Comments on: Offended by conditional probability

By: Success » Diigo Links (weekly)

Success » Diigo Links (weekly) — Mon, 29 Apr 2013 20:55:33 +0000

[...] Offended by conditional probability — The Endeavour [...]

By: SiPlus

SiPlus — Tue, 19 Feb 2013 07:41:50 +0000

What about Notepad++ users?

By: Ed

Ed — Sat, 16 Feb 2013 16:50:27 +0000

Every OS needs a good text editor: implementing Vim under Emacs-Evil.

http://developers.slashdot.org/story/13/02/16/0251239/evil-almost-full-vim-implementation-in-emacs-reaches-10

By: John

John — Thu, 14 Feb 2013 12:10:04 +0000

Dave and Seamus: Good points. When I say “makes,” I mean “causes the rational person to increase their estimated probability of.” I agree with Jaynes’ view that probability is not a statement about nature but rather of our understanding of nature.

By: Seamus

Seamus — Thu, 14 Feb 2013 07:27:38 +0000

“If A makes B more likely then B makes A more likely” is only true when the “makes” is read as being about how your subjective degree of belief changes on learning one of the events. If it’s read as something akin to “causes” then you got obvious absurdities: Going to MIT causes you to be better at programming, therefore being good at programming causes you to have gone to MIT. Clearly crazy.

On a related topic, here’s an interesting paper: http://www.colyvan.com/papers/shonubi.pdf

By: Chris Brew

Chris Brew — Thu, 14 Feb 2013 02:47:58 +0000

People who are offended by conditional probability are often also offended by symbolic logic. In the same way that logic presents itself as a distillation of patterns of valid argument, probability presents itself as being about patterns of plausible inference. Yet, for logic there is always the option of giving up on argument and building up a model-theoretic account, and for probability you can give up on inference and spend your time reducing things to measure theory and sigma algebras. In neither case does the ensuing activity feel right as a way of capturing what goes on when people construct arguments or make inferences.

By: Dave Backus @ NYU

Dave Backus @ NYU — Thu, 14 Feb 2013 02:09:25 +0000

When you say “being an Emacs user makes you a better programmer,” I think you want to avoid the word “makes.”

By: SteveBrooklineMA

SteveBrooklineMA — Thu, 14 Feb 2013 01:40:41 +0000

I agree with the comments above regarding the translation from everyday speech to probability and math. Many of the probability problems I’ve seen floating around the net (e.g. “Tuesday boy”) seem to boil down to differences in translation.

By: Derek Jones

Derek Jones — Wed, 13 Feb 2013 23:51:36 +0000

You have fallen into the trap of thinking that human communication is governed by such a simple Bayesian rule. Check out the works of Paul Grice and what he had to say about Conversational implicature. Sperber & Wilson's "Relevance communication & cognition" is worth a look.

By: John

John — Wed, 13 Feb 2013 23:21:01 +0000

Peter: Maybe so. I never said that, but someone may have heard that.

It’s amazing how touchy the Emacs/Vim thing is. Ironic too since both Emacs and Vim are obscure corners of the universe as far as Visual Studio and Eclipse users are concerned. I guess that’s how it always goes: sibling rivalries are the most bitter.

By: Peter

Peter — Wed, 13 Feb 2013 23:02:04 +0000

Perhaps the complaint of the above Vim users is better formalized like so:

Prob( good programmer | Emacs user ) > Prob( good programmer | Vim user )

By: rdm

rdm — Wed, 13 Feb 2013 19:59:09 +0000

I think it’s a mistake, though, to assume that the rules of english grammar correspond to the rules of probability.

One issue here is that contexts shift between different parts of an english phrase or discussion while probabilistic inference (such as you are describing, above) is only valid within a fixed context.

So, while it’s possible to use english to discuss probability (if we are careful) it’s easy to fall into problems if we do not carefully address contextual issues as perceived by our audiences.