You are on page 1of 3

What do you think is the most interesting thought proposed by Steven Johnson?

The most important and most interesting thing that the author espouses in the book Emergence is the idea of Emergence as a way of solving complex problems and as a basis of creating self-realising systems. He shows how it is possible using bottom up approach to create self-realising systems. The belief that there has to be a central monitor or pacemaker for creating intelligent machines has been challenged. The idea that small units, which work under simple rules, can perform complex tasks through interactions between themselves, which would otherwise seem impossible for each of them separately, seems rebellious to the general intuition. Slime mold cells which are simple single celled organism exhibit this emergent behaviour. In unfavourable weather they live as separate single celled bodies but when weather becomes favourable and plenty of food is available the individual cells start coming together to form a large common organism. They achieve this even though they have simple structure and no special controlling cells are there to control the coming together of the individual cells. This is quite confounding and the secrets behind this can help us understand ourselves better how billions of cells that constitute us are able to work coherently and how it endows us with our intelligence? It is one of the primary motivations which make the author come up with his idea of Emergence. Emergence can be defined as the movement from low level rules to higher level sophistication. This can be seen from the patterns in how a slime mold cells come together, how our neurons make our brain and how city neighbourhoods are formed. It is natural for us to look for pacemakers when we see repeated shapes and structures emerge out of apparent chaos as it provides the simplest explanation. Turings work on morphogenesis marked the beginning of shapes and patterns which form the basis for emergence which is used in lot of fields nowadays like in the recommendations that we receive on websites like Amazon and EBay. With the help of Gordons ant farm the author explains about emergent behaviour. In an ant colony we can see there are different functional segregations like harvesters, workers and queen. This despite the fact all those ants are otherwise alike and clearly there is no top down command and control system and yet they manage to form these colonies where they work in different groups towards a common purpose. It is really amazing and mind boggling and one explanation for this is emergence. All, these ants individually have no idea about their bigger world. The queen ant is not an authority figure and she is fed and cared for by worker ants. Worker ants are not picked by the queen and neither does the queen decide the tasks of the worker ants. Similarly, she does not create the group of harvester ants or assign its responsibilities or provide commands to it and yet all the ants seem to work in tandem. The explanation that the author provides is each of these ants produce pheromone trails and each of them can also sense the trails left by the other ants. Using these trails, the interactions between the neighbouring ants produce a self-organising behaviour. Since, there are lot of ants in a colony these interactions generate a common pattern and the anomalous behaviours of individuals tend to cancel out. Other social insects and animals behaviours can also be seen be explained on similar lines. On similar lines the author has explained how we humans form the city neighbourhoods where no individual has idea of the entire city at a time, instead he has idea of only his surroundings and through his interactions with his surroundings he shapes the neighbourhoods. Of course the interactions of humans with his surroundings are different because of his intelligence but yet at a higher level it is through his interactions that the neighbourhood shapes up.

The five pre-requisites, according to the author, for deriving macro-intelligence from local knowledge or for emergent systems are more is different, ignorance is useful, encourage random encounters, look for pattern in the signs and pay attention to your neighbours. Another interesting thing is that the macro-behaviour keeps evolving with time. This is again demonstrated with the analogy of ants colony and it is also true for city neighbourhoods formed by humans. Using the same theory he is able to explain how cells in our bodies differentiate when all of them have the same set of genes. An interesting thing to note here is that consciousness is not a pre-requisite for learning through emergent behaviour and therefore we can say even cities learn and evolve just like our bodys immune system learns to fight infections when it is exposed to pathogens. Author pops up the question whether web is learning as well, by connecting individual minds all across the world. Web by itself is highly disorganised space and there is no way a site knows which site links to it and the inverse is also true. So, the interaction and feedback is missing. However, with the coming of search engines and software like Alexa which try to arrange the sites and find interlinks between them and study user traffic pattern, it is slowly evolving as an emergent system which is at a still higher level than the city neighbourhoods. Feedback is also an important part of realising smarter systems. It helps in self-learning and self-organisation as displayed by the websites of Slashdot and EBay. It would enable smarter TV which can receive feeds from millions of channels and present programs according to your taste. Feedback allows the system to be adaptive by self-learning and self-organisation. Through a similar biofeedback mechanism our body is able to maintain homeostasis. It is the rules of feedback which determines how the system is going to self-organise itself and by tinkering with these rules we can change the way the system works. Resnicks slime mold simulation has been used by the author to demonstrate the emergent behaviour. The formation of mold and its organisations in this simulation is only directly controlled from the margins. At the start of the program no one knows where the mold is going to form but the rules under which it is to form are well defined and when these conditions are met the molds appear on the screen. The idea of emergence is used to an extent in the gaming industry for game development. It has also been used to generate emergent software using genetic algorithm, where smaller piece of code form part of the code gene pool. These undergo mutations across generations to emerge with the solutions. Of course here also the parameters defining the rules of mutations and selection of mutants are predefined and so it is not exactly as random as it in the Darwinian Theory but it is more pragmatic since there is no certainty if it works completely randomly that it will ever arrive at the solution in a reasonable timeframe. For a fully emergent system it would be impossible to predict the higher level behaviour in advance. Our ability of mind reading is another example of emergent behaviour. We have it in our genes but none of us are born ready with this ability and yet we learn it as we grow and this has been demonstrated by the experiments of Simon Baron-Cohen, Alan Leslie and Uta Frith on three and four year old kids. Even our ability of self-awareness comes from the ability to read status of other minds. Through several analogies, some of which has been touched upon in the above discussion, the author tries to portray his idea of Emergence. Emergence as he says is about how patterns and order is obtained at a macro-level from seemingly utter simplicity or randomness. The beauty is there is no central control and yet somehow through their interaction the micro-components project a more sophisticated behaviour at macro-level.

Does Johnsons book stimulate thought within you about what it takes to build intelligent technical systems?

Johnsons book definitely comes up with a nice theory which could explain how we derive our intelligence and awareness but he does not explicitly define intelligence anywhere in his book. The bottom up approach of emergence behaviour has been used to explain the behaviour of ants, moulds and humans as well. The self-awareness of humans and the way they come together to form city neighbourhoods and how these neighbourhoods seem to have their own intelligence has been explained. Whether he provides us with the information about what it takes to build intelligent systems depends largely on what we view as intelligent systems. If we talk of intelligence at human level which can learn and respond in an open ended manner then I think the information falls way short of requirements. Of course basic idea of emergence can explain this but we do not have any idea about the rules and parameters under which it will operate and moreover we cannot build a fully emergent system whose higher level output we can predict in advance. However, we can build systems which would have only limited intelligence with no open ended learning capability as has already been displayed in the development of games using the concepts of emergent systems. Games like SimCity utilise the idea of emergence, of course with certain limitations, to create a game where one can design and give shape to entire cities and the city seems to self-organise itself. The actions seem life like but all those are coded into the rules of the game by the programmer. From the example of the game we can see that sometimes it makes more sense to develop solutions whose higher level outcomes are more predictable because that keeps the interest of the user in mind and can target a specific set of user. In the absence of any such predictability and time limitations no one would be willing to use it. Again it would be questionable, if the system is fully Emergent, whether it can be qualified as intelligent if we cannot produce a predictable action what if it comes out with a bogus solution? As has been described by the author the idea of emergent systems can be used in the area of ecommerce to gather feedback from the users and accordingly tune the offerings to the user. Similarly, the TV programming can be customised. Internet forums and discussion boards also can be made smarter that can automatically block spammers and remove degrading posts. Thus we see it can help us achieve a certain degree of intelligence in the system but to achieve an open ended human like intelligence using the idea of Emergence as explained in the book seems farfetched and also undesirable since there would be no way to know in advance how the system is going to perform. In my opinion the author falls short of providing ideas about making intelligent systems if we mean human like intelligence but definitely the idea of emergence, as explained by him, is usable for creating less intelligent systems like in artificial intelligence.

You might also like