Monday: Where did that hour go?

Gone.  I thought it was like 9:15.  Turns out it's 10:26.  Oops.  Let's do the quick list of things that are better today than before:

• A/C back at work.  My office is now suitable for things to be done.
• Because of that, I largely finished Project A, I just need to test it.  Woo proper productivity temperature.
• My neighbor's sink is clogged too, and so they're going to fix it from that side.  Not something I need to worry about, I guess.

"Attack!"

"WTF was that supposed to be?"

• Mitt Romney: still a fucking idiot.
• This guy: also a fucking idiot.  Let's go slowly, in case he tracks incoming links:  You want to know the true population results, but know that you have three segments of the population P = [p1 p2 p3], each of which have a fraction Pf = [A B (1 - (A+B))].  Now, you sample a number of people related to the accuracy you're looking for.  This is generally a Poissonian distributed thing, so you get an idea of how big a sample you need from N ~ 1 / (error^2).  Let's say we want a good result without having to ask a bunch of people.  3%.  Sure, therefore we need something like 1000 people.  Nice round number.  Now, we ask those 1000 people if they like option X or Y better, and then ask them to identify as a member of populations P.  Now, you take the counts of population K = [k1 k2 k3] as well as the fraction of that population supporting one of the binary options F = [x1 x2 x3].  Now, assume you have the population fractions given above.  Now, calculate a weight factor from W = Pf / K.  Now do the weighted mean: sum(W .* F) / sum(W).  There.  You've corrected your survey based on the expected real population based on the observed population sampling.  Boom.  All that shit about "oversampling" works out to be fucking nonsense, isn't it?  Now, exercise for the reader: calculate the expected fraction errors for each population segment and not for the entire survey.
• This dog: best use of stairs ever.
• Michael Alexander Salzhauer, MD: pretty fucking creepy.
• I like how the postdocs are largely in the "unlikely to get a faculty position" category.
• I can't tell if this is "cool" or "cute."  Coolute?  That's not a word, is it.  Oh well. Licia Ronzulli: best Member of European Parliament ever.