Applying Genetic Algorithms to e-business strategies

The technicalities of Genetic Algorithms are relatively unimportant because the underlying principle is so elegantly simple:

Periodically select winning combinations of objects or ideas and mate them to achieve a fast route to optimisation.

The vital key to the efficiency of this process is that genes are transferred during the mating not singly, but, in groups. By keeping certain combinations of genes together - across generations - allows interdependence and relationships to be taken into account. As this is a very abstract concept it may best be explained in terms of a football team.

It has often been observed that when a football player is transferred from one team to another his performance undergoes a radical change. This may be because the player's ability had been enhanced (or reduced) by the particular combination of other players he'd been playing with previously. A star player might become quite ineffective at his new club with a different combination of players. In some cases this may work the other way, a player who has not appeared to be playing well in one team might prove to be a star in another team, where the play of some of the other players especially suit his particular style.

The way in which nature transfers genes across generations is not singly, the transference is in groups. This can be likened to football players never being transferred as individuals, but, always together with some of the other players they have been playing with. This allows not only individual skills to be transferred, but, also the indefinable associations that enhance those skills.

By arranging random break points, different combinations of elements can be tried out without permanently losing the combinations that work most effectively together. In the example of the football team this is like being able to try out different combinations of player in different teams to discover which combinations work best.

Clearly, this form of optimisation has important implications for collaborative business activity. Just as genes can be shuffled around in a mating process, so, can collaborators. If this results in the more efficient creation of solutions, the entrepreneurs or prime movers who use this strategy may be at an appreciable advantage to those who rely on fixed teams.

If we now relate this back to the strategy of a film director, who makes a series of films, we might think of him as starting out with a randomly chosen assortment of people to fulfil all the roles that are necessary for the making of the film. It is likely that for the second film, the director will choose some of the people who worked on the first film to work on the second because they worked well together. The third film might see the director again choosing combinations of people from the second film to work on this new film, dropping those that didn't perform so well. It can easily be imagined how, over a series of films, an astute director can gradually improve the efficiency by which he can create a film by judiciously selecting the right combinations of people to collaborate with.

Directors do not always have to use their own judgement though. It is quite common for stars or other important people involved in a film to insist on certain personnel working on the film with them. They get to know what people they can work best with, those that bring out the best in their performance. In this way, like the biological mating process, the elements are put together not only singly, but also in groups.

Now map this scenario across to an e-business solution. An e-business solution provider can work the same way as a film director, choosing collaborators from a pool of freelancers. The first attempt may not bring about the optimum mix of collaborators, but, the second allows an opportunity to alter the mix to retain those who produced good work in the first attempt and replacing those who didn't. As the solution provider tries more attempts at providing a solution, the team of collaborators (freelancers) is likely to get more efficient.

Just like the director of a film, the solution provider need not have to specify every single collaborator. Collaborators might have worked successfully with other freelancer specialists on other projects and might recommend that they be brought in as well. In this way, a solution provider will work with groups of collaborators rather than individuals.

If the complex nature of an e-business environment is likened to that of a biological ecosystem, even more advantages of this system of working become apparent. Just like the strategy of a species, the approach to a solution can be arranged through a number of simultaneous approaches. Several alternative solutions can be tried out at the same time. This can be likened to a film director starting several movies simultaneously, or, an entrepreneur working on several different ideas at the same time. Groups of collaborators may be used, as and when necessary. Good ones can be reemployed, bad ones dropped and the collaborators transferred between separate approaches.

Seeing collaborators as groups of genes, also allows projects to be divided up into periods of time - with each period acting as a new generation - where successful groupings of genes go forward and the least successful drop out. This will enable solutions to evolve in much the same way as species or organisms evolve in the biological world -- able to adapt quickly to any changing conditions. The successful approaches can be considered as living through successive generations, while the unsuccessful approaches are allowed to die off. In this way, a best solution can emerge as a survivor.

Unpromising or dead-end approaches can be killed off at period ends, but, it may not be necessary to kill off all the collaborators who worked on them. They may be able to be absorbed into surviving approaches or used to embark on a new approach. In this way, an e-business solution provider can emulate the strategy of nature; retaining the best sets of genes in successive generations even when there are failures at the individual level.

For maximum efficiency, the time periods - the generations - that projects are broken up into, would need to be as short as possible. In this way, risks would be minimised and the chances of going widely off course would be lessened. Mistakes could be quickly rectified and the effects of sudden or unpredictable changes accommodated without too much disruption. In effect, this will give any strategy the same resilience and adaptability as a biological system.

Summarising:

In a relatively stable environment with a reasonable amount of predictability, a conventional planned approach with a managed team concentrating all effort on a single solution would be preferable.

In conditions of uncertainty and unpredictable change, many simultaneous approaches would be the best approach with periodic re-arrangements of people and ideas.