This is interesting because I think we do want an authoritative document store and that, yes, we hence need to pay for its upkeep. So if he had mirrored and hosted all of these cases, they would've been merely snapshots of past history, not the curated corpus that PACER has.
The same could be said of scientific papers where large retractions are handled by the journals, but may be lost by some mirrors.
Information quality, provenance and current validity is more important than the trope of 'wanting to be free.' Once information passes into the 'historical' realm, perhaps it should/must be free, but when we are in the malleable phase it's irresponsible to 'mirror once' without knowing how to get pushed (or pull) updates.
Look at how the Linux kernel mirror system works, push mirroring, etc. The scrape method doesn't pass the smell test if you really want to provide a service beyond point in time archiving (aka archive.org).
Regarding depression, suicide and unfair persecution I'll withhold comment.