"They certainly aren't valid, documented sources"
--anything from which one takes whatever sort of data is a source and should be cited as such. Evaluation of quality of information is a quite different matter.
"...and should be (and usually are) ignored by most serious researchers."
--There are actually some sterling trees in which cited documentation for nearly everything is present. There are also many trees for which the owner has extensive documentation but has not supplied it on the tree (but may share when contacted with specific questions). For serious researchers, it can be helpful to find a collaborator who has posted a "cousin bait" tree.