Artificial Intelligence with Doors
Project proposal for "OpenDoors 2004", the Doors of Perception design contest
Date:
16-08-2004
Name of entrant:
Wilfried Hou Je Bek
Name of organisation:
socialfiction.org
Contact:
info [at] socialfiction [dot] org
Short description:
'Artificial Intelligence with Doors' describes a system consisting of 16 doors in a 4 by 4 grid.
We will explain this system as:
- crowd information processing unit
- read/write head of a programmable analogue computational device
- landmark and urban prism
- example of DIY urbanism
- engine of serendipity
Somewhat idiosyncratic, a small software program was written to demonstrate some of the possible applications of the 'Artificial Intelligence with Doors' system. All images included in this document are made with this program.
URL:this document is available at http://socialfiction.org/doors.html
++++++++++++++++++++++++++++++++++++
Artificial Intelligence with Doors
Functional Description
The 'monolith', as we will call this structure henceforth, is a system of 16 doors placed in a grid of 4 by 4 doors. Each row of doors is distanced from the next at doors-length. This results in a permanently fluctuating labyrinthine structure.
Each door can be only fully open or closed. From here on we will talk about these positions in binary code:
closed = 0 (= white in the software)
open = 1 (= black in the software)
Each door represents a bit, the monolith can be in 2^16 or 65536 possible states. This is to say that it can encode 65536 unique symbols. Half of this, 8 bits, can encode 256 characters, sufficient to encode the entire alphabet, capitals and a range of peripheral textual items included. For reasons of speed also, this option is preferable, as in this way the monolith is a dual processor, able to cater for 2 parallel streams of information simultaneously.
Notice that in the monolith form and information are exactly the same: when all doors are closed this is read as 2 strings of 8 bits ('00000000'). This string doesn't mean anything according to the conversion table used in the demo software. But for instance a combination of doors that amounts to '00110001' translates into "a" in normal script. The change of position of 1 door might change the encoded letter into for instance a 'd' or a 'E'.
Screendumps
Left screen: position of doors
Middle screen: each space behind a door is visualised as open(black) or closed(white)
Right Screen: a string of 2x8bits, to keep a track of the history of information passing through the monolith. The present pattern is always at the bottom.
All doors closed
All doors open
F12: list of available functions
F2: number of Random generations
F4: intelligence produced after 60 randomly generated doors changing position
Typical example of screen after some generations.
The letters 'a' (the first 2 columns) and 'z' (the last 2 columns) encoded in doors.
Other Numbersystems
8 Bit encoding is common computer practise, but not by any means the only valid choice here. For the sake of argument here are some other possible ways to encode information in the monolith.
The human DNA can be reduced to 4 amino-acids. A DNA molecule in turn is a string of 4 amino-acids. 2 Doors can encode one amino acid, 8 doors encode one DNA molecule.
Colours, sound and Chinese characters can only be very poorly encoded in 8 bits. For purposes like these all 16 doors of the monolith can be used. In the past 16 bits were pretty much state of the art, but nowadays 128 bits and higher are common. But whereas every increase in fidelity in a digital computer takes millions of Euros in R&D, the upgrade of the monolith is only bounded by the availability of space.
People can decide between themselves certain codes; one lover can inform the other, by turning the doors into a certain fixed position (like: 01010101 10101010), whether their love can be consummated tonight or not. This ad-hoc symbolic language can be very complex as 16 doors does give ample possibility to keep adding new symbols to the vocabulary.
Reading the Monolith
The demo software has an inbuilt function to randomly create a sequence of doors changing position. After every change the pattern is compared with the conversion table. Running in this mode, the Monolith will generate a stream of characters like: ttyyHF%%ccG. Or something equally as incomprehensible. The lack of meaning of this string is as predicted, as the process guiding the changing position of the doors, is random. This function is mimicking the proverbial monkeys in the basement of the British Museum typing away, statistically at one point having to hit upon a sequence of letters that are exactly the same as the 3rd Canto of Lord Byron's Don Juan. The monolith needs an external guiding-system that controls the opening and closing of doors to make the patterns we can read from it meaningful.
Crowd Intelligence
Ants by themselves are simple entities that do simple things, but the behaviour of the swarm as a collective is regarded by biologists to display a certain kind of intelligence. From this perspective of bottom-up, undirected self-organisation, cities have been subject to academic analysis since the 1920ties. Taking the cue from this ongoing research, the monolith can be used to determine if urban crowds should be thought of as stupid monkey or on the contrary as smart ants. Assuming, for one minute, the existence of an emergent urban crowd intelligence, this groupmind might be able to guide the process of changing doors. This would mean that, even though we are not aware of this ourselves, we, as crowd, produce an meaningful stream of patterns while passing through the monolith. We might even seek to communicate. But if extensive analysis on the patterns of, say, 100 days of slapping doors, doesn't result in finding a coherence that could betray intelligence, this doesn't have to mean that there isn't any; we might just not be smart enough to recognise it.
Intelligence (some observations on the side)
How can we recognize extra-human intelligence? This question becomes harder to answer the more we grasp the fundamental difficulties in understanding what constitutes intelligence in general. For instance: statistical analysis has shown that the zillion or so digits of PI are without any internal coherence, but they is a coherence as they are produced by a formula. What seems random at first might only need the right filter to turn out to be produced by something highly intelligent.
Programming the Monolith
Just because we are used to associate binary code with computers, doesn't mean that doors can't have their way with it. In the same vain are computational processes, formulised in an algorithmic language, by no means only executable by computers. To prove this last point socialfiction.org has started the .walk [pronounce dotwalk] project. This project undertakes to built a fully functional computer, constructed in the thin air of walktime; emerging from the total of interactions of people executing algorithms by means of walking.
Past experiments have shown that the predictability of encounters between the .walkers is crucial. Embedded in a .walk application, the monolith will be a powerful mechanism to facilitate the short-term information storage and transaction (input and output) between independently operating .walkers. In this set-up the monolith is a interface between pedestrianised processors.
.Walk is the logical follow-up on our experiments in generative psychogeography; the employment of simple algorithm to determine offbeat paths to enable unprejudiced exploration of cities. This close relationship between .walk and psychogeography is important to keep in mind because .walk applications are running on top of psychogeographical walks.
The reactions of people when exposed to .walk are known to vary immensely, but we find it encouraging that computer scientists immediately grasp the idea and deem it in accordance with the principles of the Universal Turing Machine. This means that a .walk computer can solve anything that can be (logically) solved by an algorithmic machine.
Programming .walk demands little programming knowledge and it has the benefit to be much less error-prone than ordinary programming languages. It's better for your general physical condition too.
More about .walk at: http://socialfiction.org/dotwalk
More about psychogeography at: http://socialfiction.org/psychogeography
The Monolith as Abacus
Children will like the monolith because children like labyrinths. Teachers can use this natural sympathy towards the monolith, to enhance students attention by re-appropriating the monolith for educational purposes. By turning the monolith into an abacus, it can be used to better explain basic mathematical operations like addition and subtraction. It could work like this: the first 9 doors are reserved to count to 9 (0 is all doors closed) while the spare doors or used to keep a tally on the tens. Clearly visualised the pupils can see how 2 doors already open and another door being opened, makes 3 doors open (=2+1). While the pupils are doing this, they immediately learn to read the monolith, making it easier for them to understand the monolith as a configurable communication device later in life.
The Monolith as Something to be Interpreted
To most people the monolith will seem like a useless and functionless mumbo jumbo of doors inconveniently filling up former empty space. Perhaps few people will ever bother about finding out the details behind the monolith, let alone in learning how to program it. This is perfectly alright because the monolith will keep up with them.
The monolith is a design that evokes serendipity: the act of finding something useful while looking for something completely different. It is meant to generate unintended consequences by just being present with the same self-convinced air of a Lord Byron, it will draw attention. The comparison is to pompous to really use, but it does make sense to call the monolith a Stonehenge in doors. Just as Stonehenge is all the more impressive and mysterious because of the doubts around its function, the monolith excites people into speculation about the nature of the object. It's the object itself that inspires the function instead of the other way around. And as you remember Stonehenge is rumoured to be computer too...
'Serendipity' is one of the key-principles socialfiction.org has defined for DIY (Do-It-Yourself) urbanism. The other 2 principles are 'generosity' and 'turriphilia'. These 3 key-principles remind us of what has been the concern of psychogeographers of all ages: the necessity for people to actively participate in the creation of their own neighbourhood, independent of professional, cultural, political and/or financial common sense. Turriphilia, the manic desire to built towers, doesn't necessarily have to result in large buildings, a construction of doors like this is a turriphiliac folly too. What unites all turriphiliacs is their contagious and generous enthusiasm.
What a Monolith can Learn
The monolith adds an enormity of empty surfaces when installed in a neighbourhood. A typical door measures 2 by 1 meters, a door has 2 sides, multiplied by 16 doors, this equals to 64 square meters of surface. To enhance the impact of the monolith as landmark, we would however recommend doors of less modest size. These doors are one giant hexapodal message-board, free to use by anyone. The monolith will prove to be its own best archive.
Groups looking for a place to meet in are welcome to make the most of the monoliths flexibility of form to make small rooms out of it. Rooms that vanish when the meeting is over. This too is a kind of programming.
The Monolith as Landmark
The quality undisplayable by the software is the kaleidoscopic feel of the monolith. Especially when the doors have their own colourful history to tell. Now and then, for a short interval, 4 door in a row, will be aligned, tuning the monolith into prism on the city behind it. This window on the city changes position between rows unexpectedly, the keen observer can walk around it, picking a good spot, perceiving the city framed in hundreds of pictures sliding by.
Psychogeographical walks result in unexpected finds by taking the trouble to step outside of daily habit: serendipity is not something that falls from the sky, it needs some sort of initial action ("finding while looking for something different") to present itself. Likewise the monolith does demand a general degree of willingness, already present in people, to make sense of the world around them. A willingness, the highly manipulative form of the monolith appeals to. All scenarios, all possible approaches, mentioned here only go as far as our limit fantasy allows us. The success of the monolith as a serendipity machine lays in it ability to provoke people in imagining uses for it beyond everything suggested here.
Finally
The sense in which we talk about the monolith as artificially intelligent, is in the sense that the monolith is a signal, that needs a filter to become intelligent. The ability of the community to make sense of the monolith (and to keep making more sense of it over time), is the true sign of the existence of urban crowd intelligence.
Too often specialists march in thinking they know it all, without contacting the inhabitants they change the neighbourhood beyond recognition, and never look back afterwards. The monolith, like any other platform for social innovation, starts as an intruding parasite on its host. To be truly innovative, it must before all other things, become intrinsic part of the social system it seeks to innovate. At the moment the monolith is discussed between people who never met before, it's production of serendipity has begun. DIY urbanism enfolds from a gift as small as a wicked idea of what can be done with a ridiculous bunch of doors; 16 doors mockingly named in remembrance to a science fiction book (and movie) that deals with the nature of intelligence.
