The simplest version would be to just duplicate each file N times, and spread the copies across the disc. Reads on an undamaged disc would be fast (faster, even, since in principle you only have to seek to the nearest version). Damaged reads will of course be slow since you may need to read all N copies (picking up good data sector-by-sector). And writes are obviously slow.
You could, in principle, cross-interleave all data across the entire disc such that you could lose (say) 20% of the disc surface for a cost of 20% overhead. And yeah, that would be very slow since you’d have to read the entire disc to get at one file. But you don’t have to do it that way.
My Bingoogling is failing me, but I recall an article a while back where someone managed to read enough data off a (music, I think) CD to play it…using some sort of high-zoot slide film scanner or something similar. The disk didn’t spin at all, the scanner was able to determine the tracks and pit locations visually – at least well enough to error correct.
I didn’t consider physically reassembling the CD for one second. Just “scanning” the individual pieces and putting the recovered data back together in software. A modest power digital microscope connected to a computer is all the hardware needed. The real effort is on the computer side.
And a polar co-ordinate system at that. And the polar system has to micro-compensate for the spiral, not concentric tracks. Gets pretty complicated, you bet.
Not hard in the sense that, if you have the equipment, it’s doable. Hard in the sense that such equipment is expensive and very few people have all of it. I’d expect only big science laboratories or actual data recovery specialists to have access to the equipment you would need.
I guess you could try to set it up where you moved the pieces by hand, and used some sort of stitching algorithm to put it back together, but I would be that would be hard when so much will look pretty much the same. It’d probably be just as hard as putting the fragments together.
Once you have an accurate image of the disk, I wouldn’t expect recovering what data is actually there to be all that hard, compared to process of putting it together in the first place.
I know the reason it took so long to find out how some SNES chips worked and the data they contained was not because there was no way to analyze the information, but because it took a long time before someone with interest had access to the equipment and scanned it in.
Take a bunch of pics. Doesn’t matter how precise the alignment is at all as long as you are sloppy and get a lot of overlap. The software can find the matching images for a given piece (in fact, overlap makes this easier). And you don’t need really great magnification.
The first hard part is writing software that finds the individual tracks, regardless of angle, curve, magnification changes, etc.
The hardware can be found in any modern bio lab. Some poor schlub is needed to take each piece and scan it. Might take a couple hours.
The real effort starts then. Not just the software needing to be written to find a decode track fragments, but to figure out how each track fragment goes with another track fragment on another piece. This is the part that means the OP has essentially nothing to worry about given the effort required.