You mean data rot, not bit-rot. And it's not a concern on basic user level most of the time. But sure, redundancies like the once implemented by filesystems like ZFS help if you're paranoid much. Also shield your PC from cosmic rays. :P

Cosmic rays? I thought its little devils flipping the bits. ;) Anyways, I already had one ECC-related event in x8sti, and recently primary CPU fan (scythe shuriken) failed and board managed to gracefully downclock absolutely without any damage. All these little tin-foil hats,.. they actually pay off. Better paranoid than depressed :D

What I'm saying is that the basic user doesn't really have to worry about such things. It's much more realistic to just worry about simple disk crashes (god help us all) than things like data rot which should concern high load and mission critical systems only.

Nobody is as enraged and concerned as a "basic wife" ^) , who finds out her 3 hour marriage video to be corrupted ^) - and all the backups corrupted as well. ^)) / LOL

Data rot does not occur only in high load or mission critical scenarios. It can happen everywhere and it isn't even uncommon to happen. If you don't check for corruption in your data, you can easily spread the corrupted data to all your backups, so backups alone won't help you there. So calculating checksums and/or using file systems like ZFS or Btrfs as mentioned before is not a bad idea if you've got any important data which can't easily be recreated.

Data rot is a very specific term that encompasses very specific types of, well, data degradation. This does not include regular cache read/write errors for example or bad sectors. It includes things like bits losing their magnetic orientation due to time and non-use.

Yeah, yeah explain this to my wife :P Bits flipped, its me who got to deal with that. :P
Either way, its ECC and btrfs for me since then.

Or multiple file backups, to various locations/sources.
Data rot happening simultaneously to however many backups you implement is unlikely.

Replicating damaged files/overwriting known good backups with data rot accidentally is always fun,
from a distant third party perspective.

Look, if you're that worried about corrupt GOG games, just use a yttrium orthosilicate crystal doped with praseodymium, cooled to 3 degrees Kelvin, held in an electric field and write on it with pulses of laser (ie. quantum memory). It's really very simple. Amateurs...

Buuhh, thats so much yesterday! Papyrus is the future!

Edit: its made by Gods chewing on weed!!.... But its a secret!

You're not wrong.
I'd rather smoke it.

Lol!

"Well, according to some new technology designed by an Indian engineering student, an extraordinary 256GB..."

Forget this thread, it won't help you! ;)

Data rot isn't really as specific as you claim it to be, but actually I was referring to the meaning of it as you describe here. When you encounter bad sectors, you usually notice something is wrong, so you can easily take action. The far worse situation is when your data is silently corrupted and this is not limited to enterprise applications at all.
Just to be clear, I was talking about data being silently corrupted for whatever reason. I guess the question of this thread was about just that. Using MD5 for verifying data integrity works well in that situation and its cryptographic weaknesses do not matter at all in this case.

I'm not disputing MD5 at all. It works™ as far as verifying a file integrity goes. I'm just saying that throwing technical terms like data rot around to non-technical users is furthering their paranoia for no good reason at all. Ask the OP. He probably had never heard of data rot before and is now hiding in an underground bunker with a tin foil hat because of it. :P