http://dablog.ulcc.ac.uk/2009/05/31/on-the-limits-of-preservation/
A recent article in New Scientist on the outer fringes of the chiptune scene prompted me to think about preservation, emulation and the fact that some digital things simply aren’t preservable in any useful sense.
Chiptunes are typically created using early personal computers or videogames and/or their soundchips. In that respect, they depend on 
technology preservation – the museum approach to digital preservation. Chiptune composers either use the systems as designed, programming them directly to create their music, or alter them in some way using techniques collectively known as ‘circuit-bending’, which makes the machines capable of producing sounds that they could not have originally produced. Some aspects of the chiptune scene utilise more modern synthetic techniques to recreate the sounds produced by these early chips – these are, in a loose sense, emulating the original systems, although not in a way that would allow you to use original software to create your sounds. But some adherents of the chiptune genre are going further, using the sounds of the systems themselves in their compositions.
The article which set my train of thought going covered Matthew Applegate’s (aka Pixelh8) concert in late March 2009 at the National Museum of Computing, which amongst other things used the electro-mechanical sounds produced by the Colossus and by hand-cranked adding machines. Now in this respect, as in many others, digital isn’t different to the analogue world. One could argue that found sounds have a long tradition in Western music, from Tchaikovsky’s use of cannon in the 1812 overture through to Leroy Anderson’s typewriter symphony and the work of the Boston Typewriter Orchestra, who eschew traditional instruments entirely and concentrate solely on percussive typewriter music, interleaved with the odd end-of-line bell.
But pixelh8′s work reminded me of some earlier, and unlikely, uses of computers to create music which I fear is already lost – although discerning listeners might well argue that we are all much the better for it being lost. One depended on the noise-making abilities of computer peripherals such as line and chain printers, the fiercest of which were capable of noise levels that drowned out conversation their vicinity. Devious users could create files which generated recognisable rhythms when printed on a specific printer. I’ve seen reference both to the “William Tell Overture” and “She’ll be Coming Round the Mountain” being produced by this method, which demanded detailed knowledge of the printer’s workings (but no programming knowledge – the effect was produced merely by printing a carefully-crafted file.) It seems that at least one recording of this exists, from the mid-1960s. Good ideas like this don’t die, or are reimagined by later generations, and 1999 brought us the Symphony for dot matrix printers.
These examples, in common with some of pixelh8′s work, are dependent on the relevant hardware still existing and being in working order in order to allow us to continue to perform them, or to create new compositions. That’s certainly a challenge for things like line printers, and I imagine the same is true of mechanical typewriters. These, however, are relatively simple mechanical devices and, so long as someone is still interested in maintaining them, they will be repairable for many generations to come, even if they are no longer being manufactured.
But other forms of of computer music aren’t so easily re-created. At about the same time as people were creating line printer music, others realised that the computer’s own electronics generated sound in an unplanned-for way. Almost all computers of the 1960s generated substantial amounts of radio frequency interference of a type that’s now illegal. The clock speeds of those computers meant that the interference fell mainly in the medium wave band, between 500 Khz and just over 1Mhz, exactly the right spot for any consumer AM radio to pick it up. (Present-day systems, if not shielded, would be generating interference for microwaves and mobile phones, not AM radios.) Enterprising programmers with time on their hands and dedicated systems to play with wrote programs which generated RF interference which approximated a tune, since the exact nature of the interference depended on the instructions being processed, the data being fetched, and probably the addresses from which it was being fetched, as well as the properties of the CPU circuitry.
I encountered one version of this, using one of DEC’s PDP10 processors (probably a KI10, although I can’t be sure) and the DECUS program library also contains a similar program for the PDP8. I’ve come across one reference to a recording being made, this time of the same technique on an IBM 1401, but the available example uses samples from a recording to make tunes rather than the original recording itself being musical.
All these examples illustrate, to different degrees, the digital preservation problem that pixelh8′s work brought to mind. The RFI music was exquisitely sensitive to properties of the original electronics. It wasn’t enough that the computer ran the same instruction set. DEC’s PDP10 series had four generations of processor – the KA10, KI10, KL10 and KS10 – each of which could have run the music program without error. But since each was entirely different at the circuit level, only one of those machines would produce the right sort of radio interference to generate the desired effect. Emulators of these systems now exist, and a number of manufacturers have produced hardware-level recreations such as the Toad 1, Foonly and Systems Concepts machines. But neither the software nor the hardware emulators will have the same radio interference properties, so again, although they will run the original program without errors, it won’t produce the desired effects.
In fact, it would be extremely difficult to produce an electronic equivalent of those systems again, and as technology moves on such problems will only get more difficult. The early systems, which used primarily discrete components (transistors, capacitors, etc) could be built again by sufficiently dedicated enthusiasts. But chip fabrication is another thing entirely. The music that those programs and systems produced is essentially unpreservable except by recording it at the time. The significant properties that emerge from the combination of software and hardware are extremely difficult to characterise, and no environment that allows the software to run will emulate the one property of the original environment that really matters.
Some stuff really, really can’t be digitally preserved. It’s good to remember that sometimes.
(And if you are aware of any other recordings of computer ‘music’ of this type, I and others would very much like to hear of them.)
[Updated to remove an embedded YouTube video which causes problems with the Iphone. Instead, you'll have to follow the link to see and hear some hardware-based chiptune hacking.]
3 Comments
Enjoyed this post very much: it’s very good fun and sets up some of the important issues in DP.
I wonder about ‘period instruments’? The authenticity claimed for historically informed performance derives a lot from the musical texture in which notes and rhythms are arguably more closely aligned with a composer’s own performances or intentions. But the value of authorial authenticity is not a given and so the search for authenticity is often naive and frequently risks becoming prescriptive. I mean Rossini could not have imagined the WT overture on a dot matrix printer; and neither Beethoven nor Schiller could have anticipated the seminal performance of the no 9 at the Brandenburg gate in 1989 where they had the audacity to change the words. But surely they would have been delighted (Rossini excepted perhaps?)
Performance is sui generis and irreducible: attempts to capture them are always/already somehow flawed. Is emulation or migration any different? What are the significant properties of music? Discuss …
As you say, William, artistic performance is something that we expect to be unique on each occasion, and we also don’t expect the same degree of fidelity to the original – and sometimes a later interpretation is seen as better than the original.
But in much of what we’re doing with software and information preservation, we’re not treating performance as an artistic concept, and there are some who expect to be able to formally validate that it’s been done correctly. I think one can do this for many useful cases, but it isn’t always possible. And part of the problem is that you don’t always know when your automated tests have failed you.
Emulation and migration always produce change, and are in some sense a reinterpretation. MS Word on a Mac is not exactly like MS Word on Windows, and there’s an extent to which we expect that to be true: the program is expected to be adapted to its environment, just as a tune needs to be adapted to be played on a piano versus a guitar.
All of the examples I gave come down to someone taking an aspect of a computer that wasn’t a part of its design, and making it fundamental to the behaviour of a piece of code, or even the rendering of a file. These examples were intentional – some others aren’t.
>If you are aware of any other recordings of computer ‘music’ of this type, I and others would very much like to hear of them.
Since you ask…you’ve already mentioned [the user], whom I saw performing in an art gallery in Dundee in 2003. It was a live interaction with a disused silo in Canada, involving making amplified sounds on stage which were transmitted to the silo over ISDN lines, and broadcast into this empty space. The reverberating sounds were then fed back to Dundee, along with some visual analogues of what we were hearing. In one sense I suppose this was a form of sound-art being enacted live over the internet, which is an intriguing proposal, but then again I also have to admit it wasn’t very interesting to listen to. See this page and this review.