Friday, October 31, 2008

Happy Halloween!

A coworker passes this along:

Ghosts in the Library!



Oh to be ... irony-free

Pharyngula points us toward this image:

Jesus People Pray That False Idol Will Save God’s Economy


If you can't afford a golden calf these days, the bronze bull on Wall Street will have to do.

Thursday, October 30, 2008

My take on Obama's infomercial

I'm no great chess player, but one of the lessons I've been taught is that when you have a winning position, you take the time to "eliminate counterplay." Since things are going your way, your opponent would like to launch a counterattack, make some threats that distract you from your own plans, hope you make a mistake - anything to cause chaos, 'cause things are going your way; unless he changes something, he's gonna lose. The good player isn't in a hurry; he pauses to make sure there'll be no effective counterattacking and then proceeds to crush the helpless foe.

Obama's infomercial was an unspectacular, yet brilliant move that I'm betting has eliminated McCain's counterplay. A half hour of calmness, reassurance, gravitas - dare I say it? Class - is going to stonewall McCain's attempts to make people nervous about Obama. He can only look petty, mean, and small in comparison if he continues his attacks. And yet he has no other play - Obama has the better position and if McCain doesn't attack, he simply loses in the obvious way. So McCain has two ways to lose, but if I read this right - that those wavering people who just might have gone over to McCain at the last minute are now, in fact, safely in Obama's camp - then he has no way to win. A chess master at work.

Salmon survival

Dams Not Main Cause of Salmon Collapse, Study Says

This would be rather a surprise, if confirmed.

Wednesday, October 29, 2008

Bad map. Bad bad map.

I came across this map of African-American population that unintentionally illustrates some bad practices:


First off, the map uses confusing symbols. The colors are supposed to illustrate the percentage of Black population by county. You aren't told that the breakdown is by county - you just figure that out after looking at the less-populated areas of the country. Which brings up point number two: using large dots which actually obscure much of the map, and obscure each other. I can get an overall impression of Black distribution across the country, but I can't examine any of the Southern states to get a more precise look at the demographics - the symbols obscure more than they reveal. The large dots are meaningless in this context, which is error number three. They may be proportional to actual population - in fact, that seems the most likely explanation - but there's no information about that at all.

Bad practice number four isn't all that awful, but it's still worth mentioning: the categories break at very odd percentage values. I mean, really - 0.09% instead of 0.10%? What's happened is that the categories have been broken down along strict quintiles - the lowest 1/5 of percentages happens to top out at 0.09, so that's the breakpoint. Better practice, though, is to look for either natural break points (if the data tends to cluster very much) or just fudge to rounder numbers so you don't give a false sense of meaning to the category boundaries.

And finally, those colors. Given the association of red/yellow/green with danger/caution/safety in our culture, what can the map creator possibly have been thinking in assigning this color scheme to such a map? Ai yi yi yi yi.