Who is responsible for detecting plagiarism in a PhD thesis?
The problem is that the PhD system is designed for people who intend to become researchers. For these cases, plagiarism is not at all a common problem. You are expected to published your research, and you will not have a successful career unless it is widely read and cited. That gives lots of opportunities to get caught, and the penalties for plagiarism are a huge deterrent.
To the extent you find plagiarism, it's generally people who do not want a research career, but instead view the PhD simply as an obstacle on the road to a teaching (or other) career. Probably the community should scrutinize these sorts of theses more carefully, but it can be hard to work up the energy to do so when most of them are OK, and when these theses really don't matter much for the research world.
The German politicians are pretty much the worst case scenario. In the US, the stereotypical case is educational administrators. Typically, you have a distinguished person who starts to feel the need for a PhD. Perhaps it's because they associate with academics and feel looked down upon, or perhaps it's because an academic endorsement would make the public value their expertise more. This student is very smart and accomplished, and nobody suspects them of any dishonesty. However, they are also very busy, often working on a PhD while pursuing other projects as well, and academic research is not a priority. At some point, they succumb to pressure and start taking shortcuts. Probably it starts with small things, but the shortcuts gradually grow larger. They rationalize that the thesis doesn't really matter anyway, because they have no intention of following an academic career track. After all, they have the knowledge and experience, and they deserve the PhD title, so what difference does one document make anyway? Meanwhile, the advisor probably doesn't spend that much time working with the student, and has no reason to suspect anything. The advisor really ought to be extra careful in cases like this, but that would seem like an insult to the student, so it's easiest just to trust them.
So my take on this is that plagiarism is not as widespread as news stories might suggest. It's just particularly likely to happen in cases where it would attract media attention.
Given the complexity of the modern PhD thesis, and the number of references to other works in the literature that such a work would normally contain, being able to catch plagiarism can be difficult. This is especially true when you only have a printed version of the thesis to work from. Moreover, the time to review a thesis is normally quite short—a few weeks to a month at most. Given the number of other responsibilities most faculty members have, it's unlikely that they're actually able to verify every fact and every citation, let alone check for evidence of plagiarism!
On a general level, it is also assumed that everybody plays by the rules, and that therefore plagiarism shouldn't be likely in a PhD thesis. Only if suspicions somehow get raised do people take a second look. It's probably an invalid assumption, but it keeps the system moving. Unless we want to move to a QA-like scheme in which every claim and source has to be verified against the original, I'm not sure what other options there are.
Plagarism in journal publications is a large concern because the professors are often running their own labs, doing their own research, with very little oversight. The only chance to catch fraud is through a careful analysis of their publications, which is done both by the publications themselves, their peers, and the public at large.
Regarding PhD theses, the student is working with an advisor, doing (hopefully) original research. Any actual publications the student submits will go through the peer-review process, which will hopefully find any plagiarized references. Regarding the dissertation itself, it's often simply cut-and-paste from their actual publications, and once submitted, never read again.
It's likely for this reason that people probably don't dedicate much effort to finding plagiarized works in their thesis. The student is doing original work, as verified by their advisor and their committee, and any publications by the student have been peer-reviewed. No one cares about their thesis, and if they try to continue their behavior as they progress in their career then they'll simply get caught when it actually matters (i.e., during the peer-review process) in the future.