Take the original Boyer-Moore paper, for instance. It seems every time I look for an algorithm the paper I see referenced is talking about implementation idiosyncrasies on a PDP-10 or PDP-11.
I know these were popular machines at one time, but not so popular as to account for the large number of references I see. Is there a specific reason these machines were used as examples in algorithms papers?
but not so popular as to account for the large number of references I see
I strongly disagree. The PDP-11 series was far cheaper than other midrange computers. Some of the cheapest models, such as the PDP-11/34 (started at about $10k) could be purchased on a research grant with money left over for the investigators. This allowed researchers to acquire their own computer during an era when commercially sold "serious" computers tended to start well over $100k. Universities were able to get copies of Unix for free back then, instead of having to pay an additional large sum of money for the operating system to go along with the hardware.
Back then, many companies like IBM were renting their hardware and software. The monthly charges would quickly devour any researchers' grants, and most were too expensive that if the university had one, spare time would be rented out to paying customers.
The PDP-11 had a very regular instruction set, which made it much easier to write code for than most of its competitors. That also made prototype implementations easier to understand. DEC's FORTRAN compilers were also very good, giving the best "bang for the buck" (performance-wise) as it were.
PDPs 10 and 11 (expecially 11/3x, colloquially known as the half past noon series) were not the only one, but they were at the time (and those were the times when you didn't usually find a computer, hell, a calculator in every building), I won't say popular, but not so rare as before.
They were relatively cost efficient, and research institutions and engineering colleges started acquiring them in computing centres for their work. In East Germany a similar (don't know how much similar, though) machine was manufactured. Programs were written for them (I still have dozens of numerical books with programs written for PDP11) and people, naturally, found no reason to not continue on what was already there.
The people thinking up new algorithms and writing about them were in environments where those were the typically available machines. There were other machines around (IBM 370, PDP-9, Burroughs, Univac, Honeywell, Xerox), but DEC had figured out how to sell to those environments where people thought most about algorithms, AI, etc.
This was smart because those environments (universities, research labs) would generate a flow of people into industry who had brand loyalty to DEC.
PDP-11s and PDP-10s were used by the leading CS departments of the 1970's: MIT, Stanford, Carnegie-Mellon, Columbia, and others. Also UNIX and C were originally written for the PDP-11. DEC's operating systems were both simpler and more powerful than offerings from other manufacturers of the time.
Before the PC era, PDP10s were the most cost effective computers to give students access.
