Fully centralized design is in fact largely obsolete. Centralized operations are still very relevant and important, but I doubt we will ever see a new version control system designed without distributed facilities.

"A few use cases" and "for one reason or another" both do sound dismissive, I agree.

Agreed, the main benefits DCVS bought is the ability to work without a constant internet connection and efficient branching/merging.

For the rest most companies are using git in a way very similar to svn, but now Github/Gitlab/Bitkeeper whatever provides the role of a single authorative central server.

Your first sentence is hard to parse, but it seems like you are saying (1) DCVS has more efficient branching than git, and (2) implying git requires a constant internet connection. In which case, (1) is dubious -- got benchmarks? and (2) is simply incorrect.

Is DCVS not just a typo of DVCS, ie Distributed Version Control System, of which git is one?

correct

Those statements are mostly true. One of those few use cases where centralized version control still has a lot traction is game dev. Because of large files and a need to manage them efficiently on relatively limited local disks.

You also have to ask the question if the "distributed" functionality is really necessary. Even if companies use git or mercurial or whatever, most have a central repository anyway. They basically offer no extra value over Subversion on that specific (but most prominent) feature in practice.

A lot of the young guys I work with don't even really know about the distributed characteristics. For them git is Bitbucket or Github. that's all they know.

They'll learn. Things like cheap branching didn't become useful to me until I had a bunch of stuff going on at once, which didn't happen until I graduated to more senior roles.

One great side effect of distributed is that even if your use case is generally centralized, you de facto have a lot of valid backup states available if something in the central repo gets borked in some destructive way.

FreeBSD uses it for the operating system. While he complains that svn is obsolete, he then goes on to say it's widely used, and so far, it seems pretty important when games and operating systems--and more--are using it.

obsolete ≠ unimportant. I think most would agree that COBOL is an obsolete programming language, yet half the worlds financial systems still run on it.

How many percentage of new projects use svn instead of git? That it a much better metric for how obsolete something is.

Obsolete means to no longer be produced or used... These things are not obsolete.

They aren't cool / widely used but definitely not obsolete if half the worlds financial systems still run on them.

From merriam webster:

    1a : no longer in use or no longer useful
         // an obsolete word
     b : of a kind or style no longer current : old-fashioned                 
         // an obsolete technology

A good example is a satellite, lots of older (obsolete model) satellites are still in orbit and performing well, and replacing them is not worth the cost. But if a new satellite is needed the newest model will be send up, not the obsolete model.

iirc FreeBSD is sticking with svn because git is GNU licensed

Agreed. The book not only needs to be fleshed out, but is severely lacking in quality.