I am all with you concerning the reliability problems of "baby name" websites. And for me there is also a difference between tiny and small, in fact a whole order of magnitude: something small can easily be 10 times as big as something tiny.
But my view concerning fabrication is different. I certainly saw much of it in the last 5 years of intensive Googling while building my name database from scratch, but often "plagiarisation" is an even bigger problem.An
example that I recently encountered: What does the name Ayanna mean? Baby name websites mostly tell "beautiful flower" in Swahili. Well, this is a likely candidate for a fabrication, because people knowing something about Swahili or even speaking it know nothing about such "beautiful flowers":http://research.yale.edu/swahili/dc/dcboard.php?az=show_topics&forum=2&page=3
But look just how many websites tell this probable nonsense? There must be dozens of them! If fabrication is really widespread I probably should see a wide choice of different purported meanings for a name like Ayanna whose meaning is probably unknown and thus attracts "fabricators".
But instead widespread copying of information seems to take place. 1 fabrication once at an unknown place and time in the past, but dozens of copies that nearly drown interesting information about the name.
And why not? Let's suppose I want to make a quick dollar on the Internet and see that baby names are all the rage. I hope to make a small fortune with advertising and products for parents like many such sites seem to. I thus need a name database pronto. Now do I start with a name list and fabricate meanings where none are readily available? Hell no, I just copy the names somewhere, and shortly afterwards my site is running...
Maybe it even overestimates the capabilities of some of these people assuming them to be able to decently fabricate.