"The astronomical community did not believe we would ever really make the data public," says Mr. Szalay. The typical practice in the mid-1990s was to guard data because it was so difficult to get telescope time, and scholars did not want to get scooped on an analysis of something they gathered.
One incident demonstrates the mood at the time. A young astronomer saw a data set in a published journal and wanted to reanalyze it, so he asked his colleague for the numbers. The scholar who published the paper refused, so the junior scholar took the published scatterplot, guessed the numbers, and published his own analysis. The original scholar was so upset that he called for the second journal to retract the young scholar's paper.
Mr. Szalay said that astronomers changed their minds once the first big data sets hit the Web, starting with some images from NASA, followed by the official release of the first Sloan survey results in 2000.
I was surprised by that anecdote, but then I only started working in physics in '97. I recall though converting one or the other figure into a table to be able to reuse the data - an extremely annoying procedure, even with the use of suitable software. However, these were figures from decade-old textbooks, the data of which I needed to check whether a code I had written would make a sufficiently good fit. And 5 years back or so, when I had a phase of sudden interest in neutrino physics, I noticed that while one finds plenty of papers on the results of Monte-Carlo simulations to fit neutrino experiments, the data used is not for all experiments listed. In one case, I ended up browsing a bulk of Japanese PhD thesis (luckily in English) till I found the tables in the appendix of one, and then I had to type them off. Not sure how much the situation in that area has changed since. But change is inevitably on its way...