-=> My computer told me that Rick Collins said to Richard St. John
 RC> And it tells them _only_ the results obtained under the conditions of
 RC> the test - extrapolating that to what the individual _user_ might
 RC> expect in _his_ environment is the problem.
In the original series of posts, as with all the comparisons that Virus
Bulletin has done the EXACT testing procedure as well as equipment is
listed for people to duplicate if needed.
 
 RC> And did the results differ from the "published" results you've spoken
 RC> of?  I'd suggest they did, and that alone proves my point.
Many did, many did not {some were right on the money, even with our more
extensive testing}, but our test bed was entirely different than the
one they published so naturaly our answers should be slightly different
especially since we shot much more at the scanner, so our percentages
were definitely different.
To describe our method we used multiple configurations of PC's everything
from Pentium Pro with 64MBytes of memory to 286's with 1 MB memory,
DOS 5.0 through 6.22, Windows 2.15 through 3.11 and NT 3.51 through 4.0.
We also shot 100,000 instances of various viruses at the scanners
{Almost took us longer to infect the server than it did to run the tests.
The servers were everything from 3.11 servers to 4.1.
 
 RC> Yes, but do they infer from that what the best product would be in
 RC> _their_ environment?  And how, pray tell, do they analyse _their_
 RC> environment to determine how close it is to the published test
 RC> conditions?
Their answers are in percentages and are just numbers. how an individual
infers these answers is that persons way of doing it. In _MY_ system
environment their testbed is rather pathetic as I administrate 750+
Novell servers 500+ NT servers and 40,000+ PC's world wide. Hell I have
more in my office than their entire test bed, but knowing what they used
and how their testing worked helped.
We, like anyone that reads the comparison {BTW their July one is done},
made our own determination as to our needs. Their testing was just a
basic guideline.
RS
--- GEcho 1.20/Pro
---------------