[Cryptography] Magnetic media destruction question

[Jon Callas]

> On Jan 8, 2026, at 16:17, Kent Borg <kentborg at borg.org> wrote:
> 
> Given crazy disk densities, I can imagine that there might be recoverable data on that. And, as erased data isn't necessarily erased, maybe there is some strategy for recovering files that get copied and more than once? Maybe it is a bit like piecing together DNA sequences.
> Were I paranoid about such things I would be happy with the current shredding size, but then insist on some post processing step. Heat? A caustic solvent? Stirring a cooking slurry of said?
> It is possible that the benefit of shredding is the destruction if data where the cuts are made not in the jigsaw puzzle of the remaining bits. 


Paranoia is the *unreasonable* fear that people are out to get you. The reasonable fear is called "caution." It's difficult to discuss real risks if there's paranoia out there. Let's look at it from a viewpoint of what's possible at reasonable cost (which is a variable -- if the USG had disks with Maduro's diaries on it, they'd spend a lot more to recover that than I would with DriveSavers). I know what you're thinking, and yet there's a lot of cryptological nonsense going on because of what people can imagine. It is, for example, difficult to have a rational conversation about the capabilities of quantum computers because people imagine what quantum computers might do and think that you just order one from Amazon and it shows up with Prime Delivery. In contrast, note the article I just sent that is at best mediocre news for quantum cracking. Similarly, I'm sure you can imagine the number 2^300, I know I can. At the same time there are only 2^265 atoms in the universe (or so I have been told, I bet they didn't count them).

Back in the old days, "bits" were written onto disks the way your imagination thinks. That's why old erasing standards from the 1980s are written the way they are. Today, there are not bits directly written on the platters, what's written is an analog waveform, and the drive itself relies on the platters not moving in relation to each other, and this is a good way to ruin the disks all on your own (and touching them with one's icky hands -- that's also a good way to ruin it). Modern disks are pushing the edge so much that they often have a quarter of them reserved as spare space to recover from errors. Also, note that many platters are made of glass, and glass has different characteristics to metal when shredded.

Also note that if someone went and made a shredder that could do 2.2mm pieces, they'd have a huge incentive to spread FUD about a mere 2.5mm shredder, and not even really FUD. They'd say that with today's requirements it's time to consider *their* shredder. That's 12% better for the same cost, after all.

I propose a test: saw a drive in half and then see what can be recovered. That's a 60mm by 120mm semicircle.

	Jon