When unanimity signals bias
January 7, 2016 4:19 AM   Subscribe

 
That was really interesting, thanks. Examples 1 - 5 all looked good to me, but the sixth seemed a little strange: I think empirical models still rely on the unreasonable effectiveness of mathematics.
posted by Ned G at 4:40 AM on January 7, 2016 [2 favorites]


So too, if everyone here fails to vote #1 quidnunc kid, should he not be swiftly elected? I think we can all agree that the only way to root out and stomp on the dark vice of bias on MeFi is to follow this wise and just policy. In that context, please either vote #1 quidnunc kid, or don't - the result will be exactly the same. And, once he is elected to a position of unlimited power, he promises to utterly crush your biases, and also your face, in a gigantic dark vice.
posted by the quidnunc kid at 4:44 AM on January 7, 2016 [61 favorites]


I can imagine that I'll find myself thinking of this article a lot, now that I've read it. I do experimental optics research, and I've learned the hard way to be suspicious of my own data when it is "too good." Often turns out that my "beautiful signal" is a reflection hitting the detector funny, a malfunctioing instrument or problem with the set up, or an illusion caused by wishful thinking and pareidolia. The first thing you do with a pretty signal is mess it up to prove you can.

I had thought of it as a kind of scientific superstition that researchers develop, this distrust of the "too good" result. But this article gives me rational grounds for it, so thank you!
posted by OnceUponATime at 4:49 AM on January 7, 2016 [15 favorites]


I was kind of hoping the article would explain why so many people saw Muslims dancing and celebrating 9/11 in NJ.
posted by klarck at 4:57 AM on January 7, 2016


I just remembered what this reminded me of- this scene in World War Z, where they explain why only Jerusalem managed to anticipate the zombie threat. Wish I could add it to the OP now.
posted by leibniz at 4:58 AM on January 7, 2016 [3 favorites]


"...also, an entire panel of Jews with only one opinion? This is what you see? Clearly not all is well and we can try again later and get some proper disagreement. Jews agreeing! What a thing to have to see in this world."
posted by Tomorrowful at 5:03 AM on January 7, 2016 [31 favorites]


Opens up an interesting reverse defence strategy.
posted by Segundus at 5:09 AM on January 7, 2016 [5 favorites]


doesn't this encourage people to avoid the middle ground and instead opt for the most barbaric crime possible? if i do something so god-awfully terrible that everyone is convinced i must be guilty, then i get off free?

oh, on edit: what Segundus said.
posted by andrewcooke at 5:11 AM on January 7, 2016 [3 favorites]


Brings to mind police officer shootings. Everyone acquit these noble defenders of justice!
posted by benzenedream at 5:26 AM on January 7, 2016 [2 favorites]


I like this whole paper but.... does ancient jewish law really say that? That story feels too good to be true. Also: TOTALY NUTSO BONKERS
posted by rebent at 5:28 AM on January 7, 2016 [2 favorites]


A detailed paper on the Talmudic provision that exonerates the accused if there is a unanimous verdict

One "novel" interpretation of this law is that the translation of the original text is not that "he is exonerated" but that "he is sent away", meaning that he is immediately executed, there being no need for delay if the judges are unanimous.

But the prevailing interpretations hold that a unanimous verdict (of no less than 23 Sanhedrin judges) indicates judicial collusion. Another view (from Menachem Schneerson (the last Lubavitcher Rebbe) is that if the crime was so heinous that all the judges agree on execution, the defendant is "undeserving of punishment at the hands of mere mortals" and must by judged and punished by God himself. Yet another view, which places a lot of confidence in the inherent good of people, is that punishment is unnecessary in the case of a unanimous verdict because the defendant, hearing such a verdict, will surely regret his actions.
posted by beagle at 5:30 AM on January 7, 2016 [52 favorites]


I guess some crimes are just too big to fail!
posted by blue_beetle at 5:35 AM on January 7, 2016 [6 favorites]


Wow, thanks beagle
posted by rebent at 5:36 AM on January 7, 2016


If the judges know about this rule, and if they are colluding, wouldn't they just agree to have a few of them be token opposition?
posted by Pyry at 5:37 AM on January 7, 2016 [10 favorites]


9 out of 10 judges agree that there is no collusion on the panel of judges investigating collusion among judges.
posted by tobascodagama at 5:41 AM on January 7, 2016 [31 favorites]


If the judges know about this rule, and if they are colluding, wouldn't they just agree to have a few of them be token opposition?

Exactly. Dumbest collusion ever amirite?

The only alternative I can think of is if there was some overwhelming extrajudicial social pressure to not be the one particular judge that found known, obvious criminal not guilty to get him on the hook, as opposed to off via the 100% rule.

Bah my brain hurts after typing that. Too early for this shit.
posted by RolandOfEld at 5:44 AM on January 7, 2016 [2 favorites]


> The first thing you do with a pretty signal is mess it up to prove you can.

The same thing applies to debugging software. The first thing you do once you fix a bug is try to break it again. Sometimes you didn't fix the bug, you just stopped being able to reproduce it.
posted by I-Write-Essays at 5:46 AM on January 7, 2016 [8 favorites]


Oh and World War Z is far better in book form. Best in audiobook form even. Carl Reiner does the Jewish chapter/speaker and is outstanding, not to mention Alan Alda and other greats.
posted by RolandOfEld at 5:48 AM on January 7, 2016 [6 favorites]


Mark Hamill!

(Just ignore the unfortunate Chinese accents as best you can)
posted by Holy Zarquon's Singing Fish at 5:51 AM on January 7, 2016


Also if everyone cleared you, did you get executed?
posted by Segundus at 5:55 AM on January 7, 2016 [1 favorite]


Brings to mind police officer shootings.

King: Record number of police officers were charged with murder or manslaughter in 2015 — not a single one convicted

(Does anybody know what source=fark is doing in a Daily News link?)
posted by bukvich at 5:59 AM on January 7, 2016 [1 favorite]


That was really interesting, thanks. Examples 1 - 5 all looked good to me, but the sixth seemed a little strange: I think empirical models still rely on the unreasonable effectiveness of mathematics.

Yeah #6 downplays an ongoing intellectual controversy - as exemplified in Norvig v.s. Chomsky on the "two cultures of statistics".

On the one hand, computer-assisted methods such as simulation and machine learning are increasingly dominating scientific research, and their effectiveness apparently obviates the need for closed-form mathematics.

Yet theoretical computer science is an area where proofs about algorithms, etc., still reign as the best practice methodology. Theoretical physics is also math heavy, and that's an understatement.

It's hard to say - maybe Abbott sees this abundance of applied computing, and is allowing that to bias his assessment on the role of math. Maybe he hasn't heard, but Proofs are Programs, so what's the difference, really?
posted by polymodus at 6:08 AM on January 7, 2016 [1 favorite]


How does this apply to things like climate science in the media, where one of the prevalent memes has been "More than 90% of scientists agree, therefore it can't be true, because science doesn't operate by consensus"? My own take is that 90 to 95% is just about right, and 100% would indicate a statistically improbable level of agreement.
posted by sneebler at 6:21 AM on January 7, 2016


My own take is that 90 to 95% is just about right, and 100% would indicate a statistically improbable level of agreement.

What percentage of scientists think that the earth is round? I bet it's larger than 95%.
posted by Pyry at 6:48 AM on January 7, 2016 [7 favorites]


Basically, there is only a 1 in 100 chance that anything would be 100%. Therefore, if something is 100%, it is much more likely NOT to be 100%. Therefore it must be some other percent, say 14%. The chance that BOTH the 100% is wrong AND the 14% are wrong = 99/100 x 98/99 = 98%, which is a whole percentage point more certain than the 99%-wrongness of 100%. If we then steadily reduce the "14%" figure to 13%, 12%, 11%, and so on, we find that the chances of the correct figure actually being 0% = 99/100 x 98/99 x 97/98 ... x 85/86, which equals 85%. In other words, if you accuse me of being 100% responsible for certain horrible, disgusting crimes, it is mathematically 14% MORE LIKELY that I am ZERO% responsible for those crimes, and maybe those orphanage children covered THEMSELVES in 11 secret herbs and spices and JUMPED in the deep fryer. So if I can just have my belt, shoelaces and lemon-scented wipe back, I will bid you all good day.
posted by the quidnunc kid at 6:51 AM on January 7, 2016 [27 favorites]


What percentage of scientists think that the earth is round? I bet it's larger than 95%

I think this falls under the apples vs. bananas scenario.
posted by layceepee at 6:54 AM on January 7, 2016 [1 favorite]


My brain hurts now. Thanks, the quidnunc kid.
posted by gmb at 6:54 AM on January 7, 2016


What percentage of scientists think that the earth is round? I bet it's larger than 95%.

Well isn't it more pear-shaped really? It's actually, like everything else, kind of a hard question.
posted by zachlipton at 6:55 AM on January 7, 2016 [3 favorites]


My brain hurts now. Thanks, the quidnunc kid.

The part of me that was battered and fried by the quidnunc kid* hurts more, I'll bet, with at least 95% certainty.

* I begged him to kill me instead of just battering and deep-frying my arm, but he replied (with a touch of that devilish asperity that we so admire), "one does't eat a commenter like you all at once." True story! For a value of true approved by 23 of 23 Sanhedrin judges....
posted by GenjiandProust at 7:00 AM on January 7, 2016 [4 favorites]


Just popping in to clarify about my eating human flesh: at least 82% of people strongly agree with me that I have NEVER eaten human flesh. I have also been eating those people who who think that I DO eat human flesh, which is reducing the number of people who think that. So soon we will have a good, round 100% of delicious people who agree with me that I have NEVER eaten ANY human flesh, no matter how damn tasty it is. I know you are as hungry for certainty about this issue as I am - and allow me to assure you, my mouth-watering friends, that I am positively ravenous for certainty. YUM.
posted by the quidnunc kid at 7:25 AM on January 7, 2016 [8 favorites]


I don't doubt the premise here, but some of the examples they cite seem a little tenuous, especially when they handwave part of their explanation away with "Well, it doesn't apply when the answer is OBVIOUS, of course." What's the ratio of Heilbronn-Killer-style systemic misidentifications to actual serial killers? If I flip a coin 20 times in a row and get all heads, does that necessarily mean I made up my results?

Again, not arguing against the premise: eyewitness testimony is indeed notoriously unreliable, and the human drive to build unanimous agreement can create some terrifying things. This is not, however, a universal indictment of unanimity. Taking it to mean that is just as bad as the problem it's purporting to address. It smacks of James Surowiecki giving motivational speeches about magically locating missing shipwrecks, because Math! If you go sifting through thousands or millions of anecdotes, you're bound to find a few outliers. In fact, not finding extraordinarily unlikely outliers would imply, well, that the data set was tainted.
posted by Mayor West at 7:37 AM on January 7, 2016 [3 favorites]


Opens up an interesting reverse defence strategy.

Shoot the moon!
posted by freebird at 7:44 AM on January 7, 2016 [2 favorites]


I like this whole paper but.... does ancient jewish law really say that?

I would note that while Jewish law does say that, it applies strictly to capital cases. (A full Sanhedrin wouldn't be needed for lesser criminal cases, or for civil cases, and unanimous verdicts would follow the ruling in those.)
posted by Shmuel510 at 7:44 AM on January 7, 2016 [7 favorites]


The Talmud also rules out imposing the death penalty in cases where the crucial evidence is provided by the accused person's confession, and Maimonides specifically cites the danger of people with suicidal urges confessing to crimes they hadn't committed. There's a lot of psychological acuity in there that American law still hasn't caught up to, judging by how many people in the legal system still can't wrap their minds around the idea of false confessions. Not to mention their high incidence in overturned death penalty cases.
posted by ostro at 7:51 AM on January 7, 2016 [12 favorites]


But the prevailing interpretations hold that a unanimous verdict (of no less than 23 Sanhedrin judges) indicates judicial collusion

This view is illogical as if the judges wish to collude, they merely need to select one or a few of their number to offer token innocent verdicts.

There's another interpretation in the same paper (in addition to the ones listed by Beagle) that explains the law by emphasizing that the role of the Sanhedrin court is to find evidence of innocence not guilt ( sort of an extreme version of the better 10 guilty people go free than 1 innocent is punished idea) . So an unanimous guilty verdict shows that the court didn't perform its proper role. But this again could be avoided with token innocent verdict collusion. If there were never any token innocent verdict collusion , the institution in this interpretation would likely collapse under victims' rights arguments.
posted by Bwithh at 8:19 AM on January 7, 2016 [2 favorites]


I like this whole paper but.... does ancient jewish law really say that?

Here's the original. Scroll down for R. Kahana's comment.

But here's the thing that must be remembered about Talmud: It is not a book of law, it is a book of discussion of law, often over centuries, between rabbis who often had never met each other but couldn't disagree more. It is possible that this was the practice of the Sanhedrin during Kahana's time, which was the very end of the Sanhedrin. But this doesn't indicate that it was the case for the entire, thousand-year history of the institution, or that rabbis generally agreed with this.

There were a lot of laws that had to be passed, or judgements that had to be made, by consensus in the Jewish courts over time. It's worth noting that this particular anti-consensus opinion seems limited to capital cases, and the Sanhedrin was famously cautious when it came to capital cases, as demonstrated by tractate Makkoth 1:10:

"A Sanhedrin that puts a man to death once in seven years is called destructive. Rabbi Eliezer ben Azariah says that this extends to a Sanhedrin that puts a man to death even once in seventy years. Rabbi Akiba and Rabbi Tarfon say: Had we been in the Sanhedrin none would ever have been put to death."
posted by maxsparber at 8:45 AM on January 7, 2016 [7 favorites]


"A Sanhedrin that puts a man to death once in seven years is called destructive. Rabbi Eliezer ben Azariah says that this extends to a Sanhedrin that puts a man to death even once in seventy years. Rabbi Akiba and Rabbi Tarfon say: Had we been in the Sanhedrin none would ever have been put to death."

While I agree with your overall point, it seems a bit of a distortion to cut that quote off just before the final sentence: "Rabban Shimon ben Gamliel said: If so, they would have increased the number of murderers [lit. "shedders of blood"] in Israel."

(Which goes back to your point about the Talmud being the record of centuries of argumentation.)
posted by Shmuel510 at 9:05 AM on January 7, 2016 [2 favorites]


So the question is: is deep-fried human flesh Kosher or treyf? My not NOT voting for #1 quidnunc kid may or may not be dependent on this.
posted by briank at 9:22 AM on January 7, 2016 [1 favorite]


You know, you can actually go to the grave of Rabban Shimon ben Gamliel. I bet he's still arguing with Rabbi Akiba.
posted by maxsparber at 9:44 AM on January 7, 2016


My dim memory recesses suggest that there have indeed been been rabbinical rulings on cannibalism, but I'll have to look.
posted by tavella at 10:00 AM on January 7, 2016


Hmm, can't find the specific discussion I was thinking of (whether it was permissible to eat dead bodies to preserve human life c.f. Alive), but yeah, it's a discussed issue with various rulings.
posted by tavella at 10:05 AM on January 7, 2016


>But this doesn't indicate that it was the case for the entire, thousand-year history of the institution, or that rabbis generally agreed with this.

The bolded part cracks me up. I guess, according to the same line of thinking as this suspicion of unanimity, if EVERYONE had agreed that this was a good policy, that would have been a red flag that the policy wasn't being proposed in good faith.

Anyway, the issue of having one or more people disagree *just* to be the token "disagree-er" is that, in practice, people are going to be afraid to be that token. The reason that collusion would be tempting in the first place would be because there's a lot of social or societal pressure to stick with the party line and decide things a certain way. Nobody wants to be the one vote against something that is clearly in "everybody else's" self interest ("everybody else" other than the accused).

Distrusting unanimity is a good protection against developing a mob mentality (or meting out mob justice).
posted by rue72 at 10:07 AM on January 7, 2016 [3 favorites]


You know what they say: Two Talmuds, three opinions.
posted by maxsparber at 10:11 AM on January 7, 2016 [1 favorite]


Now I'm wanting to see a list of MetaFilter threads in which everyone agreed.
posted by straight at 10:29 AM on January 7, 2016 [6 favorites]


That was really interesting, thanks. Examples 1 - 5 all looked good to me, but the sixth seemed a little strange: I think empirical models still rely on the unreasonable effectiveness of mathematics.
Ned G

The standard response to the unreasonable effectiveness of mathematics is that there is simply so much mathematics (literally infinite amounts of mathematics) that it is only reasonable that some of it will be effective, and it is reasonable that we tend to pay more attention to those bits.

But coming back to the problem of unanimity as it applies to applied mathematics and statistics (and computer simulation etc.): Overfitting a model is very much a real danger.
posted by yeolcoatl at 10:40 AM on January 7, 2016


This post is well-timed for the not unanimous vote to send Ken Griffey Jr. to the Baseball Hall of Fame. Is his selection more legitimate since it was not unanimous?
posted by palindromic at 11:19 AM on January 7, 2016


Human meat may not be kosher, but if it is permissible to eat to save your life, the good news is, it's pareve.

Meanwhile, if everyone is telling you the Pope is performing your third wedding...
posted by Mchelly at 12:01 PM on January 7, 2016 [2 favorites]


The standard response to the unreasonable effectiveness of mathematics is that there is simply so much mathematics (literally infinite amounts of mathematics) that it is only reasonable that some of it will be effective, and it is reasonable that we tend to pay more attention to those bits.

But that's the same argument in the article - the journalist calls it the cherry picking bias. The counterpoint is that the amount of information and the degree of parameterizability are non-issues: for I could use only 1% of the features of my sonic screwdriver, or whatever, and still decide that as a tool it is super effective. Not much of it has to be used or mapped to reality; those are just a metric and not obviously logically relevant.
posted by polymodus at 12:08 PM on January 7, 2016


Human meat may not be kosher, but if it is permissible to eat to save your life, the good news is, it's pareve.

Mornay sauce, now is your moment!
posted by maxsparber at 12:17 PM on January 7, 2016 [1 favorite]


briank: So the question is: is deep-fried human flesh Kosher or treyf?

Mchelly: Human meat may not be kosher, but if it is permissible to eat to save your life, the good news is, it's pareve.

In event of starvation, Long Pork Buns may be served with Milk?
posted by zarq at 12:21 PM on January 7, 2016 [2 favorites]


Related (NYT, 2013/06/18): The F.B.I. Deemed Agents Faultless in 150 Shootings
But if such internal investigations are time-tested, their outcomes are also predictable: from 1993 to early 2011, F.B.I. agents fatally shot about 70 “subjects” and wounded about 80 others — and every one of those episodes was deemed justified, according to interviews and internal F.B.I. records obtained by The New York Times through a Freedom of Information Act lawsuit.

The last two years have followed the same pattern: an F.B.I. spokesman said that since 2011, there had been no findings of improper intentional shootings.

[...]

Critics say the fact that for at least two decades no agent has been disciplined for any instance of deliberately shooting someone raises questions about the credibility of the bureau’s internal investigations. Samuel Walker, a professor of criminal justice at the University of Nebraska Omaha who studies internal law enforcement investigations, called the bureau’s conclusions about cases of improper shootings “suspiciously low.”
posted by mhum at 12:31 PM on January 7, 2016 [4 favorites]


Interesting. Human breast milk is considered pareve.

Some of the other questions on the site could be spliced into a hell of a story: "Do I need to feed my cat kosher pet food?" / Is Human Blood Kosher? / Is the Lab-Created Burger Kosher? etc., etc.
posted by zarq at 12:44 PM on January 7, 2016 [2 favorites]


the journalist calls it the cherry picking bias. The counterpoint is that the amount of information and the degree of parameterizability are non-issues: for I could use only 1% of the features of my sonic screwdriver, or whatever, and still decide that as a tool it is super effective.
polymodus

No, a cherry picking bias is one selects and hides data to support a particular narrative. This is different, it's an attentional bias: People only see what what's important to them. The key distinction here is perception and choice. A cherry picker sees all the data and chooses to be biased. Someone with attentional bias is biased because they don't see some of the data, even though it was right in front of them.

And your point about sonic screwdrivers is not a counter point, it supports my point. Your sonic screwdriver is effective because one 1% of it is useful, Of course it's useful. I'm not saying it isn't useful. I'm saying it's ridiculous to be surprised that it's useful.

If you create something that can do so many things, of course some of the things it can do are going to be useful, so of course the tool is useful. And if you create the tool in response to problems that needs solving, of course it's going to be useful. The "unreasonable effectiveness of sonic screwdrivers" is what happens when someone only pays attention the the useful 1% and says "wow, it's so weird that sonic screwdrivers are so useful" when there is nothing weird about it being useful at all.

My problem with the "unreasonable effectiveness of mathematics" is the "unreasonable" part -- the sense of surprise. It's all completely natural and rational for people to focus on, further study, and develop things that can be useful to them. We find out that we can create a tool that helps understand something and then forget we created it and act surprised when it helps us understand things we created it for. We make it infinitely large, and also act surprised when it can be used to understand some things we hadn't anticipated when we created it.

It's unreasonable to think mathematics wouldn't be effective. But that's what happens when we focus on just the infinitely small useful fraction of mathematics, instead of mathematics as a whole, why mathematics was created, or the millions of people who worked to create it. And we focus on that infinitely small useful fraction of mathematics instead of all those other things, because that infinitely small useful fraction of mathematics is so gosh darned useful, and thinking about where that useful mathematics comes from isn't as useful. So we usually just don't think about it. Attentional bias.
posted by yeolcoatl at 2:12 PM on January 7, 2016 [2 favorites]


If you create something that can do so many things, of course some of the things it can do are going to be useful, so of course the tool is useful.

But this is completely fallacious. Versatility a priori does not imply a posteriori utility. There are two worldviews in operation, one while designing it and one using it in field. You could get catastrophic uselessness. Or "jack of all trades, master of none". So that would be a fallacy, and any trained engineer understands this intuitively.

And if you create the tool in response to problems that needs solving, of course it's going to be useful.

Except that mathematics is more accurately a co-evolved cognitive phenomenon. I, of course, empathize with your argument as presented, as well Abott's and the journalist's; however, I cannot on face value accept the assumption that mathematics can only be thought of as a human tool. That's an open scientific or philosophical issue.

For if the argument is that people are just forgetting that math is only an intellectual abstraction, then why not try arguing that the part of mathematics that we can observe (comprehend, even) only appears as if it is a tool? I would develop an argument that we have way to tell which is the case. Either stance requires a commitment, that others are not necessarily obliged to share.

The "tool" account of math presents many problems (pretty sure I can think of at least 5 issues). I can see the appeal, because it neatly resolves Wigner's rhetoric by claiming that the mystery is a tautology. But isn't that's partly the lesson of the article!?

I also disagree with the that argument lots of wrong equations or parameters means that math is not effective. That's not what effectiveness entails, for those could be interpreted as examples that any tool can be used the wrong way. Effectiveness is an issue external to that aspect.

And I disagree with the argument—albeit plausible and I sympathize—that infinitude of math implies these people are experiencing some kind of self-selection bias. Reason: Mathematics is infinite, but it is not arbitrary. Mathematics in practice is reducible to ZF/C set theory. So therein lies Wigners' question, in a slightly different form.

While we're at that, why not point out that Banach-Tarski geometry, long thought to be an obscure mathematical paradox, is now appearing in high-energy particle physics papers (in explaining how it is possible to get something from nothing). So, things are simply not clear-cut when we even begin an attempt evaluate the phenomenon of "math".

The best reason I have against the explanation given #6 is the Curry-Howard isomorphism. This mathematical insight has to do with what people even begin to think qualifies as math. That's another example of an a priori assumption, when trying to reason about these things.

Anyways I think Wigner's semi-tongue-in-cheek quote presents an interesting conceptual conflict (which is debated in earnest between people in STEM), and there's simply a lot going on, and I've tried to write down the few that I've thought of so far. The underlying question is a simple one: how should we best understand the relationship between mathematics and physical reality? "WTF is going on??" I also think it's important to be clear about giving plausible explanations, versus evaluating whether an argument is a logical justification that answers the question.
posted by polymodus at 7:57 PM on January 7, 2016


The tone (& substance) of the article bothered me, because I suspect he's butchering the paper. But I haven't read the paper* so I'm making some guesses. I'd love it if someone with more energy or statistical chops can confirm or demolish my interpretation.

I suspect the paper is positing that you have a test with a (reasonably high but not perfect) expected accuracy, but also a (reasonably low) chance of being just plain wrong. In this scenario, near-unanimity is unlikely, because you expect a failures. At some level of accuracy, the better explanation (statistically) is the scenario you thought was unlikely--you've got a screwy test.

To stick with the lineup scenario, if you know eyewitnesses are around 75% accurate (depending on vantage point, memory, etc.) a "good" test with 20 witnesses should give you 15 people agreeing and 5 picking a random person or no one. There's only a 1-in-300 chance of all 20 people being correct. So in this scenario it's way more likely you did something silly (or malicious) in running the lineup, and not that you have 20 people who have perfect memory and got a great look.

This makes sense to me. What I think the article butchers is by saying repeatedly thinking this means things like:

The take-home message is that, in a healthy democracy, when a party wins by a small margin, instead of name-calling the 'dumb' voters of the opposition, we should be celebrating the fact that the opposing voters preserved the integrity of democracy.

If you have 30% stupid voters (or 5% or whatever), and your election results give you 100% for the smart choice, of course this is evidence of fraud. It's a sign votes weren't counted correctly.

It does not mean stupid voters are helping democracy. Not being able to count them (since we all know they exist) would be bad sign for democracy. It does not mean their existence is in any way good, or that we couldn't do with them, at least not in the context of this paper.

This is literally confusing "correlation with causation," not in some cliched over-interpreting results way, but actually in the not-understanding-the-difference way.

Unless, again, I myself have messed up what the paper's arguing.

*I'm slow digesting stats arguments, and even if I weren't the paper explaining this mind-blowing claim is the next counterintuitive thing I'm going to try to work through.
posted by mark k at 9:03 PM on January 7, 2016


Metafilter: "any trained engineer understands this intuitively."
posted by sneebler at 6:40 AM on January 8, 2016 [1 favorite]


It does not mean stupid voters are helping democracy. Not being able to count them (since we all know they exist) would be bad sign for democracy. It does not mean their existence is in any way good, or that we couldn't do with them, at least not in the context of this paper.

This might be over my head, but I took it to mean that the their presence helps us know that democracy is working, like a canary in a coal mine. Not saying "I’m glad the bird died, we’re saved". It seems like you could say "The bird dying saved our lives" in that it let you know about the problem, but the act of it dying did not magically save you. Having confidence in the electoral system is crucial to it functioning.
posted by bongo_x at 9:10 AM on January 8, 2016 [1 favorite]


« Older A Victorian booty call: Come see our new Lamp. You...   |   "Same-sex" in This Instance Not Entirely Accurate Newer »


This thread has been archived and is closed to new comments