> Because? What have you seen of those tests that makes you
> distrust them?
> And can someone post a link?

http://blog.untangle.com/?p=96 describes the test.
http://www.darkreading.com/document.asp?doc_id=131246&WT.svl=news1_5 is Dark
Reading's take, though it seems a remarkably sloppy piece of journalism.

I'm actually working on an article around this topic at the moment, and I'll
post a link here when it's up. So I don't want to duplicate too much here.
But here are the danger signs:

1) Tiny test set, of which at least one (probably more) samples was/weren't
viruses at all (EICAR test file).
2) No indication of any sample validation, particularly of those samples
supplied by the audience. No information as to the source of the "wild"
samples. No consideration whatsoever of the possibility of false positives.
3) Inconsistencies with other tests on same samples from competent source
4) Gateway and desktop products tested altogether
5) (Mis)configuration issues, certainly with Sophos and probably with
Watchguard. No obvious attempt to level playing field.
6) Just the fact that the results are (1) so widely divergent within this
test (2) so different to established, reputable tests - VB's test in April,
for instance. Maybe everyone else has got it wrong and Untangle have it
right, but given the other indicators, I doubt it.

--
David Harley
http://www.smallblue-greenworld.co.uk

[ reply ]