A bit of twiddling with a ruler on a log-linear graph of the article count vs. time appears to show a "natural" exponential growth rate of between:
"Optimistic", based on May 2001 - May 2002, ignoring the Great Slowdown, and also fitting the most recent slope:
exp(0.18 * months)
"Pessimistic", fitting May 2001 - present including the Great Slowdown:
exp(0.125 * months)
This produces these extrapolations, assuming exponential growth.
Month, optimistic, pessimistic Sep 2002, 45000, 45000 ) 50,000 is reached sometime Oct 2002, 54000, 51000 ) in these two months Nov 2002, 64000, 58000 Dec 2002, 77000, 65000
Jan 2003, 92000, 74000 Feb 2003, 111000, 84000 <--- optimistic 100,000 Mar 2003, 133000, 95000 Apr 2003, 159000, 108000 <--- pessimistic 100,000 May 2003, 190000, 122000 Jun 2003, 227000, 139000 Jul 2003, 272000, 157000 Aug 2003, 326000, 178000 Sep 2003, 390000, 202000 Oct 2003, 467000, 229000 Nov 2003, 559000, 259000 <--- optimistic 500,000 Dec 2003, 670000, 293000
Jan 2004, 802000, 333000 Feb 2004, 960000, 377000 Mar 2004, 1149000, 427000 <--- optimistic 1,000,000 Apr 2004, 1376000, 484000 May 2004, 1647000, 548000 <--- pessimistic 500,000 Jun 2004, 1972000, 621000 Jul 2004, 2361000, 704000 Aug 2004, 2826000, 798000 Sep 2004, 3383000, 904000 Oct 2004, 4051000, 1024000 <--- pessimistic 1,000,000 Nov 2004, 4850000, 1161000 Dec 2004, 5806000, 1315000