My point is that people at such position can best serve science, if they funnel some money into actual research. There are many researchers, but qualified administrators who actually cares, are very rare.
My experience is that it's not necessary the case. Even when such jobs exist ("low-salary" permanent research positions that may exist in some european countries for instance), a lot of people still prefer to take a professor position at some point.
Also, passed a certain age, some researchers find themselves less productive research-wise and want to do other things.
Hmm, I'd be interested in learning more; sounds like the kind of position I want. :-) Around here (Nordic region) I don't know of any such positions though. There are teaching-only positions, which can be a good option if you like teaching but not research: no grant or publication pressure, just a stable job with reasonable work hours. And there are regular faculty positions, with the usual teaching/service/research/management mix. But I don't know of an option to go research-only, at least in CS. In some areas you could work for a state agency as permanent research staff, but I don't know of any that do CS research (civil-service research positions do exist in areas like historical archives, social science, energy policy, international relations, healthcare policy, etc.).
edit: I did think of one place with such positions, the French state technical research institute INRIA. Maybe I should learn French...
There are stable research positions, but they tend to actually be high-prestige and hard to get, like faculty member or permanent fellow at a research institute (Institute for Advanced Study and the like). In some fields you may also be able to find research positions in industry, or at a national laboratory, and those options are pretty popular, where they exist. A really well-funded lab that can string together enough grants to insulate the research scientists from the money flow and provide them de facto permanent jobs can also be such a place, e.g. the heyday of the MIT AI Lab, which had many non-faculty researchers. Also a popular option when it exists, but fairly rare in CS (more common in areas like medicine).