[notacoward]

Despite the fact that it was mentioned in the paper, developers continued to deny that it was a serious problem or claim that it had been solved since. Periodic re-insertion and date-based redirects were both touted as answers.

As for "can't" that's not accurate. You can certainly prevent old data from being pushed out, by returning an error on insertion of new data if there's no free space. That's how many other storage systems (e.g. every filesystem ever) work. While it's true that you can't fully protect against a reduction of capacity when nodes go offline, that's a very different issue. Freenet being cache-like rather than storage-like was a decision of convenience, not a technical necessity.

[renegadus]

Filesystems don't rely on space voluntarily donated by others, and that's the critical difference here.

Freenet is designed for sharing information, not archiving it. You can't have both.

[notacoward]

Other systems have also relied on space voluntarily donated by others, and still made the choice to behave as storage systems instead of messaging systems with some history. Or caches that just happen to turn over slowly because the system's too slow and unusable to create cache pressure. If Freenet ever had really taken off, you would have had to deal with these issues. Sharing information doesn't work when the information falls out before the recipient can get it.