things that have happened to me recently

[toaster_boy]

Went on the canoe trip this past weekend, had lots of fun. Barefoot soccer really beat up my feet though; Sunday morning I discovered I had lots of tiny cuts which didn't go all the way through the thick skin, but combined to make it feel like I had a giant bruise covering my entire foot. At least as long as I was barefoot; wearing shoes mostly

Comments

[arcagentraphael]

My video card recently went haywire and I'll probably be getting another one soon. Is the RAMDAC burning out something that would cause horizontal blurs across the screen? Cuz that's what I've got.

[toaster_boy]

That sounds exactly like the problem I had: dim picture, with regions of light or dark in the image smeared all the way across the screen.

I don't actually know it was the RAMDAC, but it's a pretty good guess, I think.

[lizwalmart]

I won't be canoeing on the 16th, but have a good time and dunk Chris for me. :)

I'm eager to see Batman Begins.

[miles_of_bc]

http://en.wikipedia.org/wiki/Bayesian_inference

http://www.stat.ohio-state.edu/people/faculty/snm.html

http://www.stat.ohio-state.edu/people/faculty/mb.html

Well the problem deals with prior probabilities...so if you check out Bayesian statistics you might find some answers to the logic of the solution. It sounds like a straight forward example of this increasingly popular branch of statistics so I highly doubt it is a april fools joke

[toaster_boy]

Clearly the problem is designed to sound that way, and the author does a very good job up until (s)he attempts the "simple algebra" at the end. It is incontrovertible fact, however, that the solution given is incorrect. It can be easily verified experimentally (I didn't run the simulation; Dad did), and experiment proves that switching your guess doesn't help you.

Actually I'm quite familiar with Bayesian inference, though for entirely unrelated reasons. Among the many things sprouting from the new popularity of Bayesian inference have been numerous so-called 'naive Bayesian' approaches to automatic classification of, for example, spam versus non-spam emails. The traditional, more powerful (theoretically), but much more numerically intensive correlative methods are very difficult to justify for that engineering space. I still have a soft spot for them, though; as mathematical objects the old methods are just too unbearably cool to ignore.

Sigh.... so many years left until IBM's prediction-by-partial-match patents expire.

[miles_of_bc]

Here's a similar problem and solution from MAA

http://www.maa.org/devlin/devlin_07_03.html

regardless...it's just messing with people's assignments of prior and posterior probabilities

Fundamentally I think it should still be 50/50 though...

[curlyjones]

Yay! April's coming! I haven't seen her in forever. I'll catch you both on the 16th.