Machida Museum of Graphic Arts, Tokyo
I have carefully selected the following images either directly off the web, or hand tooled on my laptop. In each case I resized each image to suit this project and applied 2 identical filters to ensure that they are to all intents and purposes unique. I then created compressed web ready images for the site and destroyed the 'originals'.
Imagine you are a picture. You see what a picture sees, which is nothing. You just hang there being looked at.
Now imagine what it is like to be a digital image in a Jini powered, networked to all buggery, Universal Object Space.
Being digital in such an environment is quite overwhelming. For a start there are small software agents which drift by to suss you out, decode your headers and see if you're carrying any latent copyright on you, or kiddie porn. They may be checking to see if you are that famously retouched QuickTime movie of Michael Flatley sucking his own cock.
Then there are those eyes which peer in from the network and figure out what you are a picture of. If there's a heap of whitey blue at the top, and whitey yellow at the bottom with a dash of brown, pink, green etc in the middle, then you may be classified as a girl on a beach.
Then come the offers of service. "I can sharpen you for this..." "are you crazy? i can get twice as sharp as that, and have my contrast balanced for half that." "who's that?" "That's Tron, he fights for the users." "Fuck Tron, what would he know anyway. If he had his way we'd all be fucking KPT again."
Tiny software objects now behave, if only to themselves and other of their kind, like a hyper free market where everyone speaks XML city speak. I knew the lingo, every good cop did.
And the users. "Only poor old Tron thinks the users are not actually real, or that he's a user, or some such shit" - "hey am i getting paid for this chat?" "no" . "goodbye, here's my rates, i gotto go get replicated."
Everything is capable of both requesting and providing services, and can remain persistent and unique within the Universal Object Space. A picture in this world loosely shares an inherited set of properties, such as a name, an author, what went into the making of it. On request it can provide a concise textual description of the contents of the image, marked up in some formal graphics markup language.
A modern image is a small service provider. So is a filter, a printer and an artist. You have dues to pay to the directory services, or no-one will know you exist. So you have to extract dues from those who may want what you have. It's a whole new world in there.
It's good to toss some meat into such a primeval soup.
In any complex ecosystem, be it biological, or strictly informational, small degrees of relative autonomy and a whole mess of randomness constrained by evolutionary forces spit up results which you could never predict, but which in hindsight seem sort of logical. Then they go and surprise you all the more because you left them running for another million years.
The photobots are small multilegged, many coloured creatures; great and small. They feed on colour, specifically the luminosity of colours. They don't all feed on the same colours, or even the same range of colours. Many photobots will be highly omnivorous, whereas many will also be extremely focussed about what colours they can and can't eat. They have different metabolisms, in that they use ingested colour to grow to full adult size, and then have the capacity to store additional colour within their bodies as fat. They have bowels, and when full photobots will shit a small blob of colour, a mix of whatever undigested colour was left in the bowel. They can smell colours with varying degrees of accuracy, and can tell living colour such as another photobot from dead colour such as shit, a background image, or the splattered remains of a fat photobot worn down and torn to shreds by circling hunters.
Photobots have 9 primary urges and 4 basic 'states of mind' which influence their behaviour. Their brains are Surging Urge Networks(my term) which reacts dynamically every time step. The basic urges are Move, Turn Left and Turn Right. Then there are slightly more complex urges such as eat, fight,or push as we call it so as not to confuse it with f for flush. When a photobot flushes, its colour changes for a few seconds, then receded back to its original. To the observer this looks like a blink, but to the photobots it smells like a sudden change of flavour, which gradually comes back if the flushing bug doesn't make a run for it. The urges are all directly hardwired either to their nose, bowel, heart (ie are we scared yet?), hale (ie are we happy and well) and hearty (happy and full).
When an urge is stimulated by a sense it gets a bit more excited. the more excited an urge the more chance it will pop it's cork (so to speak) and the urge is physically enacted. The photobot takes a bite, throws a punch, or drops a shit. When an urge fires, it consumes the energy and trickles a portion of that energy back into urges which are related. Related urges can be inhibited or stimulated by the firing of an urge and adjust themselves on a moment by moment basis by comparing the photobots overall sense of well being post urge with pre urge. This is as close to having a memory as the photobots get.
The second form of photobot learning, also evolutionary in strategy, is by passing on knowledge to their offspring. When two photobots love each other very much the lady photobots may well get pregnant. They are carried around in the mother's womb and consume a load of her energy and often either the energy expended in giving birth to them kills the mother, or the babies turn around and eat her. The baby at time of conception gets a scattering of genes from both parents, with a built in tendancy for symmetry in leg numbers. no-one has ever been able to explain to me why there is so much symmetry in the world.
The embrionic photobot is also imprinted with a superimposition of each parent's Urge Network. This provides both darwinian and lamarkian evolutionary models in a manner which mirrors our own education of children with none of the associated memory blowouts theorists have traditionally associated with lamarkian evolution.
Of course the third form of photobot learning is group learning, where a whole colony of many thousand photobots will stratify into a sort of caste system, with a cattle class at the centre providing food for other photobots nearer the edges. The cattle are effectively surrounded and hemmed in and die brutish, ugly and rapid death and devouring. At the periphery of such a colony adventurers wander about sniffing for food. The whole colony will, once formed, proceed in a disorderly fashion over the image, gradually, over a period of many hours, leaving a nasty black smear in its wake. This surprised us as they usually didn't touch the gray levels in the images they'd devour. Not at least until they had clumped and started turning the image below into a killing field.
I have once seen two independent colonies form but the machine was running so slowly that it became impossible to continue.
The purpose of this web site is to both display the photobots devouring a loosely related series of images, and to provide instructions for exhibiting the photobots as a long term physical installation.
Ideally you will have a number high end computers capable of running at least java1.0 within a version 4 web browser. They will be connected to a local network, and to the internet proper. Note this is for simplifying the set up procedure, not so that the photobots can 'get loose'.
Simply go to each terminal and, from this web site, open each image on a single machine. Try to run Internet explorer or netscape in full screen mode, and to provide touch screens if you want people to be able to actually look at the photobots up close, but for the most part hang them with bloody big flat screens on the gallery walls and some powerful data projectors.
Once the photobots have started eating, you may safely disconnect the network, but it is not essential. They remain isolated within their own browser windows. Allow them to evolve, uninterrupted for as long as you can.
The photobots were developed by Dave Sag and Mike Cooper as part of a project to 'add a little life' to Virtual Artists' home page, and based on theories developed over the preceeding few years by Dave Sag.
This exhibition was developed specifically for Machida Museum of Graphic Arts, Tokyo's Art On The Net 1999.
The Photobots are the result of over 2 years research and development. You can find more information on the photobots from either: