Both hippo and raiph are right. In 2000, Perl was at its zenith. But other languages were catching up, rapidly, and even though Perl was growing still in both userbase and features, other languages grew faster. And Perl was suffering badly from flamewars, lots of flamewars, hundreds of messages after each other about what now seems trivial things. I know quite some people from that time that were driven away from Perl because of the hatred and nastiness among Perl-developers at that time, and I was nearly among those. Since less than a decade, I meet some of them again, and they are quite happy with the many changes of Perl (both 5 and 6), most importantly the lack of flamewars.

The arrival of Modern Perl (both the style and the book) made a big difference too. As did the downfall of obfuscated code, the continuation of Perl workshops and conferences, and at some point the first QA Hackathon. Books like Perl Best Practices helped too (most people understood that the chapters were not to be followed religiously, but to provoke you to make a well-discussed decision).

The brain drain from Perl 5 to Perl 6 was minor, I think. Just one or two handfuls of people spent a lot of time on Perl 6. Later on, people from outside the Perl-community came in, like Jonathan Worthington.