Will heating the platters beyond curie temp make disk-splicing impossible?

On Thu, 15 Apr 2010 14:15:32 -0700, GreenXenon wibbled:

This is the answer:

(watch if you enjoy pyro and bad things happening to French made cars)

How about cooking the platters directly above the blue flames of a gas stove? Will that eliminate the magnetic properties of platters and render disk-splicing useless for recovering data?

What do you have on these disks that's so important to hide?

That's nonsense.

John

You're just not that important that anyone would bother spying on you, RIchard.

To a large extent, yes, but I think it depends greatly on how much effort one goes to in destroying the media -- a disk that's just been broken into, e.g., a half-dozen pieces is probably well worth putting back together.

Sending the platters through a chipper should be pretty effective, I expect.

I've been told that during the cold war years intelligence agencies would meticulously splice back together paper documents that had gone through a shredder.

But of course all governments have an interest in suggesting their intelligence capabilities are far greater than they really are too.

One newer topic in security is "deniable encryption," wherein you purposely setup your encrypted hard drive (or whatever) in a manner that goes no obvious sign whether it's a bunch of encrypyed data or if it's a just a disk full of radom gobbledeegook that you put there when you erased the drive for completely legitimate (e.g., privacy) reasons. Cool idea... I've never encrypted an entire hard drive, but I definitely have used programs like "disk eraser" that fill the drive with random data when I've sold off an old drive, precisely to ensure there wasn't anything of use left.

---Joel

Dissolve them in acid, dilute it with 1000 gallons of tomato juice, scatter one cup at a time in different rivers, lakes, and oceans.

John

IBM and the German government announced a program to do that on a huge scale with the shredded Stasi archives--which occupy thousands of garbage bags of confetti. They're doing it by scanning and recombining the scanned images. Could be pretty cool if it works. Fortunately the perps have had another 20 years to finish dying off....

Cheers

Phil Hobbs

IBM and the German government announced a program to do that on a huge scale with the shredded Stasi archives--which occupy thousands of garbage bags of confetti. They're doing it by scanning and recombining the scanned images. Could be pretty cool if it works. Fortunately the perps have had another 20 years to finish dying off....

Cheers

Phil Hobbs

Older disks could be treated with a magnetic suspension and make the transitions visible in a microscope. I think modern disks are too dense for that. An atomic-force microscope could map the magnetic domains of even a fragment of a disk.

Yup, turn it into filings. Scatter it along the highway. Toss some into the ocean. Feed some to your cats.

I believe there is software to do that now, based on scans of the shreds. That technology is used for old crumbled documents like Dead Sea Scroll sorts of things.

A constant stream of them could burn up all the supercomputer arrays the NAS has, and bring down the power grid to boot.

One could also XOR into the files all sorts of spurious and silly and contradictory documents.

John

No. Alas, you haven't any clear idea what the 'curie point' is for any given disk, since you don't know the magnetic formula. Disk-splicing will lose data at each stress region, since stress leading to fracture or bending is also capable of demagnetizing.

Similar, no. Recovery, yes. The volatility has a time decay constant of a second or so, and it takes a long, temperature-dependent, delay after power-off to thermalize the information to nonexistence.

If the time-decay-constant is a second, then will it take a second for the data to be completely lost when the power is offed?

Vertical magnetic domains makes this a bit tougher, too.

Stick a gouge into the spinning disk.

Encrypted data has a pretty high entropy. Garbage data isn't likely to look anything like it.

Like the Health Care Bill?

Thermite ignition will be pretty ruinous to its stored data.

The most famous recent one was the Iranian student reassembling shredded US intelligence documents after they invaded the US embassy in Iran.

Most users seem unaware that format does little more than alter the drives directory table. That action is easily undone. Secondhand computers from big organisations that *should* know better crop up with monotonous regularity. A full long format writing all zeroes is better but would not really challenge a data recovery specialist.

After you have rewritten the drive with random data a few times a la disk eraser a government intelligence agency might be able to get at least some of the old data back but they would only bother trying if they had very good reason to do so. Single bit errors in encrypted and/or compressed data are fatal to decoding.

You can hide pretty much anything in JPEG files without affecting the decoded image if you know what you are doing. See steganography

BTW GreenXenon is a paranoid netkook.

Regards, Martin Brown

Belt sander might work fairly well.

True.

False; the usual 'long format' involves a search for bad blocks, and DOES erase and rewrite all data; nothing useful remains of the data on the disk, because MODERN disk drives don't waste any of the platter area. Thirty-five years ago, there might have been some data residue.

True, and there's an implication here: all hard drive data is modulated in a scheme like Manchester coding, RLL, GCR, eight-fourteen modulation... there's lots of schemes and lots of names. These are all history-dependent in some way, i.e. a ruined bit kills ALL the subsequent data to the end of the block. There's also error-correction data, which assures the reader that his as-recovered data is wrong. There's no way that the hypothetical 'intelligence agency' can reconstruct the block well enough to match the error-correct code, and that means clobbering 'most of' the bits is good enough to delete the data.

Even a few bits bad means the error-correcting code will make the result into evidence-unusable-in-court.

Of course not. If you heat them beyond the curie point, all the magnetization will disappear.

Are you skybuck or numnuts under a new nym?

Good Luck! Rich

Why would you do that?

Send the drive an ATA secure erase command, wait for it to finish, and recovering any data will be somewhere between "impossible" and "not worth it unless you're at the level where the national security advisor uses your mugshot as a dartboard". And probably closer to the former.

Even overwriting with zeros ("dd if=/dev/zero ...") will be adequate for most purposes; however, that doesn't deal with any blocks which have gone bad and been remapped by the firmware, while a secure erase will overwrite everything, including blocks affected by physical defects.

Yes. At room temperature, the half-life is typically measured in seconds, but cooling can increase this dramatically. An experiment on DRAM:

found that fewer than 1% of bits decayed after 10 minutes when cooled to

-50C, while cooling to -200C with liquid nitrogen resulted in a decay of

Some SRAM chips can retain 80% of bits for over a week at -50C:

Quite the opposite, modern drives reserve about 3% to 5% of the platter for "live spare" blocks, and remap the sectors on the drive invisibly to whatever OS/hardware is accessing the disk. It is older drives where the space on the disk was actually physically mapped to the CHS address.

Not always.

not for 30 years

Your lack of solid understanding is glaring here. You are mixing varied techniques at different levels of the read/write chain with wild abandon here.

constant

No. About 1 second for 63% of the data to become thermalized. It is an exponential tail like a capacitor discharge, which is what is going on.