Data Recovery Case File · Case 1450 · Three Things Have to Survive

He Preserved the Disk Perfectly and That Was Never Enough

This enquiry has been waiting thirty years to be made. "I have a floppy disk saved from a 1990s computer that has my father's book manuscript on it. I'm hoping to find a place that can recover the file from the disk so I can get a copy printed for my siblings, as it's the only writing we have left" of his. Somebody kept that disk safe for three decades, through moves and clear-outs, on purpose. And keeping the disk was one third of what preserving the manuscript required — the other two thirds nobody thinks about until a moment like this one.

The decode: what he preserved, and the two things he could not

The first dependency — the medium — and he did this perfectly. Magnetic media of that era is more durable than its reputation. A disk stored indoors, away from heat and strong fields, frequently reads after thirty years with no difficulty at all. Somebody kept this one deliberately through three decades of house moves, and there is every reason to expect the disk itself is fine.

The second dependency — the reader — and nobody preserves this. Computers of that family did something unusual to fit more data onto a disk: they varied the rotation speed depending on which part of the disk was being accessed, packing more sectors onto the outer tracks. An ordinary drive spins at one fixed speed and physically cannot read a disk written that way. Not a formatting difficulty — a mechanical incompatibility, in a drive type that has itself disappeared from every machine sold in twenty years.

The third dependency — the format — and this is the one that surprises people. Even with every byte lifted off the disk, a manuscript written in a 1990s word processor is a file whose internal structure is documented in places that no longer exist. Something has to know what those bytes mean. A file you cannot open is not meaningfully different from a file you cannot read.

So the position, stated plainly: the disk survived, the machine that made it did not, and the program that wrote it did not either. Preserving one of three dependencies is not preserving the data — it is preserving an object which happens to contain it, and hoping the rest can be reassembled later. Which, in this case, it can. But that is fortune rather than planning.

The doctrine of case 1450: data survives by moving, not by being kept

The instinct is to store carefully — to put the important thing somewhere safe and leave it there, because that is how physical objects are preserved and it works beautifully for a photograph in an album. It does not work for data, because data is never a thing on its own. It is always a thing plus a machine that reads it plus a program that understands it, and only the first of those can be put in a drawer.

What works instead is migration. A file copied forward every few years — onto whatever medium is current, into whatever format is current — outlives every technology it passes through, because at no point does it depend on anything that is about to disappear. The file that has been moved eight times is safe. The disk that has been kept perfectly is a bet on three separate technologies still existing.

Why nobody does this: because it feels like the opposite of preservation. Copying something onto new media and letting the old copy go feels careless, and keeping the original feels responsible. The instinct is exactly inverted, and this archive is full of the consequences — the tape nobody can play, the drive nobody can connect, the file nobody can open.

And then there is what he plans to do with it, which is the best part of this enquiry. He wants it printed, so that each of his siblings can have a copy. Paper has no reader dependency and no format dependency — it needs no drive, no software, no operating system, and no compatible anything. It is the only storage medium in this archive that has been continuously readable for a thousand years.

Which makes the ending of this case the answer to it. The thing he intends to do with the manuscript is the thing that would have meant this enquiry was never necessary. Printed, in several houses, it survives everything. A book that exists on paper in four homes is backed up more robustly than anything else described in these fourteen hundred and fifty cases.

Why this is case 1450. Almost every file here is about a medium that failed — a platter, a chip, a card, a board. This is the one where the medium is the only part that worked, and it still took thirty years and a specialist to read a document that was never damaged at all.

On the bench

The disk was read on period-appropriate hardware capable of variable-rotation encoding — computers of that family varying rotation speed by track position to increase capacity, which a fixed-speed drive cannot physically follow. Magnetic condition was assessed before any read attempt. Flux-level capture was performed rather than logical reading, recording the raw magnetic transitions so that decoding could be attempted repeatedly without returning to the disk. The document format was identified and converted to a current open format, and output delivered in printable form.

The outcome

The disk read on period hardware, captured at flux level, decoded and converted to a current format, and delivered ready to print. Free assessment, one fixed written figure including VAT, and no recovery, no fee. The decode, and the doctrine of case 1450: data has three dependencies — the medium, the machine that reads it, and the format that gives it meaning — and only the first can be put in a drawer. You preserved the disk perfectly and the disk was never the part at risk. Data survives by being moved forward, not by being kept still. And the printing he plans is the most durable thing anybody in this archive has proposed.

Our case files are drawn from genuine enquiries received by our laboratory over the past ten years, anonymised to protect client confidentiality. Each one describes the diagnostic and recovery procedure our engineers apply to that fault, using the equipment listed.