Surely low enough not to be overcome by you being impressed or you agreeing with his philosophy
“Here is a very simple example of Bayesian reasoning, that most people are in fact capable of. Suppose we draw a random number between 1 and a million; the prior for any particular number between 1 and a million is straightforwardly very low—one in a million, of course. Now, I have just generated the number 493250 using random.org. Surely this prior of 1 in a million that I have generated any specific number like 493250 is low enough to not be overcome by you being impressed by looking at this comment and see ’493250′ in it? The prior for you having very special powers of perception of the right number is likewise proportionally low to how very special it is, and so on.”
“Here is a very simple example of Bayesian reasoning, that most people are in fact capable of. Suppose we are looking at people who write clip art web comics; the prior for any particular clip art being the best or most popular is straightforwardly very low—one in a million, say, or what ever is your number. Now, we look at http://www.qwantz.com/index.php Surely this prior of 1 in a million is low enough to not be overcome by you being impressed by looking at this Dinosaur Comics? The prior for you having very special powers of perception of clip art is likewise proportionally low to how very special it is, and so on.”
The ensuring debates and demands for evidence that something with very low prior isn’t true, are particularly illuminating with regards to just how incapable certain self proclaimed Bayesians are of the most basic forms of probabilistic reasoning.
Yes. I agree. Some of these self proclaimed Bayesians cannot even fully specify their examples or prove their arguments or explain the crucial part of what they were probably arguing.
“Here is a very simple example of Bayesian reasoning, that most people are in fact capable of. Suppose we are looking at people who write clip art web comics; the prior for any particular clip art being the best or most popular is straightforwardly very low—one in a million, say, or what ever is your number. Now, we look at http://www.qwantz.com/index.php Surely this prior of 1 in a million is low enough to not be overcome by you being impressed by looking at this Dinosaur Comics? The prior for you having very special powers of perception of clip art is likewise proportionally low to how very special it is, and so on.”
So, putting the analogy into reverse, the top post is wrong. You can judge N levels above your own.
I was making the point that Dmytry’s claim was flawed in 2 separate ways; ‘you can judge N levels above your own’ is closer to the point of the random.org example than the DC example. (The DC example was more about neither DC nor EY being a random selection, not the strength of personal judgment.)
“Here is a very simple example of Bayesian reasoning, that most people are in fact capable of. Suppose we draw a random number between 1 and a million; the prior for any particular number between 1 and a million is straightforwardly very low—one in a million, of course. Now, I have just generated the number 493250 using random.org. Surely this prior of 1 in a million that I have generated any specific number like 493250 is low enough to not be overcome by you being impressed by looking at this comment and see ’493250′ in it? The prior for you having very special powers of perception of the right number is likewise proportionally low to how very special it is, and so on.”
“Here is a very simple example of Bayesian reasoning, that most people are in fact capable of. Suppose we are looking at people who write clip art web comics; the prior for any particular clip art being the best or most popular is straightforwardly very low—one in a million, say, or what ever is your number. Now, we look at http://www.qwantz.com/index.php Surely this prior of 1 in a million is low enough to not be overcome by you being impressed by looking at this Dinosaur Comics? The prior for you having very special powers of perception of clip art is likewise proportionally low to how very special it is, and so on.”
Yes. I agree. Some of these self proclaimed Bayesians cannot even fully specify their examples or prove their arguments or explain the crucial part of what they were probably arguing.
“Here is a very simple example of Bayesian reasoning, that most people are in fact capable of. Suppose we are looking at people who write clip art web comics; the prior for any particular clip art being the best or most popular is straightforwardly very low—one in a million, say, or what ever is your number. Now, we look at http://www.qwantz.com/index.php Surely this prior of 1 in a million is low enough to not be overcome by you being impressed by looking at this Dinosaur Comics? The prior for you having very special powers of perception of clip art is likewise proportionally low to how very special it is, and so on.”
So, putting the analogy into reverse, the top post is wrong. You can judge N levels above your own.
I was making the point that Dmytry’s claim was flawed in 2 separate ways; ‘you can judge N levels above your own’ is closer to the point of the random.org example than the DC example. (The DC example was more about neither DC nor EY being a random selection, not the strength of personal judgment.)