http://piratebayztemzmv.onion/top/all https://youtube.com/ https://en.wikipedia.org/wiki/List_of_most-viewed_YouTube_videos https://en.wikipedia.org/wiki/Distributed_hash_table https://en.wikipedia.org/wiki/Exabyte https://en.wikipedia.org/wiki/Zettabyte https://en.wikipedia.org/wiki/Zettabyte_Era https://www.youtube.com/watch?v=iIKPjOuwqHo A new globally distributed P2P autonomously self maintaining deduplicating plug and play slack space object storage network could easily surpass 100EB... "slack space" - network utilizes say 80% of free space on disk for itself, maintains that by progressively removing its own extents upon sensing storage pressure encroaching the remaining 20% (from users other apps), adds more extents if users space later decreases. Some data expiry models exist as natural byproduct in a number of types of storage networks... - Equal resource sharing / trading, pay based... prioritizes the equal player, highest payer, etc. - Curation, review, governance... follows the whim bias demand and censorship of those actors / crowd - Seed based networks... survive until the last seed. These types also don't support insert and disconnect operations very well... indigent artists, first leaks, bulk publishers, etc. Prior attempts at distributed self maintaining object storage networks may have assumed that demand fallof is the only valid metric to expiry. http://piratebayztemzmv.onion/top/48hall In absense of the three above, short term flash demand is news, thus always valid. Long term, things start looking different, unpopular and or old hashes often begin to hold term value and importance. Models for a network open to the masses, that happens to prune historical top N hashes at each hashes respective M year accounting point, may be unlikely to lose anything of importance perhaps due to eventual popular reuploading from other sources. Designers of new nets may want to consider priortizing any space reclamation needs based on some analysis of real world "popular pop trash" content falloff patterns, some physical library, museum, and attic holdings models, inverse falloff metrics, internally guaranteed levels of object redundancy, etc. Are any current networks autonomous, free from three models above?