Or, in the case of software tools in education, always a step behind, but for a good reason. David Wolever recently wrote about how SVK has made his life better; he thinks it's less painful to use than Subversion, and has hinted more than once that DrProject ought to switch. I'm willing to agree with the first, but not (yet) the second---not until there are tutorials that have had their tires kicked, a couple of books, plugins for Eclipse and other IDEs, desktop GUI integration, and everything else that comprises a mature, student-accessible software ecosystem.
I'm trying to teach good practices, which really means that I'm trying to persuade people to try something new based on my promise that it will make them more productive. SVK's advantages aren't apparent to beginners, who are really using version control for backup and as an alternative to emailing files to one another, so the "other stuff" that SVN offers makes it a winner. I feel the same way about SCons and Rake vs. Ant or Make, or JSON vs. XML: the newer tools would win if all other things were equal, but they're not, so they don't.
I sometimes hear people argue that if someone isn't smart enough to figure it out on their own, they shouldn't be programming anyway, but I don't buy it. The productivity gains of newer tools are uncertain, but the costs of working with incomplete documentation and flaky integration are not, so the intelligent thing to do is actually to be conservative and let someone with a higher tolerance for pain (more charitably, someone who measures productivity in a different way) thrash their way through the undergrowth.