1131: "Math"

This forum is for the individual discussion thread that goes with each new comic.

Moderators: Moderators General, Prelates, Magistrates

User avatar
dudiobugtron
Posts: 1098
Joined: Mon Jul 30, 2012 9:14 am UTC
Location: The Outlier

Re: 1131: "Math"

Postby dudiobugtron » Tue Nov 13, 2012 7:12 am UTC

In order to prove 'white wins', then, you need to show that white starts with the 'initiative'*, and also that the initiative exists (ie: that black can't force a stalemate).

*(which seems intuitively correct - although it's possible that every opening move puts white in a worse position, it doesn't seem likely.)
Image

User avatar
Velexia
Posts: 196
Joined: Thu Nov 01, 2012 1:10 pm UTC

Re: 1131: "Math"

Postby Velexia » Tue Nov 13, 2012 7:43 am UTC

A quick google search.
Hail Eris!

rmsgrey
Posts: 3634
Joined: Wed Nov 16, 2011 6:35 pm UTC

Re: 1131: "Math"

Postby rmsgrey » Tue Nov 13, 2012 1:16 pm UTC

The Wikipedia page on whether White should win at Chess starts off confidently, portraying it as a question of whether White can force a win or Black can force a draw, but goes on to note that there are "expert" opinions that Black, by reacting to White exposing White's plans, should be able to counter them and force a win.

It's intuitively appealing that White's initiative is an advantage, but it's not formally proven.

SerialTroll
Posts: 82
Joined: Mon Dec 20, 2010 5:28 am UTC

Re: 1131: "Math"

Postby SerialTroll » Tue Nov 13, 2012 2:10 pm UTC

rmsgrey wrote:The Wikipedia page on whether White should win at Chess starts off confidently, portraying it as a question of whether White can force a win or Black can force a draw, but goes on to note that there are "expert" opinions that Black, by reacting to White exposing White's plans, should be able to counter them and force a win.

It's intuitively appealing that White's initiative is an advantage, but it's not formally proven.


Kinda interesting that this conversation veered into chess after the 'think logically' thread. Also interesting that Velexia has adopted some of my points after throwing a hissy and ignoring me.

Anyhow, I'll give some of my background. I've played chess for 30 years in national tournaments. I've had FIDE titled instructors and have used computer chess programs for over 20, have a background in computer programming and worked as a hobby on chess programs. My children are involved in chess, and my 7 year old daughter recently won a USCF rated tournament with a perfect score in her age group (girl power).

There is absolutely no advantage in playing black. Chess has really been formally studied for approximately 400 years. you could argue for a couple hundred years longer than that, but chess rules were not even 100% consistent across the world, so I take the 1600s as the point where you see score sheets being kept, opening lines studied and other items that we would recognize as being modern to the game.

For a long time, before computers, three moves were considered to be worth a pawn in the opening. This dynamic advantage of three moves was considered to be the same as the static advantage of one pawn. White, therefore, has an advantage of 1/3 of a pawn. Computers come up with a similar assessment. Turn off the opening book of a commercially available chess program, let it run for a couple days and it will come up roughly a 0.33 pawn advantage. The number will of course vary based on the algorithm the computer uses, but regardless the computer will always come up with an advantage for White.

Modern Chess Openings is considered the Bible of the chess openings world. It is a comprehensive overview of chess openings as you might surmise from the title. I have an old edition that I inherited from my grandfather and a 13th edition (15th is now out). It is based on theory from tournament play and I believe computer analysis now. At the end of each line, the book will give its assessment of whether an opening is better for black, white or equal. It also gives degrees of "better". You will not find a single forcing line, one where white must play into it, where black is better. In fact, you'll find that white can always force an opening where he at least has a marginal advantage.

If you want further evidence, you only have to look at tournament play. Regardless of the skill level of the tournament, white has a large advantage over white. Doesn't matter if it is correspondence chess, traditional tournament time controls, blitz or bullet chess. If two equally skilled players go at it, white wins more often than black. As you go up the skill tree, draws become more prevalent, but White retains the advantage.

Black with best play is looking for equality. He simply cannot force an advantage. I cannot imagine what the wikipedia article is referring to with the "expert" opinions. Can you point to the page that refers to this? I looked through "Chess", but I see nothing. I assume there is another wikipedia page, that the "expert" opinion was put in without reference and might refer to some guy's chess buddy rated 1200 USCF.

So, in one sense you are correct. Chess hasn't been solved. And no GrandMaster would argue white or black can force a win in chess (unless he is trolling). No other titled player would either. Those involved in computer chess would also agree. Yet, there is undoubtedly an advantage for White because he starts with the initiative -- born out by tournament play, computer play and opening study.

if you could send me the link to the page you found, I would appreciate it. I do edit wikipedia and love chess, and would like to review and correct if needed.

User avatar
gmalivuk
GNU Terry Pratchett
Posts: 26767
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

Re: 1131: "Math"

Postby gmalivuk » Tue Nov 13, 2012 2:45 pm UTC

I'm curious how it drifted to the topic of chess in the first place. It's almost as if several people aren't paying the least bit of attention to which thread they're posting in, preferring instead to just drag their same arguments into whatever thread suits them at the moment.
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

rmsgrey
Posts: 3634
Joined: Wed Nov 16, 2011 6:35 pm UTC

Re: 1131: "Math"

Postby rmsgrey » Tue Nov 13, 2012 3:20 pm UTC

SerialTroll wrote:if you could send me the link to the page you found, I would appreciate it. I do edit wikipedia and love chess, and would like to review and correct if needed.

http://en.wikipedia.org/wiki/First-move_advantage_in_chess

I am suspicious that much of the analysis of Chess builds on earlier work, so the fact that later conclusions and later statistical evidence agrees with the earlier results should be discounted to some extent. For example, if it's accepted that black should play for a draw while white plays for a win, then players who do well at achieving their goals in the game will produce a record where white wins more often than black - even if there were no truth in the idea that white has the advantage, once it became established, it would naturally produce the results that then get used as confirmatory evidence.

What the evidence supports is that, within the limits of existing analysis, white has the advantage and should either win or draw with best-known play. Incidentally, this is not the same as saying that black should play for equality - there's a difference between a draw being the best possible outcome and aiming for a draw being the best possible approach. The Wikipedia page includes reported opinions by Fischer and Kasparov that the secret to playing black well is to play to win, not to draw.

It remains possible (but unlikely) that black actually has the advantage once the game is solved, but that it's harder to play black "well", and will do so until someone manages to prove otherwise.

SerialTroll
Posts: 82
Joined: Mon Dec 20, 2010 5:28 am UTC

Re: 1131: "Math"

Postby SerialTroll » Tue Nov 13, 2012 3:36 pm UTC

rmsgrey wrote:
SerialTroll wrote:if you could send me the link to the page you found, I would appreciate it. I do edit wikipedia and love chess, and would like to review and correct if needed.

http://en.wikipedia.org/wiki/First-move_advantage_in_chess

I am suspicious that much of the analysis of Chess builds on earlier work, so the fact that later conclusions and later statistical evidence agrees with the earlier results should be discounted to some extent. For example, if it's accepted that black should play for a draw while white plays for a win, then players who do well at achieving their goals in the game will produce a record where white wins more often than black - even if there were no truth in the idea that white has the advantage, once it became established, it would naturally produce the results that then get used as confirmatory evidence.

It remains possible (but unlikely) that black actually has the advantage once the game is solved, but that it's harder to play black "well", and will do so until someone manages to prove otherwise.


1) Thanks for the link. I'll read this more closely
2) While it is possible that somehow humans have retained some sort of corporate bad habit for black, it is unlikely. There are many lines where black plays for an advantage and gets it if white plays improperly (think center counter gambit), black gets a larger disadvantage if white plays properly.
3) While it is possible that somehow humans have retained some sort of corporate bad habit for black, computers brute forcing openings would not. You are free to conduct this experiment for yourself. There are inexpensive chess programs where you can try it for yourself and I believe a few free ones. I use two chess engines, Fritz (technically contains multiple engines) and Chessmaster. Older versions of both can practically be purchased for free.
4) While you are correct that there could technically be some type of advantage for black, it is not born out by any evidence. If I am wrong, present the evidence.
5) At this point, the argument (read "discussion of points", not "heated") boils down to it is theoretically possible that all of the chess minds of the past 120 years or so are all wrong, that computer capable of trillions of chess positions per second of analysis and beating grandmaters, and practical play are all wrong. Yes, it is theoretically possible. But would you bet $100 on it even with 1,000 to 1 odds? If so, I'll gladly take it.

Now, I'm off to read the wikipedia article.
Last edited by SerialTroll on Tue Nov 13, 2012 3:42 pm UTC, edited 1 time in total.

User avatar
Velexia
Posts: 196
Joined: Thu Nov 01, 2012 1:10 pm UTC

Re: 1131: "Math"

Postby Velexia » Tue Nov 13, 2012 3:37 pm UTC

rmsgrey wrote:The Wikipedia page on whether White should win at Chess starts off confidently, portraying it as a question of whether White can force a win or Black can force a draw, but goes on to note that there are "expert" opinions that Black, by reacting to White exposing White's plans, should be able to counter them and force a win.

It's intuitively appealing that White's initiative is an advantage, but it's not formally proven.


Indeed, I also read that particular article and it was very interesting. I noted that most everyone except perhaps the last person to suggest that white can force a win thanks to initiative has been heavily ridiculed by other experts.

I think it is highly possible that this unpopular idea is true though... It'll be a while before it is proven, but I have used the "initiative" in chess to great effect, to win many games. I always keep in mind who has initiative, and what move I can make to regain it, or keep it.
Hail Eris!

User avatar
gmalivuk
GNU Terry Pratchett
Posts: 26767
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

Re: 1131: "Math"

Postby gmalivuk » Tue Nov 13, 2012 3:40 pm UTC

Seriously, folks, are you sure you know which thread you're discussing this in?
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

SerialTroll
Posts: 82
Joined: Mon Dec 20, 2010 5:28 am UTC

Re: 1131: "Math"

Postby SerialTroll » Tue Nov 13, 2012 3:48 pm UTC

Velexia wrote:
I think it is highly possible that this unpopular idea is true though... It'll be a while before it is proven, but I have used the "initiative" in chess to great effect, to win many games. I always keep in mind who has initiative, and what move I can make to regain it, or keep it.


Yes, Velexia, you are clearly the chess genius, and others like Steinitz, Fischer and Kasparov are the fools. What does a GM title or World Champion title mean anyhow when we have a chess genius of your caliber here? A week ago you were arguing that chess is "solved" and a proven win for white. This shows how little you really know. Do you really think these players don't look for the initiative?

What is most sad about you isn't your ignorance, but the fact that you don't recognize your ignorance. There are many subjects where I am not an expert. The difference is that i don't pretend to be an expert in those subjects and am willing to learn from those who are. You fail on this account.
Last edited by SerialTroll on Tue Nov 13, 2012 3:52 pm UTC, edited 1 time in total.

SerialTroll
Posts: 82
Joined: Mon Dec 20, 2010 5:28 am UTC

Re: 1131: "Math"

Postby SerialTroll » Tue Nov 13, 2012 3:49 pm UTC

gmalivuk wrote:Seriously, folks, are you sure you know which thread you're discussing this in?


Sorry. the subject went to chess here and it is a love of mine. Too many incorrect statements for me to ignore.

User avatar
Velexia
Posts: 196
Joined: Thu Nov 01, 2012 1:10 pm UTC

Re: 1131: "Math"

Postby Velexia » Tue Nov 13, 2012 3:52 pm UTC

At Page 3, 26 posts down, I think a majority of people posting got distracted and forgot which thread this was. It would certainly not be in error to move it to the proper thread (Chess! or some thread to that effect...) (although it might take some work).

J Thomas wrote:Back on the original topic, it doesn't work to make a proof that there's a tree of moves for white that is guaranteed to win, without looking at all the possible responses.
Hail Eris!

rmsgrey
Posts: 3634
Joined: Wed Nov 16, 2011 6:35 pm UTC

Re: 1131: "Math"

Postby rmsgrey » Tue Nov 13, 2012 4:06 pm UTC

SerialTroll wrote:3) While it is possible that somehow humans have retained some sort of corporate bad habit for black, computers brute forcing openings would not. You are free to conduct this experiment for yourself. There are inexpensive chess programs where you can try it for yourself and I believe a few free ones. I use two chess engines, Fritz (technically contains multiple engines) and Chessmaster. Older versions of both can practically be purchased for free.


The chess programs use some sort of heuristic to judge the value of the end-states of their brute-force openings. While I wouldn't want to bet that black does have the forced win in Chess, I would be happy to bet that the heuristics used to evaluate the brute-force openings in deciding whether they generate an advantage for either side are based on legacy analysis of the game rather than being independently derived from first principles.

The comic works pretty well for any major chess tournament (aside from using blue and red rather than white and black) - representing the percentage of checkmates for each colour - though the error-margin looks a little wide.

SerialTroll
Posts: 82
Joined: Mon Dec 20, 2010 5:28 am UTC

Re: 1131: "Math"

Postby SerialTroll » Tue Nov 13, 2012 4:56 pm UTC

rmsgrey wrote:The chess programs use some sort of heuristic to judge the value of the end-states of their brute-force openings. While I wouldn't want to bet that black does have the forced win in Chess, I would be happy to bet that the heuristics used to evaluate the brute-force openings in deciding whether they generate an advantage for either side are based on legacy analysis of the game rather than being independently derived from first principles.


To a certain point you are correct that the algorithms are based on what a human think. But then I would suggest that it would be up to you or others to suggest algorithms that are better suited to black play. It would be a relatively trivial exercise and would absolutely revolutionize the game. Your name would go down in the history of chess. Programs with different algorithms and different weighting of inputs are frequently pitted against each other and this is how different engines get different ratings. The algorithms themselves are not terribly complex, the engines themselves are responsible for how to prune search trees and identifying transpositions of positions. if you can put together such an algorithm and have it beat the traditional algorithms, you will literally rewrite centuries of chess theory.

The argument I am hearing is there is still a theoretical possibility, then my response is "maybe, but I am 99.9999999999% certain white has an advantage". Add a couple more 9's, and you will have my confidence level for 2 + 2 = 4.

rmsgrey
Posts: 3634
Joined: Wed Nov 16, 2011 6:35 pm UTC

Re: 1131: "Math"

Postby rmsgrey » Tue Nov 13, 2012 5:29 pm UTC

SerialTroll wrote:I would suggest that it would be up to you or others to suggest algorithms that are better suited to black play. It would be a relatively trivial exercise


Relative to what???

SerialTroll wrote:The argument I am hearing is there is still a theoretical possibility, then my response is "maybe, but I am 99.9999999999% certain white has an advantage". Add a couple more 9's, and you will have my confidence level for 2 + 2 = 4.


Being certain at the one-in-a-trillion level strikes me as probably going beyond the data - with a branching number of ~30, it takes about 8 moves to get from an arbitrary initial position to a trillion possible descendants - if one line in a trillion is winning for black, then picking it out is loosely parallel to looking ahead for 8 moves by each side.

For that matter, while I can think of reasons why someone might put their confidence in 2+2=4 at less than 100%, I can't think of any that leave it higher than about six 9's (one in a million).

Of course, I don't have a lot of hard data to base my figures on, so if you have a derivation, I'd be interested in reading it.

SerialTroll
Posts: 82
Joined: Mon Dec 20, 2010 5:28 am UTC

Re: 1131: "Math"

Postby SerialTroll » Tue Nov 13, 2012 5:40 pm UTC

rmsgrey wrote:
SerialTroll wrote:I would suggest that it would be up to you or others to suggest algorithms that are better suited to black play. It would be a relatively trivial exercise


Relative to what???



My point is simply that when one is arguing that established truth "x", and the established truth is backed by an enormous amount of evidence that it is up to the arguer to demonstrate the argument.

rmsgrey wrote:
SerialTroll wrote:I would suggest that it would be up to you or others to suggest algorithms that are better suited to black play. It would be a relatively trivial exercise


Relative to what???

rmsgrey wrote:Being certain at the one-in-a-trillion level strikes me as probably going beyond the data - with a branching number of ~30, it takes about 8 moves to get from an arbitrary initial position to a trillion possible descendants - if one line in a trillion is winning for black, then picking it out is loosely parallel to looking ahead for 8 moves by each side.

For that matter, while I can think of reasons why someone might put their confidence in 2+2=4 at less than 100%, I can't think of any that leave it higher than about six 9's (one in a million).

Of course, I don't have a lot of hard data to base my figures on, so if you have a derivation, I'd be interested in reading it.


I wasn't actually counting the 9's, simply expressing that while nothing is certain, these two facts are really close to it.

At this point, I'm not sure there is much more to really be said about this discussion. You are free to respond, but it doesn't sound like we have any real difference of opinion and that now we are simply talking around the issue. I'll wish you a good day and see you in other threads.

User avatar
CZeke
Posts: 52
Joined: Tue Jan 09, 2007 9:45 am UTC
Location: Waterloo, Canada
Contact:

Re: 1131: "Math"

Postby CZeke » Tue Nov 13, 2012 6:18 pm UTC

dudiobugtron wrote:
J Thomas wrote:
CZeke wrote:Ah, but perhaps you could show that some other kind of proof exists without actually exhibiting one...


I can't, and I'm not going to work on it. Maybe somebody else can. But -- what's the difference between showing that a proof exists that white wins, compared to proving that white wins?

Nothing. I think the difference was supposed to be between proving that white can force a win if they know how, and explicitly showing how a white player might go about it.


It was a joke, guys. I was being meta.
FiveMinute.net: Because stuff is long and life is short

J Thomas
Everyone's a jerk. You. Me. This Jerk.^
Posts: 1190
Joined: Fri Sep 23, 2011 3:18 pm UTC

Re: 1131: "Math"

Postby J Thomas » Tue Nov 13, 2012 6:57 pm UTC

SerialTroll wrote:
gmalivuk wrote:Seriously, folks, are you sure you know which thread you're discussing this in?


Sorry. the subject went to chess here and it is a love of mine. Too many incorrect statements for me to ignore.


That's worth careful attention. If you can't ignore incorrect statements on the Internet, then it would be a good idea to ignore the Internet until you achieve that discipline.

If someone can troll you just by saying something wrong and sticking to it no matter how well you prove they're wrong, you can get stuck in an argument that's no fun. (Unless it does continue to be fun for you, in which case never mind.)
The Law of Fives is true. I see it everywhere I look for it.

User avatar
gmalivuk
GNU Terry Pratchett
Posts: 26767
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

Re: 1131: "Math"

Postby gmalivuk » Tue Nov 13, 2012 7:34 pm UTC

J Thomas wrote:That's worth careful attention. If you can't ignore incorrect statements on the Internet, then it would be a good idea to ignore the Internet until you achieve that discipline.
You obviously haven't spent much time on this forum, or read the comic it's attached to, if you really believe that.

Saying something wrong is often as effective a nerd-snipe as asking an interesting and difficult question, and this forum is full of nerds.
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

SerialTroll
Posts: 82
Joined: Mon Dec 20, 2010 5:28 am UTC

Re: 1131: "Math"

Postby SerialTroll » Tue Nov 13, 2012 7:49 pm UTC

J Thomas wrote:
SerialTroll wrote:
gmalivuk wrote:Seriously, folks, are you sure you know which thread you're discussing this in?


Sorry. the subject went to chess here and it is a love of mine. Too many incorrect statements for me to ignore.


That's worth careful attention. If you can't ignore incorrect statements on the Internet, then it would be a good idea to ignore the Internet until you achieve that discipline.

If someone can troll you just by saying something wrong and sticking to it no matter how well you prove they're wrong, you can get stuck in an argument that's no fun. (Unless it does continue to be fun for you, in which case never mind.)


Please don't act superior. It was you that started the chess discussion on an unrelated thread.

J Thomas
Everyone's a jerk. You. Me. This Jerk.^
Posts: 1190
Joined: Fri Sep 23, 2011 3:18 pm UTC

Re: 1131: "Math"

Postby J Thomas » Tue Nov 13, 2012 8:41 pm UTC

SerialTroll wrote:
J Thomas wrote:
SerialTroll wrote:
gmalivuk wrote:Seriously, folks, are you sure you know which thread you're discussing this in?


Sorry. the subject went to chess here and it is a love of mine. Too many incorrect statements for me to ignore.


That's worth careful attention. If you can't ignore incorrect statements on the Internet, then it would be a good idea to ignore the Internet until you achieve that discipline.

If someone can troll you just by saying something wrong and sticking to it no matter how well you prove they're wrong, you can get stuck in an argument that's no fun. (Unless it does continue to be fun for you, in which case never mind.)


Please don't act superior. It was you that started the chess discussion on an unrelated thread.


Seriously, if you feel the need to respond to such things, to argue that you aren't inferior, does it get you what you want?

No problem if you enjoy it, but I've gotten some sense that you might be feeling irritated at the whole thing. If so, it's so much more pleasant when you can give yourself permission to just drop it.
The Law of Fives is true. I see it everywhere I look for it.

User avatar
gmalivuk
GNU Terry Pratchett
Posts: 26767
Joined: Wed Feb 28, 2007 6:02 pm UTC
Location: Here and There
Contact:

Re: 1131: "Math"

Postby gmalivuk » Wed Nov 14, 2012 12:07 am UTC

Are you once again falling into that same "must respond again!" mindset that you advised Velexia to avoid, way back in the post where you managed to drag this thread off the rails into chess territory?
Unless stated otherwise, I do not care whether a statement, by itself, constitutes a persuasive political argument. I care whether it's true.
---
If this post has math that doesn't work for you, use TeX the World for Firefox or Chrome

(he/him/his)

User avatar
dudiobugtron
Posts: 1098
Joined: Mon Jul 30, 2012 9:14 am UTC
Location: The Outlier

Re: 1131: "Math"

Postby dudiobugtron » Wed Nov 14, 2012 12:31 am UTC

CZeke wrote:
dudiobugtron wrote:
J Thomas wrote:
CZeke wrote:Ah, but perhaps you could show that some other kind of proof exists without actually exhibiting one...


I can't, and I'm not going to work on it. Maybe somebody else can. But -- what's the difference between showing that a proof exists that white wins, compared to proving that white wins?

Nothing. I think the difference was supposed to be between proving that white can force a win if they know how, and explicitly showing how a white player might go about it.


It was a joke, guys. I was being meta.

I would respond to this, but it might also be a meta-joke in disguise! ;)
Image

J Thomas
Everyone's a jerk. You. Me. This Jerk.^
Posts: 1190
Joined: Fri Sep 23, 2011 3:18 pm UTC

Re: 1131: "Math"

Postby J Thomas » Wed Nov 14, 2012 12:38 am UTC

gmalivuk wrote:Are you once again falling into that same "must respond again!" mindset that you advised Velexia to avoid, way back in the post where you managed to drag this thread off the rails into chess territory?


;)
The Law of Fives is true. I see it everywhere I look for it.

wumpus
Posts: 546
Joined: Thu Feb 21, 2008 12:16 am UTC

Re: 1131: "Math"

Postby wumpus » Wed Nov 14, 2012 1:53 am UTC

But somebody on the internet is wrong!

User avatar
Velexia
Posts: 196
Joined: Thu Nov 01, 2012 1:10 pm UTC

Re: 1131: "Math"

Postby Velexia » Wed Nov 14, 2012 2:07 am UTC

Pretty sure that J Thomas made an honest mistake about which thread he was in after referencing that other thread... so let's not try to use it as ammo, that's silly. Also, I have really nothing further to add about the election... so I'm outta here =P
Hail Eris!

User avatar
addams
Posts: 10274
Joined: Sun Sep 12, 2010 4:44 am UTC
Location: Oregon Coast: 97444

Re: 1131: "Math"

Postby addams » Wed Nov 14, 2012 3:44 am UTC

Whew. Thank God.
Only Nov. 13 and people are done with the election.

Has anyone explained what is different; Yet?
Life is, just, an exchange of electrons; It is up to us to give it meaning.

We are all in The Gutter.
Some of us see The Gutter.
Some of us see The Stars.
by mr. Oscar Wilde.

Those that want to Know; Know.
Those that do not Know; Don't tell them.
They do terrible things to people that Tell Them.

wumpus
Posts: 546
Joined: Thu Feb 21, 2008 12:16 am UTC

Re: 1131: "Math"

Postby wumpus » Wed Nov 14, 2012 1:08 pm UTC

addams wrote:Whew. Thank God.
Only Nov. 13 and people are done with the election.

Has anyone explained what is different; Yet?


I think California (state government) had a result that may have a difference. Other notes:

House: Rs have the initiative on any tax bill
Senate: Ds can take initiative on any other bill. Rs can still block anything.
White House: Ds can take initiative on scotus replacements. Likely the most important (or at least long lasting) effect.
Essentially you need a 2/3 majority of the senate and either the presidency or 2/3 of the house to pass legislation over an active opposition (the 2000-2006 Democrats didn't count as any opposition). I can't see that happening any time soon.

In roughly a month both the latest extension of the "temporary" Bush tax cuts will expire, along with the "temporary" deficit funded spending boosts. If nothing changes here, then don't expect anything else to be different till at least 2014.

Oh, and we were done with the election. Chess, not so much :).

User avatar
CZeke
Posts: 52
Joined: Tue Jan 09, 2007 9:45 am UTC
Location: Waterloo, Canada
Contact:

Re: 1131: "Math"

Postby CZeke » Wed Nov 09, 2016 6:10 am UTC

So remember this strip and its smug certainty that Dewey always beats Truman, and those who disagree are saying MATH IS WRONG?

Image

I really hope we get a followup.
FiveMinute.net: Because stuff is long and life is short

User avatar
Copper Bezel
Posts: 2426
Joined: Wed Oct 12, 2011 6:35 am UTC
Location: Web exclusive!

Re: 1131: "Math"

Postby Copper Bezel » Wed Nov 09, 2016 6:45 am UTC

I don't think the comic claimed that. Subsequent discussion may well have done. Elections really are usually a game of invented conflict and hyperbolic commentary to make them interesting enough to maintain ratings. That doesn't mean unexpected upsets can't happen, just that the news media would generally like to give the impression that that's the only kind of election there is. Stuck clocks and all.
So much depends upon a red wheel barrow (>= XXII) but it is not going to be installed.

she / her / her

User avatar
cyanyoshi
Posts: 413
Joined: Thu Sep 23, 2010 3:30 am UTC

Re: 1131: "Math"

Postby cyanyoshi » Thu Nov 10, 2016 11:21 pm UTC

In fairness, math does say that 80% != 100%.

cupric
Posts: 30
Joined: Mon Nov 07, 2016 5:29 pm UTC

Re: 1131: "Math"

Postby cupric » Fri Nov 11, 2016 7:08 pm UTC

BREAKING: TO SURPRISE OF RANDALL MUNROE, NUMBERS CONTINUE TO BE AN INADEQUATE TOOL TO FULLY EXPLAIN AND PREDICT HUMAN BEHAVIOR

Glad to see I wasn't the only one who thought of this comic on Tuesday night.

Of course I'm (mostly) kidding. The election results were well within fivethirtyeight.com's margin of error. Nate Silver said multiple times during the last week before the election, in effect, "Don't get cocky, Democrats."

I too am really hoping for a follow-up.

User avatar
TvT Rivals
Posts: 41
Joined: Wed Oct 26, 2016 2:27 am UTC
Contact:

Re: 1131: "Math"

Postby TvT Rivals » Sat Nov 12, 2016 4:03 am UTC

cupric wrote:Nate Silver said multiple times during the last week before the election, in effect, "Don't get cocky, Democrats."

Did he? Where?

But besides of that, it's time to look where the error happened. Were the polls done in the correct way? Did the polled people answer honestly? Is Nate Silver's math correct? (Probably, he wouldn't make the same mistake every time.) Do we have to use a different kind of statistics?

User avatar
Soupspoon
You have done something you shouldn't. Or are about to.
Posts: 4060
Joined: Thu Jan 28, 2016 7:00 pm UTC
Location: 53-1

Re: 1131: "Math"

Postby Soupspoon » Sat Nov 12, 2016 5:24 am UTC

Nate's predictions were not "80% Clinton, 20% Trump, therefore definitely Clinton", or similar, but "80% Clinton, 20% Trump, so maybe Trump..." To be fair, so was practically every other predictor, or prediction aggregator, whether they knew it or not and whether or not the headlines (internal or external) went further than" It looks like Clinton will win...".

538, in particular, showed both histograms and probability curves (IIRC showing "80% confidence" zones) that indicated clear possibilities of both inverted and more extreme results.

If I'm rolling a D20 and a multiple of 4 is unlikely, I can't act surprised if it does land on such a multiple, however inconvenient. Not surprised, but dissapointed, and leaves me suspiciously looking at the dice if I'd previously been using them for things like Brexit and they'd rolled against my preferences, desires and expectations in Every. Single. Instance...

rmsgrey
Posts: 3634
Joined: Wed Nov 16, 2011 6:35 pm UTC

Re: 1131: "Math"

Postby rmsgrey » Sat Nov 12, 2016 12:23 pm UTC

Yeah, if the outcome of the election is "really close, but marginally Trump" and the polls were predicting "really close, but marginally Clinton", then that's less of an error than if the result had been "Clinton landslide" instead - particularly since the popular vote came down "really close, but marginally Clinton".

The problem is not so much in the maths of the pollsters, as in the reporting of their results, and is largely a result of trying convert a continuous variable into a binary result through an arbitrary cutoff. For most predicted values, their error-bars all fall on the same side of the line, so just which side is ahead is all you need to know; for values close to the line, the correct prediction looks more like "too close to call" than it does like "X wins" as the number of times the measured and actual values lie on opposite sides of the cutoff increases as you get closer.

Better understanding can reduce your error bars but you can't eliminate them entirely (though Asimov did write a short story about exactly that - a world where the polling was so accurate that, rather than bothering to hold the popular vote, elections were decided by a carefully designed poll of a single, carefully chosen citizen...

User avatar
Soupspoon
You have done something you shouldn't. Or are about to.
Posts: 4060
Joined: Thu Jan 28, 2016 7:00 pm UTC
Location: 53-1

Re: 1131: "Math"

Postby Soupspoon » Sat Nov 12, 2016 2:24 pm UTC

rmsgrey wrote:(though Asimov did write a short story about exactly that - a world where the polling was so accurate that, rather than bothering to hold the popular vote, elections were decided by a carefully designed poll of a single, carefully chosen citizen...
I did wonder if it would be too obscure to mention that story, having somehow forgotten what forum I was on. Also please let me close your parerentheses...:))

As for "too close to call", related screenshot here, if all goes well, one of a number I made during the turbulent night. Cropped to show Michigan (not yet sure) and Nevada (on a wild-ish ride, but now flipped), indicating the refined predictions (until 100%, where possible) as results modified expectations.

Photo - 02016-15-12-02-15-51.jpg
Nice surprise. Turns out the attached file isn't as cubic-scaled to unreadability as the image-cropper /itself/ suggested. Stupid tablet app interfaces...

cupric
Posts: 30
Joined: Mon Nov 07, 2016 5:29 pm UTC

Re: 1131: "Math"

Postby cupric » Sun Nov 13, 2016 2:37 am UTC

TvT Rivals wrote:
cupric wrote:Nate Silver said multiple times during the last week before the election, in effect, "Don't get cocky, Democrats."

Did he? Where?

But besides of that, it's time to look where the error happened. Were the polls done in the correct way? Did the polled people answer honestly? Is Nate Silver's math correct? (Probably, he wouldn't make the same mistake every time.) Do we have to use a different kind of statistics?


Here's a good example of him explaining how much we didn't know at the time: http://fivethirtyeight.com/features/election-update-dont-ignore-the-polls-clinton-leads-but-its-a-close-race/

Statistics as a scientific discipline is doing just fine. This was a case of garbage in garbage out. Silver is a good enough statistician that he could see how much uncertainty there was in the polling data, and hedged his predictions accordingly.


Return to “Individual XKCD Comic Threads”

Who is online

Users browsing this forum: DwayneSa, Google Feedfetcher and 25 guests