> Bitrot does set in over time, though, and the less distributed a work is the worse its odds of living through each new tech transition.
Eh, I feel like we've almost fully gotten away from bit-rot. Bit-rot was a problem because individuals don't bother to go to the effort of constant IT administration, and don't have the economies of scale required to afford good data redundancy.
But individuals don't have to back things up "personally" any more. With one command, I can send three copies of any file to three object-storage providers. Each one (according to the Dynamo architecture, which they mostly all implement) is holding 17 copies of the data on 17 shards, and replaces copies from the good ones whenever they go bad. Each provider's copies only exist in one region, but the region is different for each provider selected. As individual object-storage providers die, I can find new ones and sync my data over from the surviving ones. As long as "cloud [redundant] object storage" as a concept exists, I won't have to do much at all to ensure continued integrity of my data. All the details—including porting my data to new physical substrates when old formats die—are being handled under the abstraction, and converted into (tiny!) monthly fees that are lower than what even a zero-redundancy tape library would cost me.
And, of course, the data I'm sending is encrypted, and not even with a symmetric key, but rather a PKI key where the decryption half of the key is held in cold storage (i.e. in hardcopy base64 in a safety deposit box at a bank; and a few other places.) If I'm uploading rips of something copyrighted, nobody will ever know that but me. I'll be able to get the data back out when the time comes to dust it off and share it.
In the HD movie torrent community – people sharing full Blu-ray images – a person’s collection can run into the many terabytes. My own collection amounts to 8TB and is only content I have watched personally and am interested in, while torrent communities greatly benefit from obsessive hoarders who download and seed content beyond their personal interests. There is no economical solution to back up that much data on the cloud, and bitrot does set in.
If you're not spending at least as much money (+ labor opportunity-cost) on your own data-storage solution as it costs to back the data up to the cloud, your data isn't safe.
Eh, I feel like we've almost fully gotten away from bit-rot. Bit-rot was a problem because individuals don't bother to go to the effort of constant IT administration, and don't have the economies of scale required to afford good data redundancy.
But individuals don't have to back things up "personally" any more. With one command, I can send three copies of any file to three object-storage providers. Each one (according to the Dynamo architecture, which they mostly all implement) is holding 17 copies of the data on 17 shards, and replaces copies from the good ones whenever they go bad. Each provider's copies only exist in one region, but the region is different for each provider selected. As individual object-storage providers die, I can find new ones and sync my data over from the surviving ones. As long as "cloud [redundant] object storage" as a concept exists, I won't have to do much at all to ensure continued integrity of my data. All the details—including porting my data to new physical substrates when old formats die—are being handled under the abstraction, and converted into (tiny!) monthly fees that are lower than what even a zero-redundancy tape library would cost me.
And, of course, the data I'm sending is encrypted, and not even with a symmetric key, but rather a PKI key where the decryption half of the key is held in cold storage (i.e. in hardcopy base64 in a safety deposit box at a bank; and a few other places.) If I'm uploading rips of something copyrighted, nobody will ever know that but me. I'll be able to get the data back out when the time comes to dust it off and share it.