One problem with the second source stuff is that
it's more difficult to get
documentation for it. So archiving documentation for very rare or probably
non-existent hardware should be given some priority over the stuff that
everybody has (speaking of redundant piles of pdp11 and pdp8 documentation).
That is exactly the reason, why I take time to scan field engineering documention amongst
others about Nuclear Data systems, Unisys System-80, IBM S/370 and Honeywell
Datanet-systems, even though I don't have the hardware. When I'm uncertain about
the gain for bitsavers, I just ask Al Kossow about its relevance or if there is any
interest for it.
Especially hardware produced in little amounts or for very special purposes without any
documentation (and software) is pretty useless, unless you have time and the ability to
re-engineer its function.
Pierre