It's a full-time job to keep up with the industry, let alone change it with research. I think a little lag is OK so long as it doesn't teach BAD practice (e.g. thank god academia has moved from CVS to git).

Why do you think that position to be extreme? (serious question)

Because many of the topics covered in a typical CS curriculum have direct applicability in industry. For instance, many courses in the area of software engineering make it a goal to try to simulate the "real life" conditions of a software project (e.g. working in a team, having stakeholders with different needs). So it would be hypocritical to then claim to not pay attention to what industry says it needs.

Of course, this should not mean that whatever industry asks for is law - sometimes they simply get it wrong (it is usually not very difficult to realize when), which is exactly the point where universities can help by not educating students to repeat the same mistakes. Something like saving companies from themselves, I guess.