It's a bastardized summary of a technical paper; this is not a surprise if you've ever read _USA Today_ trying to explain what XML is.
That said, they link the paper so you can read it for yourself. So far it makes a lot of sense to me; turns out you can eliminate a great many buildings and locations by selecting a few characteristics.
For example, bin Laden's height, need for dialysis, entourage, and defensible walls rules out all but 16 structures in the Kurram region (http://en.wikipedia.org/wiki/Kurram_Agency) if you look for only 5 out of the 6 criteria. (Page 4 of PDF.)
As for your specific criticism; you're quite right. The OP has misstated his link, which says:
> According to a probabilistic model they created, there was an 88.9% chance that bin Laden was hiding out in a city less than 300 km from his last known location in Tora Bora: a region that included Abbottabad, Pakistan, where he was killed last night.
That's much more sensible, and does follow directly from their geographical theory - if you think of animals as random particles moving in Brownian motion, there is an exponential dropoff. Random walks don't go very far.
On a side-note, they were largely right about the room characteristics. The compound bin Laden was in was in fact, a tall building with multiple structures with more than 3 rooms and walls over 3 meters tall (http://en.wikipedia.org/wiki/File:Osama_bin_Laden_hideout.jp...).
In fact, the only characteristic I see missing is 'trees for cover when outside' (few trees in http://en.wikipedia.org/wiki/File:Osama_bin_Laden_hideout.jp... ) and arguably, the need for aerial cover was met by being so near the military academy. (I've seen reports and speculation that the helicopters had to spoof Pakistani transponders to avoid triggering air defenses.)