/lit/ - Literature

>>15299939
The solution to epistemology is that knowledge isn't possible

E=mc^2

boom. solved.

BOOM.

>>15299939
The Lesswrong crowd has ruined Bayes for me.

>>15299949
if knowledge is not possible, then so is the knowledge that it's not possible
>>15299951
that has nothing to do with epistemology

>>15299958
Going to talks from mathematicians has ruined Bayes for me.

>>15299961
>if knowledge is not possible, then so is the knowledge that it's not possible
No, as the idea of knowledge is phenomenological, not noumenological.
You can not understand how things actually are, you can only understand representations in your mind. This is not real knowledge. REAL knowledge is not possible, only phenomenological representations are possible.
Also, all reputable statisticians reject Bayes' "theorem".

Can someone translate this for a pleb?

>>15300069
don't bother, it's just wank

>>15300069
That's simply Bayes' Theorem. It's a simple probability equation, I'm sure you could find a lot of explanations of it for laymen. Saying that it solves epistemology is basically the equvalent of saying that Occam's razor solves the existence of God.

>>15300060
>You can not understand how things actually are, you can only understand representations in your mind.
If our minds is so divorced from 'reality', what makes you so sure our words refer to this 'reality' in the first place?

>>15300060
>Also, all reputable statisticians reject Bayes' "theorem".
Lol.

>>15300060
>Also, all reputable statisticians reject Bayes' "theorem".
elaborate

>Also, all reputable statisticians reject Bayes' "theorem".
I would've expected it to be some sort of Nassim Taleb's arrogant 'I'm going to enter your field without any prerequisites and completely reject half of it', but Taleb himself is literally a bayesian.

>>15300399
It's a mathematical truth. "Rejecting" it would be like rejecting 2+2=4.

>>15299951
based? noeticpilled?

>>15300529
fuck off

>>15300069
I don't know what this gay explenation in OP is, but .. P(a|b) means te probability of event a given that you know event b has ocurred. Bayes formula shows how you can calculate this. The point is that some information about what has ocurred gives us better predictions to know wether an other event has ocurred. Unless the two events are completely independent ( p(a^b) = p(a)*p(b) ) , so that p(a|b) = p(a)

>>15300060
> Also, all reputable statisticians reject Bayes' "theorem".

the fuck are you talking about? its logic is commonly used for medical research .

haha bayes goes brrrr

>>15299939
Someone post the macro showing that since two identical things can not exist, numbers are absurd and math basically can't be applied to reality.

>>15302703
> math basically can't be applied to reality
is this some gay cope for smallbrains who failed math?
You wouldn't be able to use your computer if math couldn't be applied to reality

>>15299939
based

>>15299999
>>15300000

>>15299939
except that it doesn't, even if it wasn't retarded

>>15300060
Debate isn't around Bayes' theorem but around "Bayesianism" which is a different albeit related thing

>bayes
lol

>>15302723
Only Angloid bugmen care about such things.

>>15302935
What's so funny?

>>15303027
If you want a real life example of Bayesian statistics failing hard, look at the coronavirus models.

>>15302723
Literally brainlet who doesn't understand high tier mathematics.

>>15299939
Why the hell does this image not simply write the denominator as Pr(E)? What does it gain by being written this way? It looks longer and more scary to the uninitiated when really it's just stock-standard Bayes' Theorem?
Besides, trying to identify formal probability as a distribution over a given formal event space with real life notions like "evidence" is a retarded positivist idea to begin with. Among many issues, you can't do things like quantify the real-life event space or distribution.
>>15300060
Bayes' Theorem is so simple as to be tautological (yes, I understand that formally all such mathematical equations are tautologies). It's just P(A⋂B) = P(B⋂A) expanded out with the definition of P(A|B). There's no deep meaning to it or anything.
>>15300069
"The probability of "A and B" when you adjust for how likely B is, is just as likely as the probability of "A and B" when you adjust for how likely A is." This is because the event A and B is the same event either way.
Very simple. In this case with the meanings given here, it means "The probability that your evidence supports your hypothesis, given how likely the evidence is to occur randomly, is the same as the probability of how likely the evidence is to occur when the hypothesis is true, given how likely the hypothesis is to be true." Which you immediately observe is a stupid interpretation to give to the formula because how on earth can you quantify "how likely the hypothesis is to be true" in a vacuum (that's Pr(H)). When applied to real-life things like this it's a nonsensical statement.

People making mathematical claims need to actually know the mathematics. Then they would know that the mathematical claims have very rigorous settings wherein they can be applied and that real life isn't one of them.

>>15303046
In real life, you don't know actually how likely anything is. So you're adding your expectations and destroying the observations. The result in practice is always the same, failed models.

>>15303031
Which ones? Link?

>>15300060
>Also, all reputable statisticians reject Bayes' "theorem".
Dumbest thing I've ever read on this site. Let me guess, you took "stats for psychology" or "stats for sociology"?

>>15300069
Ignore the shit about epistemology, the theorem is just a basic theorem in probability theory for determining the conditional probability of a hypothesis given the evidence.

Say our hypothesis is that OP is a fag, and the evidence is that he made this stupid post. Then the theorem states that the probability OP is a fag, given that he made this post, is equal to the (probability that a fag would make such a post multiplied by the probability a random person is a fag), divided by (the probability a fag would make this post multiplied by the chances a random person is a fag plus the probability a normal person would make this post multiplied by the probability that a person is normal.)

Essentially the theorem tells is how sure we should be that OP is a fag, now that we've seen him make this dumb-ass post.

>>15303031
The model for Italy was so bad, it was literally failing to predict the next day. Why? Because instead of looking purely at the data (the days before and their real measured trajectory), it also injected their retarded a priori expectations that "it should be 0 already". The result was hilariously low predictions that got beaten every single day. That's Bayes for you when you actually try to apply it, instead of wanking over the philosophical implications. That's why top statisticians will warn you about misusing it save for very very specific circumstances.

>>15303072
http://archive.is/www.healthdata.org

Because the model is dynamic, archive.is is the only place that stored forever just how bad their predictions were every single day.

>>15303088
That's an indictment of retarded priors, not Bayes theorem.

>>15303119
It's an indictment of Bayesian modelling overall if it's so hard to create good priors. Bayes' Theorem itself is trivially true in any case, of course.

>>15303184
>if it's so hard
It's not hard. The world just has a lot of retards running about. What in your opinion is a better modelling paradigm?

>>15303031
Bayesian statistics can't account for fat tails

>>15303184
Even if retarded priors bayes work, you just have to update them with every iteration.

>>15303184
It's not hard, people are just stupid. The White House literally fitted a cubic polynomial to the recorded deaths today

>>15299939
This is some real anglo shit.
Read German Idealists you midwits.

>>15299939
Aristotle already solved it more than 2000 years ago when he formulated the law of noncontradiction

>>15303246
Lmao.

>>15303246
>Bayesian statistics can't account for fat tails
What the fuck are you talking about.

>>15299958
Just forget Less Wrong and read Bruno De Finetti's writings

>>15300060
>Also, all reputable statisticians reject Bayes' "theorem".
Way to show you know jack shit about what you're talking about.
So this is the power of philosophers. You guys are even worse thanetic physicists, who think they're better than everyone at every field.

>>15300060
the noumenon doesn't exist.

>>15300060
What about analytic truths?

>>15303307
>>15303308

>>15303346
>So this is the power of philosophers.
Most philosophers are Bayesians, dumbfuck. The person you are responding to has Downs.

What the fuck are anons even arguing about in this thread? What books can I read to get in the know?

>>15304690
In terms of philosophy?

>intro articles
https://plato.stanford.edu/entries/bayes-theorem/
https://plato.stanford.edu/entries/epistemology-bayesian/

>intro books
Colin Howson: 'Scientific Reasoning: The Bayesian Approach' -- https://www.amazon.com/Scientific-Reasoning-Bayesian-Colin-Howson/dp/081269578X
John Earman: 'Bayes or Bust?: A Critical Examination of Bayesian Confirmation Theory' -- https://www.amazon.com/Bayes-Bust-Critical-Examination-Confirmation/dp/0262519003

>more
https://philpapers.org/browse/bayesian-reasoning?freeOnly=&newWindow=on&new=1&showCategories=on&sort=impact&langFilter=&start=0&onlineOnly=&cn=bayesian-reasoning&categorizerOn=&limit=50&filterByAreas=&proOnly=on&cId=5879&publishedOnly=&sqc=&hideAbstracts=&format=html&jlist=&ap_c1=&ap_c2=

>>15304747
Much appreciated anon, have a mouse

>>15303475
Again, that's about retarded priors not Bayes.

>>15300069
start with a guess, refine said guess as more evidence comes to light.