Strategy for cleaning old IQ files

My script (discussed in Uploading compressed IQ files to Dropbox (or any other cloud storage provider) ) has now uploaded 1.64 TB IQ files (42,716 observations), which means I will run out of space on my 2TB Dropbox account in a few months.

The easiest strategy that makes sense to me is to delete the oldest observations, but any other suggestions are welcome. I would prefer to keep the files that are most valuable to the community, but which files would that be?