Ants in the Net

Luís Ángel Fernández Hermana - @luisangelfh
7 August, 2018
Editorial: 225
Fecha de publicación original: 25 julio, 2000

People who stand still don’t go anywhere

Ants have been with us for 80 million years. Theirs is the story of the colony and colonisation’s unstoppable race for success. Their 10,000 species have populated almost every corner of the earth. Only some icy regions of the planet and an island or two have escaped them. After all, nobody’s perfect. And now they are about to colonise one of the last bastions left to their policy of expansionism: the artificial environment created by man, the virtual space of the Information Society. Whether they are robots which self-construct according to the tasks they assign themselves, or intelligent agents working collectively and capable of rationalising the use of the networks, the latest advances in computation have been using ants as their models. We are moving into the era of “swarm intelligence”.

Ants have been an endless source of inspiration to scientists and engineers (see the editorial “The virtual ant-hill”). Ronald Kube and Eric Bonabeau, leading expert in “ant engineering” at the Santa Fe Institute in New Mexico (USA) until recently, have been applying the organisational principles of the anthill and beehive to solving complex problems in the field of robotics. What is attractive about these insects is their distributed intelligence, their capacity for developing strategies and taking multiple decisions without any need for central control. In short, the metaphorical expression of the networked world. The hard-working nature of these creatures is what made them attractive to capitalist socio-biologists and state Socialism. Now, it is their ability to weave and deploy cooperative networks without maps or preconceived plans. This is the era of intellectual capital exercised within a network.

The way that ants successfully negotiate difficult tasks, such as carrying a leaf, transposed to complex artificial intelligence computing programmes means that robots which are completely idiotic on an individual basis, become capable of resolving complicated tasks collectively. The aim is not, however, to get these robots mowing the grass on a football field without going up into the stands, although that would be amazing anyway. The idea is that they are able to self-assemble using the most suitable parts for the task they are assigned to (or which they assign themselves), in the same way that ants form the best “work teams” for each mission.

Nevertheless, it is the Internet which is making “swarm intelligence” experts lick their lips in anticipation. The Net appears to be structured like an anthill, but nobody really knows what goes on inside it. The traffic of bits is conducted along hundreds and thousands of branches and routers which negotiate the best routes in real time, but it is not clear that this is done by using the resources available in the most efficient and sustainable way possible. As we all know — and have suffered ourselves– sudden bottlenecks crop up which slow down the speed of communication considerably, add to lack of security in the circulation of data and push user costs up. These are some of the problems that could be resolved by intelligent agents, and if they behaved like a real ant army from the same anthill there is no knowing what they could do.

Intelligent agents are programmes capable of independently fulfilling a wide range of tasks in unpredictable habitats. For example, by just giving them a series of ambiguous instructions about the information we need, they will speed through the Internet on their own, literally visiting hundreds of millions of pages at a vertiginous rate and then returning to our computer with a range of options that no search engine would be able to offer. Not content with gathering information directly related to the subject requested, they also use their own discretion and select material that might be of interest in some way to the person who asked for it.

The importance of these agents on the Net is growing all the time, although their development is still in its infancy. But in just a few years they have shot ahead in leaps and bounds comparable to the Net, both in their functions as well as possibilities, as was demonstrated at the Fourth International Conference on Autonomous Agents held in Barcelona last June. Presented there, amongst other marvels, were the Information Society ants, a colony inhabiting Marco Dorigo’s computer. This Italian scientist, who we will publish a long interview with in September, showed us a cultural vision of a biological landscape through the eyes of computation: algorithms capable of working together as though they were a colony of real ants.

The secret of the collective behaviour of these insects is the pheromones they secrete. Ants have as many as up to 39 different glands of these chemical substances, which help them communicate in a complex language and allow them to recognise one another, recruit individuals for specific tasks, sexually attract one another or send out alarm signals when suddenly attacked or faced with natural disasters, such as an unexpected moving shoe. The volatility of the pheromone is a decisive factor in “elaborating” particular phrases. The least volatile ones, for example, mark the ground for food routes and the shortest way back to the anthill. The most volatile are used to raise the alarm when there is danger.

This, but using computer programmes (intelligent agents) instead, is what Dorigo is doing and he is aiming it at the Internet. The idea is that from a computer hundreds and thousands of “virtual ants” go out, along the internal pathways of the Net. The pheromones in this case are numbers which the agents secrete wherever they go. Consequently, those that follow know who went before, what they did and what happened after that. In this way, amongst other things, data traffic can be regulated. Agents will indicate which routes are saturated, how often this happens and what the best strategy for getting rid of traffic jams in these areas are. Individually, each one of them is not much use. Collectively, they turn into a community of enormous strategic value. It is not at all surprising then that, when the idea first occurred to Dorigo in 1992, NATO immediately offered him all the help he wanted to develop it. Today it is a field of research with a personality (antability?) of its own. In a very short time, Information Society ants will reach us as part of the everyday facilities of any portal. And before we can blink, we’ll find we won’t be able to live without them.

Translation: Bridget King