It seems like part of the motivation here is that it's possible to run many contemporary and planned very-big-science projects at all, since nobody's going to b

It seems like part of the motivation here is that it's possible to run many contemporary and planned very-big-science projects at all, since nobody's going to be sitting around analyzing centuries' worth of unvetted sensor data, but there are plenty of people prepared to spend a few years of their lives checking and massaging the regions of interest marked by a computer cluster. The simplest cases will have been long enabled by simpler regressions and such, of course, but the more complex pattern recognition appears to be appreciated.