Compiled Trees: Sources as a Filter

For some families and individuals, there are hundreds or thousands of online "trees." (An earlier post put out the thought prompt that should these trees be part of an "exhaustive search.") If one decides to use the trees as clues (and never as established fact), what are ways to sift out the reasonable from the not-so-reasonable.

This sifting is not an exact science. We are not mixing chemicals where there's pretty much a known reaction and response. This is more like surgery--sometimes you don't know what you are getting into until "you open things up."

If I'm using compiled trees for clues, I tend to loosely put the trees into three categories:

• no sources
• files as sources (stuff.GED, stuff2.FTW, stuff3,RTM, etc.)
• sources

The first two are frequently put aside but unsourced trees can be accurate. It just may be that compiler didn't want to share their sources. Of course, "sources" can vary widely as well as the research and analytical skills of the compiler. If I have many to sift through, I tend to initially concentrate on those compilations that include more material of a primary nature--courthouse records, vital records, etc. But it is an inexact science to be certain.

I'm interested in hearing thoughts on how others sift. 

And, as always, online compilations are best used as clues and suggestions, not as Gospel.