With current storage technology, it's almost impossible for physical media size or price to become relevant, the human's ability to add to the collection just can't keep up, unless there are some truly extreme conditions, like they automate the collecting, or insist on uncompressed video.
This is true, but some may not realise the important distinction of "actual stuff you personally archive", not "I downloaded this from somewhere".
I suppose even DVR platforms such as MythTV recording off of TV would be excluded here, as per your 'automated'... although it's a bit of a grey area, VCRs could be automated, but not really without changing the tape in a home setting.
Another grey area is more detailed "collecting", such as older video/audio, rarer stuff which does often disappear from torrent sites and such. An example being old TV series recorded direct from TV, to tape -> digitized. But even this, with personal intervention at each collection event, fits in terms of it being hard to fill up drives.
Really, preservation is the cost. Primary live/online, along with a RAID method (hardware over software raid typically) capable of per-disk patrol reads and overall raid consistency checks are vital*. Checksums of all files are a requirement too, and an offline secondary server with a full backup synced every so often.
Otherwise bitrot sets in, and you don't know. Either at the disk level, the raid level, or the filesystem level. And that's where the secondary comes in.
Of course, that doesn't help in case of explosions, aliens, or fire. One needs a secondary offsite for that. But my point is, actual real archival isn't simple.
* if you have a raid, even software raid and you're not doing patrol reads and consistency checks regularly, you're not really doing it right. LSI cards tend to require patrol reads and consistency checks set on, and consistency checks schedules (say, Sundays).
And of course if you don't have a script to dump megacli logs to syslog or what not, you don't really know if the raid is having issues. And you don't even know if consistency checks and patrol reads are running.
(In LSI terminology, patrol reads scan entire disks individually, looking for block read errors, and if found, that block is re-written from redundant data in the array. Consistency checks look at the status of the raid, especially checksums of all disks per virtual disk block. Different checks, both required.)
I think most data hoarders aren't technically knowledgeable enough to get that deep into futureproof archival strategies. I suppose priorities are individually different, though. Some might obsess over preservation, others are happy just accumulating stuff without backups, and data loss is a brief pain, quickly forgotten. The key problem of hoarding is the reluctance/inability to deliberately dispose of things.
A new ycomb application; platforms to enable safe data hording.
Play on fears of losing everything. How it's too difficult to secure data. How the risks are too high.
It's really no different than convincing people they need all sorts of weird lotions, pills, or gadgets, lest their lives will fall apart, their health decline.
The bonus here is, it's a client base unable to help themselves. Akin to selling medicine to a hypochondriac. I envision entire divisions doing deep-research into clients, with sliding scale cost. The ultra rich hoarder will have their collection stored on the back side of the moon, SpaceX delivered data pots, with solar, redundancy, robotic maintenance, and more.
