This is a great reference, in theory. As a bunch of people have said, there are a lot of reasons we'd expect manufacturers' stated power numbers aren't going to be comparable (which is a delicate way of saying truthful).
Here's another anecdote (actually, let's say it's a hypothetical, since it is stupid and dangerous to race on public roads, even in the middle of nowhere, where I certainly would have been had I tried such a stupid thing): Just yesterday I thought a car (APR stg 2 w/ RSC TBE) felt faster than mine (Revo stg 2 w/ downpipe, no cat-back) when I took a ride in it, so we ran them to see if it was true. From 30 and then 20 to about 100-110 we were within a half-car length the whole time. This is the same thing I experienced the couple other times I ran against other GTIs with different but comparable tuning brands at stages 1 and 2. To think that I'm producing 15% more boost or 12% less horsepower at any point in the band doesn't seem right to me, empirically.
Anyway, it would be *really* great to add columns for dyno-confirmed peak HP and torque, dyno type, octane/program, and ideally dyno run location/altitude (or a data table to draw averages from for (hopefully) multiple observations if google docs can handle it). Almost all boost meters, dynos, or any other gauges of a complex measure are going to suffer in some way for accuracy. The law of large numbers can account for a lack of accuracy, but it takes a lot of data for it to start to work.
So, I vow to get my car on a dyno as soon as possible, and, if feasible, I'll run VAG-COM the whole time to get boost and other interesting readings. A bunch of people have already posted their dyno graphs on here and elsewhere.
I work with a lot of numeric data for a living, so if you want me to compile it, send what you have my way and I'll happily build the database. Please join me in this noble pursuit. Let's aggregate some data and speak truth to power.