Friday, September 28, 2007


Star (Trek) Crossed Pundits

Recognizing the polyscifi in the blogosphere.

This weekend is Star Trek weekend on national review.

Sample articles:


Wednesday, September 26, 2007


Ultimate Showdown

of inconsequential destiny.

The Furries are facing off against the Klingons. Again. Shockingly, there's more here.


Sunday, September 23, 2007


Confirming My Biases

Stories which confirm your biases are always a dangerous thing as it leads you to overlook confounding evidence, but this article still tickled my fancy because it confirmed two things which I've long believed but lacked decent empirical evidence to support.
  • First, CEO (or President depending on the incorporation model) performance has an enormous impact on company performance.
  • Second, the death of family members can either be traumatic or liberating depending on the context.
I should note that the stats are for Denmark and that the researchers don't think the mother-in-law effect was statistically significant, but it certainly is an interesting study.



Wednesday, September 19, 2007


We don't need no stinking arms....

Chuck Norris, Jack Bauer - they've got nothing on this guy.

The intrepid staff at polyscifi managed to secure a photograph which we believe could be of the actual fight. However, at this time, we are unable to verify its authenticity. In there interest of expediency, we have placed it in the extended entry and will let you, the reader, make up your own mind.

(show photo)



Saturday, September 15, 2007


More comparative health care stats

The Wall Street Journal weighs in on the Lancet study (unnamed in the publication, but that's surely the same study)

Last month, the largest ever international survey of cancer survival rates showed that in the U.S., women have a 63% chance of living at least five years after diagnosis, and men have a 66% chance -- the highest survival rates in the world. These figures reflect the care available to all Americans, not just those with private health coverage. In Great Britain, which has had a government-run universal health-care system for half a century, the figures were 53% for women and 45% for men, near the bottom of the 23 countries surveyed.
They also mention a study from Respiratory Medicine:
A 2006 study in the journal Respiratory Medicine showed that lung cancer
patients in the U.S. have the best chance of surviving five years -- about 16%.
Patients in Austria and France fare almost as well, and patients in the United
Kingdom do much worse with only 5% living five years.

I'll see if I can get a hold of the original publication and flesh out the details like I did before. But, caveat lector...

I think this is the RM article. Upon reading it, it confirms the cited stats, but the actual thrust of the article is that the reporting standards across countries are inconsistent with the authors suggest that the US methodology may be overstating results. FYI, here's the sorted average 5-yr survival for lung cancer as self-reported by the countries:
US 16%
Austria 14.7
France 14.5
Spain 12.6
Germany 10.65
Northern Ireland 9.85
Denmark 6
England/Wales 6
Scotland 5.85



Lies, Damn Lies, and Statistics

I always thought that 73% of all statistics were made up on the spot.

But who knew the phenomenon extended to science journals too?

Labels: ,


Friday, September 07, 2007


A programming hypothesis

Admittedly formed after poking around with it for an hour and still not being able to run someone else's code, but I still think the following insight is valid:

A programmer's love of Python is approximately equal to his love of Linux.

I hate Linux. But I know lots of people who swear by Python (as opposed to swear at). They're all Linux enthusiasts.

(Background: The first time I did it, it took me 30 seconds to figure out how to make a nice looking GUI in Matlab. 1 minute to use MFC classes in C++. 1 minute to make a GUI in Java. I can't even run someone else's Python compiled GUI because I can't find out what I need to tell Python to make the code/modules readable. The fact that I had to go out and find modules also annoys me to no end. I'm also annoyed that the little I have poked around with coding in Python has revealed all sorts of annoying inefficiencies - like how it handles loops. Seriously, having to copy anything rather than passing by reference rubs me the wrong way. Cycles count. Memory matters. Now I know why Civ4 is such a slow, buggy, memory-hogging monster.)


This page is powered by Blogger. Isn't yours?