That is ignoring that a sufficiently motivated actor can ensure that doesn't happen. In one of my private trackers there is a person with a seedbox that downlo

That is ignoring that a sufficiently motivated actor can ensure that doesn't happen. In one of my private trackers there is a person with a seedbox that downloads every single torrent as soon as it is uploaded, and they have been doing so for quite a few years now. This ensures that while some things will indeed, be seeded more, nothing quite vanishes. Then again the form media of that specific tracker is fairly small, so it is not prohibitively expensive to archive everything. One raw ISO Blueray movie file elsewhere could be thousands upon thousands of torrents in that specific tracker. Maybe something of utility would be creating a distributed torrent system that is a bit more closely tied to the tracker. Where membership would require you to integrate to the swarm by automatically downloading a percentage of the entire corpus, ensuring the health of the tracker. So a new peer would be bearing part of the load of having everything be accessible. I think this would require decently heavy curation, but I could see how it could be useful for something like the OP specifically, where having scientific papers lost for good would be a shame.