WGA problem

[NoStop]

Yeah, looks like I took a wrong turn there somewhere. Thanks for
straightening me out on this.
However, my intention was only to try to quantify the extent of the
"flaw" in WGA that Alias had mentioned. Calculations based on my
erroneous assumption that 1 in 5 WGA tests on valid OSs failed gave
rise to the near impossibility of getting through 50 successive tests
without a failure. This, indeed, very strongly suggested that this
assumption MUST be wrong.
Given that there is no data available regarding the relative number of
tests on pirated OSs in comparison to tests on valid OSs, it is not
possible from this data to calculate the incidence of WGA failure on
valid OSs.
Nevertheless, given that I had gone through 50 successive tests without
a failure, we can reasonably assume that the probability of this must
be high. Let's conservatively assume 0.95 (or 95%) in order to give an
order of magnitude, or ballpark, estimate.
Based on this assumption, we can calculate that the probability of a
valid OS getting a positive result from a one-off WGA test is 0.999
(0.95 to the power of one over 50). This means there is only one chance
in 1000 of a valid OS being failed.
Now, certainly it will be extremely inconvenient for the one in a
thousand OSs that fail the test, as I have found myself.
Nevertheless, a failure rate of one in 1000 doesn't seem to me to
deserve being termed "flawed".
Paul

Unless there are 250 million computers that need to pass the test. Then a
quarter million would end up being "flawed" and a quarter of a million
pissed off customers is a marketing nightmare. Even for MickeyMouse that
doesn't give much of shit about their customers.


--
WGA is the best thing that has happened for Linux in a while.

The ULTIMATE Windoze Fanboy:

http://video.google.com/videoplay?docid=-2370205018226686613

Is this a modern day equivalent of a Nazi youth rally?:

http://www.ntk.net/media/developers.mpg

A 3D Linux Desktop (video) ...

View Some Common Linux Desktops ...
http://shots.osdir.com/

 

[paulmd]

Yeah, looks like I took a wrong turn there somewhere. Thanks for
straightening me out on this.

The probibllity funticons you're using are off a bit.

Take the coin toss. 50% chance of getting heads.

What's the probability of two heads in succession? 0.5*0.5 =0.25. = 1
in 4.
And what's the probablity of 50 heads in a row? 0.5^50. =8.88x10^-16 =
1 in 2^50

let's say the WGA test is 99.99% accurate. The probiblitiy of 50
vailitations would therefore be about 0.995 or about 199/200. 1 in 200
gets screwed.

This does'nt sound too bad on the face, but it translates to tens(maybe
hundreds) of thousands of microsoft's customers getting screwed.

Further consider that WGA, until recently talked home DAILY. This means
that over the course of a year, there'd be 365 chances to go wrong.
Let's revisit the failure rate.

0.9999^365 =0.964. About 19/20. Then consider this will screw 10 times
the number of people as before.

Over the course of 5 years, which is about how long you could expect to
keep your computer. The probability of esaping unscathed is 0.833. 1 in
6 are screwed. Still like those odds?

Dropping the frequency of tests to once every 14 days, as MS now claims
it has done helps. But there's still 130 chances to go wrong. =0.987. 1
in 75 customers screwed.

This transates to a flawed system.