Digital Ephemera

Steven Pemberton, CWI/W3C, Amsterdam


An Einstein letter Einstein: 14 March 1879 – 18 April 1955

CWI is in possession of a couple of letters from him to a colleague (and the replies)


MC Escher CWI also has a letter from MC Escher thanking the MC (Mathematisch Centrum, former name of the CWI) for helping his work to achieve recognition.

Who knew?


That such letters still exist is a lucky stroke.

Usually such pieces of ephemera are not appreciated for their value until much later, often long after they have been thrown away.

Apple Lisa

Apple Lisa Example: in the 80's I was working on the programming language ABC (the forerunner of Python), and we had amongst other computers an Apple Lisa (the forerunner of the Apple Macintosh).

Apple Lisa

It was a big thing, and after the project it sat on a desk unused and getting in the way, until the point that it was just a worthless slow lump, with the result that it was unceremoniously chucked out.

Big mistake: an Apple Lisa is now a historic computer, and would be worth way more than what it cost when we bought it; just the mouse would fetch more than $500 now... It is one of a few computers that has increased in value over the years.

But at the time, no one considered that this heap of junk would actually ever be valuable.


Dr WhoBefore 1978 the BBC had no archiving policy.

So many seminal events were erased (for economic reasons!):

See "Wiping" on Wikipedia for more.


Luckily people are beginning to understand that we need to preserve our computing heritage before it is too late, while the people who knew how they worked, and still have surviving copies, are still around.


An ICL 1900 in the NetherlandsI recently discovered an emulator for the very first computer I ever worked on as a teenager, the British ICL 1900, which was an enormous room-filling, heat-producing machine, and cost of the order of a million pounds.

Luckily someone somewhere had preserved the original operating-system software (900 files, 17MB, all in assembly code), and so they could recompile it on the emulator.

Now I can rerun some of my earliest Algol 68 programs on the emulator running on a Raspberry Pi, which cost me €25, and the result runs much faster than it did then (when each program was allocated a maximum of 60 seconds runtime...)


One of the unfortunate side-effects of digitisation is that in the future we probably won't have any paper artefacts like letters to find in library archives.

And what organisations do when the information is no longer needed? Delete the lot!


You can't argue it from a cost perspective: 1 gigabyte of diskspace costs 3 to 5 cents now (and dropping), and spread over the lifetime of a disk is negligable.

Plan to keep your future history

Not just backing up the data and deleting everything.

Personal webpages should be preserved. I recently found some important data on the homepage of someone at MIT who had left in 1996!

And webpages should be preserved at the same URL they always had! As Tim Berners-Lee says: Cool URLs don't change!


Of course that means a lot of cruft will be kept.

But who cares? It is almost free to keep.

I still have all my email from the 90's: there's a lot there that's not of interest (at least, not yet), but thanks to it we now know when CWI first had a website, and I know how much filestore I was using in 1992 (117MB if you're interested, much of it binaries).


The only way to find out which ephemera will be valuable is to wait and see, and that means we have to keep it around long enough to let it accrue its value.