Oh my goodness the HST Archive search, access, and retrieval tools are very bad if you want data on thousands of pointings! Phil and I beat our heads against this all day. But we do have all the datasets in which we are interested identified and tomorrow we hope to get them all downloaded.
What's wrong with the following numerology? HST takes an average of about 60 Mb of data per orbit, and orbits every 90 mins for 20 years. That's 1 Gb per day, or a lifetime total of about 7 Tb. We can spin this all at NYU, and give the community better (faster, more survey-oriented) access tools to boot.
Ted Baltz gave a wonderfully comprehensive talk about direct detection of dark matter.