But I do think we should discuss it... is it better to have 1000 stubs or 100 long well-written articles?
I think the key point here is when you want the encyclopedia to be good. If you want it to be good now, then 100 long articles is best. If you want it to be good in a couple of years, then 1000 stubs (assuming they are good stubs, but that's another debate) is better.
This is based on the assumption than stubs will eventually become articles (and faster than non-existent pages do). This would suggest that younger wikis will have more stubs. To test this hypotheses, I copy and pasted the table linked to in the parent email into excel and plotted edits (well, log of edits, actually) against stub ratio (I'm using edits as a measure of age, seems better than time since creation). The result surprised me. For wikis with less than 20000 edits there is a very wide range of stub ratios, but there is an overall (weak) upward trend, as I was expecting. At 20,000 edits there is a very sharp cutoff and the range of ratios becomes much smaller (from about 0.5 before 20,000 to 0.2 after, excluding a few outliers in both categories, which include the English wikipedia) and there is no trend visible at all.
Any guesses on what happens to wikis at the 20,000 edits mark? It's an amazingly sharp cutoff.