The dangers of code

A moment searching reveals numerous companies who will help you with a persistent technology problem - accessing data written using technology that is now out-of-date or hard to find. A great example of this was the BBC Doomsday project - unlike the paper (well, vellum) based original, this was difficult to read a mere 19 years later as it was based on the by-the-outdated BBC micro. See here for more details.

The LRB has an unusual example of this phenomenon this week - shorthand. You see, there isn't a single shorthand. Most people these days learn Pitman, but the modern version of this is a simplification of the Victorian original. And then there are the now-dying if not dead historical systems due to Gurney, Tailor, Byrom and so on. (The picture is from an article in Wikibooks - it shows the Lord's prayer in some common shorthand systems.)

Different systems are often mutually incomprehensible: an expert in one cannot read any of the others, especially if the writer is an advanced shorthandist. There is also the issue that many of these systems evolved significantly over time, so an expert in modern Pitman can't even necessarily read something from a hundred years ago. Thus a diary, say, written in 1895 Universal Stenography is about as difficult to decode as a message in 512 byte RSA. Both are breakable (even without any hardware, user, trust model or failure recovery based attacks), but both require quite a bit of effort. Perhaps the military should take up obscure shorthand systems before quantum computing blows RSA completely out of the water?

More to the point, never underestimate the difficult of decoding a lot of stuff. The problem isn't the decoding: it is reading the decrypt and figuring out what is valuable. That's why those Victorian diaries are still lying around unread: the combination of a really small signal to noise ratio and moderately difficult cryptanalysis is a killer.