Gah. I had initially replied to the parent post with the Great Zero Challenge link, submitted and saw that you beat me. :)
AFAIK, the technique for magnetic disk recovery after one dd pass is to remove the heads and attempt to retrieve the data using laboratory equipment (more sensitive than the original heads and _vastly_ more time consuming than the commercial data recovery procedures).
You can't do that with SSDs, maybe you could attack the flash controller and exploit something like a timing attack, or maybe you could try your luck against the load leveling algorithms and dump the data from the flash chips themselves, although this would involve at least desoldering the flash chips in a big device or doing something nastier in an integrated package like a CF card. If you lose there, the next stop would seem to be something involving an electron microscope and trying to identify cells that used to be a zero and are a one now, and vice-versa -- judging from: http://hackaday.com/2008/01/01/24c3-mifare-crypto1-rfid-comp... it seems like state of the art currently is individual gate analysis, I think this would have to go farther.
I agree with you that it ought to be possible in principle, although I find it unlikely that data recovery companies would pass up revenue when they're quite able to do it. The only explanation to me would be that they're being paid off by somebody, e.g. governments, but again... WHY? My theory is that it used to be possible, back when the data density was much lower, larger chunks of material were involved.
SSDs:
I would have thought that decoupling the flash chips from the controller logic is the easiest way to get at any hidden data. You'll have to piece the data together unless you can reverse-engineer the load-balancing algorithm, although there won't be much left after one pass of zeroing anyway. I have my doubts that you can get at residual charge from previous bit values in a useful way.
AFAIK, the technique for magnetic disk recovery after one dd pass is to remove the heads and attempt to retrieve the data using laboratory equipment (more sensitive than the original heads and _vastly_ more time consuming than the commercial data recovery procedures).
You can't do that with SSDs, maybe you could attack the flash controller and exploit something like a timing attack, or maybe you could try your luck against the load leveling algorithms and dump the data from the flash chips themselves, although this would involve at least desoldering the flash chips in a big device or doing something nastier in an integrated package like a CF card. If you lose there, the next stop would seem to be something involving an electron microscope and trying to identify cells that used to be a zero and are a one now, and vice-versa -- judging from: http://hackaday.com/2008/01/01/24c3-mifare-crypto1-rfid-comp... it seems like state of the art currently is individual gate analysis, I think this would have to go farther.
From a commercial perspective, it seems like there still is some drive to build better tools: http://www.theinquirer.net/gb/inquirer/news/2008/10/28/flash...
tl;dr: I agree, unless you are doing something criminal, I would use a single dd if=/dev/zero of=/dev/diskdevice and get on with life.