The problem of the final choice

The idea of dividing a database up into small sections of people who might be of interest to each other might be more clearly understood with another example. For this, let's return to the real world, matchmaking experiment I tried out in my home town.

In that experiment, it wasn't possible to provide client side data processing and selection software because very few of the people owned computers. There was also the problem that in human matchmaking there is no program that can be devised to replace the natural way humans can judge each other through face to face interactions in real life.

As a solution to this problem, a space was created in the memory of my own computer to simulate the local town. Into this space were placed representations of all the town's pubs, bars and other social meeting places. I then randomly distributed all the clones (records of questionnaires) into these virtual meeting places.

A program was then devised so arrange for each clone to "look" at the records of all the other clones in the meeting place it had been assigned to and compare them with its own. Through a system of rules and weightings, the clones could assign a compatibility value to each of the other clones they were grouped with – calculated in much the same way as described for emotional responses (see figure3 12.3 to 12.8).

By adding together all the compatibility values and dividing by the number of people at the meeting place, each clone would be able to calculate a value to indicate how generally compatible it was with the average of the whole group. Those clones whose compatibilities were furthest away from from the average were judged as "feeling out of place" at the meeting place and were moved on to a different meeting place where they might find clones that were more compatible.

By repeating this process with all the clones in every meeting place, moving the misfits around and repeating the whole sequence many times over, the population of clones in the simulation gradually sorted themselves out into compatible groupings: each clone moving around the meeting places until it found itself in a "comfortable" environment.

This is very similar to real life. People sample various social venues and choose to frequent those where they feel most attuned to the people present. This results in social meeting places becoming "cliquey": catering for people of a similar type.

The idea was that at the end of this computer processing, the owners of all the clones would be told which of the meeting places their clone had ended up in. The owners could then be given a specific date to visit the real life meeting place to meet the owners of all the other clones that had favoured that particular venue. In this way, the clones were making a rough selection based upon rules and weightings, but, the final choice was left to the humans.

This real life example illustrates the principle of a people space that is divided up into compatible groupings. Computer programs cannot make a final choice, but, they can do the donkey work, helping to create short lists of the most promising prospects from a large number of possible choices. This makes it possible for a human to concentrate upon only a relatively small number of people: a number low enough for the human brain to be capable of dealing with efficiently.

It was with this experiment in mind that I started to write the book "Magical A-Life Avatars". It led to a completely new way of thinking about the Internet.