Tsk, tsk-- you'd think Microsoft would have learned something from all those "creatively enhanced" test results they wheeled in for the "Redmond Justice" trial. Remember the faked videotapes showing how Windows 98 supposedly runs slower when Internet Explorer is removed, and the hilarity that ensued when government mouthpiece David Boies spotted the proof that the before and after results involved two different computers? Remember later in the same trial when Microsoft delivered test results showing Internet access speed comparisons between two operating systems and neglected to mention that one computer had a much faster modem? Check it out-- some folks just never learn.

The latest bit of skullduggery to issue forth from Redmond comes to us courtesy of faithful viewer Jerry O'Neil, in the form of an article in The Register. It seems that Microsoft got a bit overzealous in its marketing efforts for the forthcoming Windows 2000. Evidently the company feels that the much-delayed successor to Windows NT 4.0 needs a bit of a hard sell to pump up the sales numbers, especially after such a long wait. That's why they're claiming that, according to a ZD Labs speed comparison, Win2K "outperforms" NT 4.0 by "up to 24 percent" and is therefore "significantly faster." There's just one problem: the ZD Labs report apparently doesn't say that at all.

In fact, what ZD Labs found was that on systems with 64 MB of RAM (the vast majority of workstations would have at least that much, we imagine), Windows NT 4.0 was actually faster than Win2K. And the speed gap actually widened when the same tests were conducted on systems with 128 MB of RAM: "Windows 2000 was three per cent slower than Windows NT 4.0." Sadly, ZD Labs didn't test higher-RAM systems, which are very likely to be used in real-world scenarios, but if you'd care to extrapolate from those two data points, it sounds like NT 4.0's performance advantage grows as more RAM is added. And why bother with a 32 MB test at all, when Microsoft recommends a minimum of 64 MB for Win2K, unless it's just to provide at least one scenario in which Win2K is actually faster, albeit marginally so? Then there's the little issue of NT 4.0 being deliberately crippled in the tests by using a slower file transfer mode. The part that kills us, though, is that even with every advantage stacked towards Win2K, the test results were so lackluster that Microsoft still had to spin them for the marketing hype.