My reaction to the Norwegian 22th of July massacre #3 - the bug (warning:statistics)

[clray123]

On another note, with a little practice it is rather easy to identify conspiracy theories and dismiss them outright based on feature similarity to other such theories. But there is also possibility for erring in that direction. Are there good examples out there of theories that were falsely dismissed as CT early on, but actually turned out as accurate later? (theories that were complex, sinister and right)

[trondreitan]

@clray123 Hmm, I think the Iran-Contras scandal (selling weapons to Iran to finance a war by proxy in Nicaragua) in the 80s were considered a conspiracy theory for a while. The scope and the complexity wasn't huge, but there had to be a few and it did go on for a while. Denying the US claim of WMDs in Iraq(2002) might also have been called a conspiracy theory by some, but those claims could have come from severe confirmation bias instead of outright lying, so it might not have been a conspiracy.

[wisling]

@trondreitan Hey, I agree with what you've said about how people react to conspiracy theories.  As to whether or not this was the motivation behind the massacre, I am not sure, but I think there's something related to statistical reasoning to be said here. In your video, at about 8:55-9:15, you mention that adding complexity can perfectly account for

[wisling]

what people observe and this holds more value to us(when it is sinister) than a simpler explanation that accounts for what is observed(that is not sinister). I think another large reason for this skewing our view is our lack of creativity in coming up with other scenarios that are just as complex but explain the data just as well. If we could do this, we would give equal weight to all the complex scenarios, which would be small as they are so numerous.

[wisling]

-Our view is basically-

Complex scenarios - only one and fits the data well, and sinister with much effort put into the explanation

Simple scenarios - only one and fits the data well, but not sinister

-But Realistically-

Complex scenarios - hundreds that fit the data well, all sinister and much effort put into the explanations

Simple scenarios - only one and fits the data well, not sinister

,in which case we would have no reason to accept one complex scenario over another.

[trondreitan]

@wisling Well, stuff that happens in human society will always be complex, though when I say complex models, it's about how much of a complicated pattern is seen (instead of complicated things happening arbitrary reasons having nothing to do with a pattern). As I said, the simple (pattern-less) models should be taken as the default position, but our mental bugs may stop us from seeing this and going for it.

[trondreitan]

@wisling (cont) However, I kind of like your approach, as a way of revealing the buggy thinking. If someone sst with a conspiracy theory they liked, why not see if another conspiracy theory also can be seen to "fit". Or how about a secret cabal to improve mankind? Or aliens? Or an overall secret plan by a deity? When all these complex scenarios turn out to work equally well, that should reveal their weakness as well as their seeming strength. Thanks for the input.

[clray123]

Another way to put it would be that conspiracy theories, like religions, explain everything, but predict nothing. That is why they can accomodate any data whatsoever.

[trondreitan]

@clray123 Thanks for the input! It's a very good expression you gave in your first sentence. If I should get technical (and Bayesian) I would say that conspiracy theories(CT) can predict, it's just that when one don't do prior-data feedback, they predict very very vaguely (and poorly). Also, a theory where each new data is incorporated into the theory by a new fudge factor (religion might be an even better example than CT), it doesn't really explain anything, it just sounds explanation-like.

[trondreitan]

@clray123 (cont): Just to clarify that last sentence of mine. What I look for in an explanation is a set of assumptions which leads to what is to be explained and which has previously been established as (at the very least least) plausible. When the theory can and do accommodate any new fact by using a new fudge factor, rather than using some of it's previously established factors, it thus fails as an explanation in my eyes. Though it may still sound like one. What do you think?

[clray123]

@trondreitan I agree - a good explanation should imply all the known facts while precluding other (hitherto unknown) facts. It should say just as much about what happens (or happened) as about what does (or did) *not* happen (similar to how a good test must avoid both false positives and false negatives). Thus a good explanation certainly is predictive (even when the prediction concerns past, not future facts), but not all explanations are like so - bad explanations just (over)fit what's known.

[websnarf]

If you think its down to a conspiratorial obsession, then do you think maybe you can act against it by teaching a class in rational thinking in school? There is a class in university (at least at mine) that teaches critical thinking class that includes symbolic logic etc. But it seems to me that such a course, could be moved to high school, and expanded to include material like that in Sagan's "Demon Haunted World".

[trondreitan]

@websnarf A class in rational thinking sounds not just as a great idea, but pretty much essential, even at early grades. Sadly, I do not know of any country which teaches this at elementary school. We were taught a little bit about critical analysis in literature and history at high school here in Norway, but that was it. Something like Carl Sagan's baloney detection kit would be very nice for a start (though as in a previous video, I would cut "absence of evidence isn't evidence of absence").