I've said this before and I will say it one more time: Do we want to keep the piss-poor definition for automatic article detection?
I propose (again) that the current definition be used for "entries" and a more stringent definition be used for "probable articles" (everything, of course, is still a page).
We can simply take the count for entries and exclude anything that is less than 500 bytes and has a link on it to [[Wikipedia:Disambiguation]] (or is listed on one of links of disambiguating pages pages). That would give us about 80,000 probable articles in the English Wikipedia (yes that still includes about 30,000 rambot articles but so what? They are far more useful as /articles/ than are many other entries we call articles).
The current prediction is that we should hit the 100,000 mark the middle of next week. Press release or no press release, that number on the Main Page will get some attention (thus the need for a press release to explain things).
-- Daniel Mayer (aka mav)