> On Sun, 30 Jul 1995, Mike Meyer wrote:
> > (*) Anneli was absolutely fuming at NetScape, with comments on "making
> > fortunes helping millions of clueless people do the wrong thing." A
> > magazine she likes had recently bragged about their web pages. Turns
> > out they aren't navigable in anything but NetScape because of
> > NetScapes broken attribute parsing. The response from the magazine was
> > that 25% (their estimate) of their market wasn't worth spending a few
> > minutes fixing the problem, or even doing it right next time.
> I have my doubts about this statistic ("75% of all browsers are Netscape").
> Where did this figure come from ? Did it come from Netscape, I wonder ?
The 75% figure comes from many independant sources. I have confirmed it
myself by collecting stats on a half dozen different servers with daily hit
rates anywhere up to 1/4 million hits per day. I get numbers in the low
70% range day after day - even after correcting for the undercounting
of Lynx. Lynx gets undercounted in most reports by anywhere from 2 1/2
to 5 times, depending of the mix of graphics and text on a system. But it
has such a tiny proportion to start with that the correction is only on
the order of 5-10%.
If you want a general source for several such collections of stats:
While it might have a certain appeal to claim that the 75% figure is
somehow just a publicity claim - it isn't. It is fairly close to the real
If you are interested in collecting your own user agent stats for
has a perl script that can do it automatically for a NCSA 1.4.x server.
-- Benjamin Franz