We had a party at my house this summer to celebrate/commiserate. What became AFNI (originally called FD3) came out of a meeting at Froedtert/MCW in late June 1994, at which several researchers there demanded the ability to transform FMRI results to Talairach space. One thing led to another, and soon (within a week or so) I had the idea of creating a tool to allow surfing FMRI analysis results in 5 "dimensions": space + scan + subject. Several months of furious programming ensued, and the first version was demo-ed to the complaining investigators (they know who they are) in September 1994. By November 1994, I was sick of AFNI, having written 30,000+ lines of code since early July, and decided to stop work on it. As you can see, that resolve didn't last long.
I suppose I should come out with a boxed DVD set called "AFNI X", with various commentary tracks by people who've used and contributed to AFNI over the years.