledge or belief can be described by probabilities to be a category mistake. Strength of belief or epistemic position, maybe. But not the belief itself.


Posted by: Cala | Link to this comment | 11-30-08 1:31 PM
horizontal rule
118

116/100 -- Transparent nonsense. Two plus two is four, and fifty three is a prime number. Period. It's a closed system, and the meanings of the concepts compel this result.

I'm willing to entertain the possibility that certain scientific pursuits might require, as a fiction, pretending that two plus two might not be four. One shouldn't confuse useful fiction with reality, however.


Posted by: CharleyCarp | Link to this comment | 11-30-08 1:34 PM
horizontal rule
119

To use the Cold War formulation, Pars, I found you soft on reality-denial.

Hm. Possibly the personality I project here. Certainly I have a complicated response to the issue.


Posted by: parsimon | Link to this comment | 11-30-08 1:48 PM
horizontal rule
120

So were the people concerned about the 1/10^31 chance that the LHC would destroy the universe just opponents of natural language?


Posted by: SP | Link to this comment | 11-30-08 1:49 PM
horizontal rule
121

121: Is that a response to my link?


Posted by: paranoid android | Link to this comment | 11-30-08 1:51 PM
horizontal rule
122

123: Was 1/10^31 the result of an actual mathematical calculation based on established facts and a valid formula? Or was it just mush, like taking two ballpark figures and multiplying them?

Once you start doing exponentiation, all numbers look unimaginably big/small except to someone familiar with the specific field being discussed. Nineteen letters from The Two Gentlemen of Verona after 42,162,500,000 billion monkey years gives you a fair idea of how unlikely Monkey Typewriter Shakespeare would be. "Much more improbable than that" gives you a vague idea of how unlikely the self-evacuation of a liter bottle of air would be. My guess is that 1/10^31 is a tiny number relative to these.


Posted by: John Emerson | Link to this comment | 11-30-08 4:15 PM
horizontal rule
123

125: No, it was a speculative theory.


Posted by: Walt Someguy | Link to this comment | 11-30-08 4:43 PM
horizontal rule
124

I hope it includes original thoughts.

A few too many, I'm afraid. I'll have to cut some of them out as I revise it down to two pages from three.


Posted by: teofilo | Link to this comment | 11-30-08 4:53 PM
horizontal rule
125

The way this thread has developed confirms my initial suspicion that this argument, typically for Unfogged, was ultimately reducible to lexical semantics. In this case the word is "knowledge."


Posted by: teofilo | Link to this comment | 11-30-08 4:57 PM
horizontal rule
126

LB gets it exactly right in 110.

The probability that the LHC will destroy the universe is 0. I'm more worried about the chance that it will turn on at all before 2010....


Posted by: essear | Link to this comment | 11-30-08 5:04 PM
horizontal rule
127

128: We call that epistemology, teo.


Posted by: parsimon | Link to this comment | 11-30-08 5:07 PM
horizontal rule
128

128: That's what I was saying from the beginning, too. Did anyone believe me?


Posted by: paranoid android | Link to this comment | 11-30-08 6:10 PM
horizontal rule
129

Did anyone believe me?

I did.


Posted by: teofilo | Link to this comment | 11-30-08 6:28 PM
horizontal rule
130

In my use of the world "knowledge", it's is somewhat interchangable with "belief".... beliefs are nothing more than probabilities, there's really no subjective bright line you can draw between belief and knowledge.

That's a pretty bold theory of knowledge, and you shouldn't be surprised that it roused resistance.


Posted by: John Emerson | Link to this comment | 11-30-08 6:37 PM
horizontal rule
131

128: I see actual areas of disagreement, at least to the extent that this is an interesting discussion. E.g., I don't think belief is just a probability, and I don't think knowledge is just a really strongly held belief. If you want to say, well, by "knowledge" I just mean "strongly held beliefs with very high probability", that's fine, but it's still open to me to say, well, what about what I meant by "knowledge?"

It's more than just redefining terms.


Posted by: Cala | Link to this comment | 11-30-08 6:49 PM
horizontal rule
132

I see actual areas of disagreement, at least to the extent that this is an interesting discussion.

So do I. But those areas of disagreement still seem to involve defining (not necessarily redefining) words.


Posted by: teofilo | Link to this comment | 11-30-08 6:56 PM
horizontal rule
133

That is, when I say that a dispute is underlyingly about semantics I don't mean to dismiss it as meaningless. Semantics is important.


Posted by: teofilo | Link to this comment | 11-30-08 7:02 PM
horizontal rule
134

A semantic is important.


Posted by: heebie-geebie | Link to this comment | 11-30-08 7:20 PM
horizontal rule
135

I hate Symantec.


Posted by: paranoid android | Link to this comment | 11-30-08 7:24 PM
horizontal rule
136

Semantics be important. Yo.


Posted by: teofilo | Link to this comment | 11-30-08 7:25 PM
horizontal rule
137

Some antics are unimportant, though.


Posted by: Stanley | Link to this comment | 11-30-08 7:29 PM
horizontal rule
138

We still matter to ourselves.


Posted by: unimportant antics | Link to this comment | 11-30-08 7:35 PM
horizontal rule
139

141: Go jump off a cliff, antics.


Posted by: paranoid android | Link to this comment | 11-30-08 7:42 PM
horizontal rule
140

In case I was unclear, all I was wanting to do at 59 was to get on the table a certain conventional fallibilist picture of the meaning of "know". I wasn't intending to assume the cogency of global scepticism. Even if global scepticism is incoherent, it's hard to deny that there are some situations where most of us justifiably believe something, but where it's barely possible that we're wrong - and that's all that's needed for the conventional fallibilist picture to have a point. If you don't think it's at all possible that the election was a hoax engineered by The Onion, it's easy to pick something else that seems more imaginable, e.g. that the moon-landings were faked, or that George W. Bush isn't actually George H.W. Bush's son, but Barbara's love-child with Roy Rogers.

This is all very boring, of course, but we Nietzschean types exist beyond boredom and entertainment.


Posted by: Amit | Link to this comment | 11-30-08 9:18 PM
horizontal rule
141

142: I think you're confusing us with the hijinks.


Posted by: unimportant antics | Link to this comment | 11-30-08 9:19 PM
horizontal rule
142

Barbara Bush was Aleister Crowley's love-child. Fact.

The internet doesn't lie.


Posted by: John Emerson | Link to this comment | 11-30-08 9:24 PM
horizontal rule
143

145: Surely that would have been mentioned in this incredibly dumb avant-garde movie, if true.


Posted by: Cryptec Nid | Link to this comment | 11-30-08 9:26 PM
horizontal rule
144

Beyond boredom and entertainment is more boredom. Fact.


Posted by: Walt Someguy | Link to this comment | 11-30-08 9:30 PM
horizontal rule
145

I am saddened that my only actual philosophical contribution to the blog in all these years, 34, was ignored.


Posted by: Crypetc Nid | Link to this comment | 11-30-08 9:32 PM
horizontal rule
146

Even if global scepticism is incoherent

The burden is to show this. I'm not sure I recommend it in blog comments.


Posted by: parsimon | Link to this comment | 11-30-08 9:36 PM
horizontal rule
147

Returning to pdf's statement: "We don't have unmediated access to the external world, so we can never judge a belief to be true, except probabilistically."

It's very hard for me to imagine a coherent viewpoint in which we (a) can never judge a belief to be true but (b) can judge a belief to have a certain probability of being true.

I mean, dude, what's the probability that the sun will rise tomorrow?


Posted by: essear | Link to this comment | 11-30-08 9:42 PM
horizontal rule
148

what's the probability that the sun will rise tomorrow?

Quite high, happily.


Posted by: Sifu Tweety | Link to this comment | 11-30-08 9:59 PM
horizontal rule
149

Re 147:

That's why my comment was very boring, even though I exist beyond boredom and entertainment.


Posted by: Amit | Link to this comment | 11-30-08 10:05 PM
horizontal rule
150

150: What's the probability that the universe is being simulated in a computer in some larger universe and the simulation will be shut off tonight? -150 dB? -200 dB (.00000000001)? Let's call it -400 dB if you're extremely skeptical of the simulation argument. In fact, the probability that the sun will go prematurely supernova tonight might be higher. But I don't know physics to well, maybe it's lower. (In which case wouldn't it take a few days to engulf the Earth? But you see my point.) The probability that the sun will rise tomorrow is 1 minus all the atronomically unlikely things that could prevent it, which have astronomically tiny (though conveniently measurable on a log scale) but non-zero probabilities.

Just because it's 1 for all intents and purposes doesn't mean that in a philosophical sense it's 1. I mean, probability 1 on a log scale is positive infinity. Why bring infinity into the picture when you don't need to? It's so messy.


Posted by: paranoid android | Link to this comment | 11-30-08 10:13 PM
horizontal rule
151

143: I hope you don't think I'm advocating global skepticism. Extreme subjectivism is not the same as skepticism, you platonists.


Posted by: paranoid android | Link to this comment | 11-30-08 10:16 PM
horizontal rule
152

Why bring infinity into the picture when you don't need to? It's so messy.

Not to mention the cost of finding a nice frame in that size.


Posted by: Sifu Tweety | Link to this comment | 11-30-08 10:16 PM
horizontal rule
153

Applications: submitted! Now, pie.


Posted by: teofilo | Link to this comment | 11-30-08 10:16 PM
horizontal rule
154

155: I hear they sell ordinal frames down at the Cantor Barn. For Picasso, aleph null is fine, but for Dali they recommend at least an aleph-2.


Posted by: paranoid android | Link to this comment | 11-30-08 10:19 PM
horizontal rule
155

153: Why do you think "the probability that the sun will rise tomorrow" or "the probability that the universe is being simulated in a computer" are meaningful quantities? Or rather: given that you seem to think they are meaningful quantities, what do you think they mean? Why do you think it is possible to know them, when you don't think anything can be known with certainty?


Posted by: essear | Link to this comment | 11-30-08 10:20 PM
horizontal rule
156

Also, the last time I checked log(1) = 0.


Posted by: essear | Link to this comment | 11-30-08 10:22 PM
horizontal rule
157

158: The probability is determined by the knowledge of the knower. Modulo inferential bias, the meaning of "Q is true with probability P" is that in P of all possible worlds in which an agent exists with your knowledge and experience, proposition Q will be true. That's if you don't make any mistakes estimating P based on your knowledge.


Posted by: paranoid android | Link to this comment | 11-30-08 10:24 PM
horizontal rule
158

159: I think I meant to say log-odds/logit scale. log(p/(1-p)).


Posted by: paranoid android | Link to this comment | 11-30-08 10:25 PM
horizontal rule
159

Modulo inferential bias, the meaning of "Q is true with probability P" is that in P of all possible worlds in which an agent exists with your knowledge and experience, proposition Q will be true.

So your position is that we can't know whether a belief is true, but we can know in what fraction of all possible worlds consistent with our experience it is true? Surely not.


Posted by: essear | Link to this comment | 11-30-08 10:27 PM
horizontal rule
160

162: If we were perfectly rational agents, we could know in what fraction of all possible worlds consistent with our experience a belief is true, in other words, how likely it is to be true. And since we approximate rationality, basically, yes.


Posted by: paranoid android | Link to this comment | 11-30-08 10:29 PM
horizontal rule
161

Mmmm, delicious pie.


Posted by: teofilo | Link to this comment | 11-30-08 10:29 PM
horizontal rule
162

I enjoy pie.


Posted by: Sifu Tweety | Link to this comment | 11-30-08 10:30 PM
horizontal rule
163

And since we approximate rationality, basically, yes.

So, so not true. I heart you, crazy android!


Posted by: Sifu Tweety | Link to this comment | 11-30-08 10:31 PM
horizontal rule
164

Well, for a certain loose sense of "approximate".


Posted by: paranoid android | Link to this comment | 11-30-08 10:31 PM
horizontal rule
165

That's if you don't make any mistakes estimating P based on your knowledge.

That's one hell of a big "if."

And since we approximate rationality, basically, yes.

How closely do you think we approximate rationality?


Posted by: teofilo | Link to this comment | 11-30-08 10:32 PM
horizontal rule
166

Actually, I rather dislike pie.


Posted by: Stanley | Link to this comment | 11-30-08 10:32 PM
horizontal rule
167

You're remarkably consistent in your wacky beliefs.


Posted by: essear | Link to this comment | 11-30-08 10:32 PM
horizontal rule
168

Tell me, android, what does the space of all possible worlds look like? Knowing that would help a lot in my day job.


Posted by: essear | Link to this comment | 11-30-08 10:33 PM
horizontal rule
169

Sifu-pwned (approximately). At least we'll always have pie.


Posted by: teofilo | Link to this comment | 11-30-08 10:34 PM
horizontal rule
170

148 -- Sorry, but I suppose 34 was turd in the punch bowl at the reality denying party.

143 -- You could fill vast libraries on every continent with all the things I don't know -- and I'd be surprised if it turned out that all the things I believe are actually true. Nonetheless, there are plenty of things I do know: a philosophical sense that hesitates to put the probablility that our universe is a simulation in a computer in some bigger universe at zero is a nonsense.

153 Why bring infinity into the picture when you don't need to? -- You're the one wanting to use that measuring stick. Don't complain if it doesn't work.


Posted by: CharleyCarp | Link to this comment | 11-30-08 10:36 PM
horizontal rule
171

168: Right, humans make mistakes and are wrong all the time. But they can hold well calibrated beliefs, i.e. beliefs with the appropriate strength based on the evidence. If people's beliefs weren't well-calibrated to some degree, they wouldn't be able to consistently feed themselves. The more directly related to fulfilling one of our animal drives a belief is, the more likely it's well-calibrated. The more likely a belief will be seriously tested often, the more likely it's accurate.


Posted by: paranoid android | Link to this comment | 11-30-08 10:36 PM
horizontal rule
172

I have no idea what 173.3 is saying.


Posted by: paranoid android | Link to this comment | 11-30-08 10:39 PM
horizontal rule
173

See, 174 I kind of agree with, but it's not like there's an explicit, conscious, rational process of evaluating relative probabilities. It's all internal to how the brain works, and to evert things such that rationality is described as maximized probabilistic inference.... well, actually, I kind of like it. It's crazy, but I like it!


Posted by: Sifu Tweety | Link to this comment | 11-30-08 10:40 PM
horizontal rule
174

i.e. beliefs with the appropriate strength based on the evidence

This is very much weaker than saying we can assign meaningful numerical probabilities to whether a belief is true.


Posted by: essear | Link to this comment | 11-30-08 10:41 PM
horizontal rule
175

Sifu pwns again.


Posted by: essear | Link to this comment | 11-30-08 10:42 PM
horizontal rule
176

Further: in my field we often say things like "this scenario is excluded at 95% confidence level by these experimental results." What we mean by that is a socially agreed-upon numerical measure of confidence; it's well-defined, but it is a huge mistake to think it's a probability, and it certainly has nothing to do with counting possible worlds.


Posted by: essear | Link to this comment | 11-30-08 10:45 PM
horizontal rule
177

Well, I would of course agree that we don't have (much) conscious access to the numbers that would accurately describe the strength of one of our beliefs. I'd be surprised if we could get within 20 dB most of the time. Especially the extreme probabilities we have trouble gauging. But the sort of idealized process I'm describing is, I think, an accurate description of idealized rationality, and one that our brains try to approximate to some degree. I don't think our beliefs have little XML tags in our brains representing their strength as an IEEE floating point number.

I wish I knew why people got that impression in the first place.


Posted by: paranoid android | Link to this comment | 11-30-08 10:51 PM
horizontal rule
178

I fully admit up front that much of this conversation is over my head, but, PA, since you keep using dB, I was wondering: what's the reference level here, knowledge-wise? Decibels are a measurement based on a reference measurement, right?

I grant you permission to laugh if that's a dumb question, but I'm genuinely curious.


Posted by: Stanley | Link to this comment | 11-30-08 10:55 PM
horizontal rule
179

But I think you're conflating reinforcement learning with rationality. Reinforcement learning is an excellent way to maximize expectations based on past circumstance, but it's not a very good way to estimate a true state, or to rationally estimate the probability of highly unlikely events. Which is not to mention that even those aspects of cognition that are plausibly (in theory) explained by reinforcement learning (as far as we know, which is not very far) aren't like, plugged neatly in to creating accurate, stable models of the world (or universe!).


Posted by: Sifu Tweety | Link to this comment | 11-30-08 10:58 PM
horizontal rule
180

Well, I'm not sure if there's really a reference level here. Decibels need a reference measurement when they're limited to positive or negative numbers, but in this case they're not. 0 dB = 50/50 chance, 20 dB = .9 chance, 40 = .99 chance, -20 = .1, -40 = .01, etc.


Posted by: paranoid android | Link to this comment | 11-30-08 10:58 PM
horizontal rule
181

181: I was wonering about the dB thing myself. I get that it's a logarithmic scale, but past that I'm puzzled.


Posted by: Sifu Tweety | Link to this comment | 11-30-08 10:59 PM
horizontal rule
182

180: I don't care about the accuracy of the numbers. What I would like to understand is why you think our confidence in some piece of knowledge should be related to how often it happens in "all possible worlds", and what that means. It seems to me that if our beliefs are calibrated based on anything, it's repeated similar occurrences in the one particular world we happen to live in.


Posted by: essear | Link to this comment | 11-30-08 11:00 PM
horizontal rule
183

If Heidegger smoked weed, was it hydropontic? And can we ever really know for sure?


Posted by: Stanley | Link to this comment | 11-30-08 11:07 PM
horizontal rule
184

185: In essence is similar to the repeated experiment. Is your friend home right now? You estimate it's more likely than not. What does that mean? It means in a large number of instances where you have exactly the same knowledge and expectations you do now (the same brain state, more or less) but the rest of the world could be in other states, then your friend will be home at ratio P. Now, often situations are very similar to others, so you could do repeated experiments, but sometimes they're not, and many-worlds is necessary to be able to define it coherently all the time. Can you give an answer without using many-worlds? I can't think of one.


Posted by: paranoid android | Link to this comment | 11-30-08 11:08 PM
horizontal rule
185

I thought I had explained the dB thing earlier. Sorry.


Posted by: paranoid android | Link to this comment | 11-30-08 11:09 PM
horizontal rule
186

188: You do play piano, right? That's in my Unfogged commenter memory bank, so I was thinking about the dB thing in a musical sense.


Posted by: Stanley | Link to this comment | 11-30-08 11:11 PM
horizontal rule
187

in a large number of instances where you have exactly the same knowledge and expectations you do now (the same brain state, more or less) but the rest of the world could be in other states, then your friend will be home at ratio P.

A large number of instances of what? Is the set of possibilities discrete, or continuous? Probably in either case it's infinite? What does "P" mean, then? You're being cavalier about assuming probabilities are meaningful, and I don't see that they are.

The alternative is that by estimating "it's more likely than not", I'm giving some unquantifiable description of my confidence. If it must be quantified, ask how much money I'll bet on it. I'm not convinced there's a non-subjective meaning here.


Posted by: essear | Link to this comment | 11-30-08 11:13 PM
horizontal rule
188

I do. I'm quite familiar with musical dB. They have the same property, that +20 dB is 10 times the amplitude. (Subjectively, about 4 times as loud.)


Posted by: paranoid android | Link to this comment | 11-30-08 11:14 PM
horizontal rule
189

OK, a possible world is one in which there is a brain with atoms configured identical to your own in this present world at a given time. (Same physical laws.) No other constraints.


Posted by: paranoid android | Link to this comment | 11-30-08 11:16 PM
horizontal rule
190

And when estimating if your friend is likely to be home right now, you do a quick mental survey of all of these worlds based on the configuration of atoms in your brain at the given time, count the number with "friend at home" versus "friend not at home", and take a ratio? What if, for reasons yet unknown to science, we live in the only possible world?

Isn't it more likely you form a hunch based on previous experience?


Posted by: essear | Link to this comment | 11-30-08 11:18 PM
horizontal rule
191

Sorry, I'm bored with this. For further amusement, google "the measure problem in eternal inflation", which should turn up people trying to formulate a precise definition of how to count possible worlds. As far as I can tell it's a literature full of highly intelligent people who really should be doing something more useful with their time.


Posted by: essear | Link to this comment | 11-30-08 11:21 PM
horizontal rule
192

Also, if you're talking about the brain, as far as I understand the evidence it's a lot more likely that your brain takes your past experience of the friend being at home or not being at home, judges which is most likely, and then decides that's the fact (for definitions of "friend at home" which vary massively and don't necessarily have to do with friends, homes or physical locations). You don't hold a continuous distribution in your head, you jump to the current expectation and treat it as known.


Posted by: Sifu Tweety | Link to this comment | 11-30-08 11:23 PM
horizontal rule
193

194: Ha! I wasn't going to respond to your last thing anyway, because I find possible worlds boring. But despite their problems, I think they're probably the right approach. And really, in this case, it's about how much your knowledge determines the actual state of the world, so perhaps there's a rigorous way to define this stuff that doesn't need possible worlds.


Posted by: paranoid android | Link to this comment | 11-30-08 11:24 PM
horizontal rule
194

194: Your presence in this comment thread precludes you from drawing the conclusion in your last sentence.


Posted by: Walt Someguy | Link to this comment | 11-30-08 11:30 PM
horizontal rule
195

197: On the contrary, Unfogged has only made me more adept at recognizing intelligent people who are wasting their livestime.


Posted by: essear | Link to this comment | 11-30-08 11:33 PM
horizontal rule
196

Re 63:

Literally FTR, I'm actually not unsympathetic to something like what you say.


Posted by: Amit | Link to this comment | 12- 3-08 3:03 PM
horizontal rule