Wednesday, March 17, 2010

Where is the intelligence?

When a system behaves intelligently, we may want to discover where the intelligence is located. In smart technical systems, we may imagine that there is some clever lump of software somewhere doing the smart bits. In intelligent organizations, we may imagine that there must be some clever people pulling the strings. So we may wish to take the system apart to discover how the system works.

But I prefer to see intelligence as a holistic property of the whole system, emerging from the interaction between all the parts of the system, rather than something that can be specifically located in certain components or subsystems.

Let's consider any team sport - baseball, cricket, or any kind of football. The top teams are successful not just because they have the most talented (and expensive) players, but because they respond quickly and fluently to the tactics of the opposing side, which involves all of the elements of organizational intelligence.

The team may have a captain, but in many sports this is merely an honorary position for the oldest or wealthiest member of the team and there may be no visible difference in behaviour between the captain and the other players. The collective intelligence of a good team is not just the eleven men on the field but also the man on the touchline in the suit; not just the 90 minutes on the field but the whole training regime that enables them to perform at that level. When our team plays another team, where is the "memory" located of the other team's recent performance? It may not be the players themselves who watch and analyse the TV footage; it may be the coaches who feed this knowledge into the pre-match training.

So does that mean the knowledge is somehow "owned" by the coaches (on behalf of the team as a whole) and then "transferred" to the players before the match? But that is a gross simplification - it makes it seem as if the players were merely blank slates for the coaching staff to write upon. All we can accurately say about this kind of knowledge is that it is derived from observation and interpretation of TV footage, contained somewhere in the whole-system, and somehow influences the behaviour of the whole-system. It's not about transferring knowledge from one subsystem to another, but about communication and collaboration that results in the whole-system using the knowledge effectively.

Describing intelligence in this way allows us to focus on a series of critical questions - how does the team (as embodied whole-system) use knowledge intelligently, what kinds of signal does it pay attention to, how does it learn from past experience, how are new tactics and counter-tactics developed, and so on.

Those primary questions raise some secondary questions about the kinds of organizational structure and team culture in which this kind of intelligence may flourish, as well as the tools, platforms and other mechanisms that may be helpful, but we must always be wary of simplistic means-ends thinking - implement this regime and install these practices, and these outcomes will magically ensue.

for VPEC-T, see comments below


  1. But I prefer to see intelligence as a holistic property of the whole system, emerging from the interaction between all the parts of the system, rather than something that can be specifically located in certain components or subsystems.

    I agree.

    Just as an individual person's "intelligence" is made up of both conscious and unconscious parts, I think that an organization's intelligence consists, at least in part, of these two aspects as well.

    To use a team sports example, consider (American) football: the quarterback is the conscious "decision maker" for the offense. Once the decision is made and the play put into action, the actions of the team are "unconscious" in the sense that the quarterback has no direct or indirect control. (Obviously, each individual on the team is conscious of their own action.)

  2. But presumably the quarterback is conscious of whether the team-as-a-whole is executing the play effectively. And are not the other members of the team also conscious of this to some extent?

    I'm not familiar with American football, but in most sports there are signals that pass between team-mates during play. Some of these signals are conscious - raised arms, shouts - while others may be more unconscious or instinctive.

    So I think there are two separate distinctions here, both of which are important.

    - planned/controlled versus situated/emergent
    - perceived versus unconscious

    The division between conscious and unconscious is very interesting. Sometimes we can improve the performance of the team by increasing the amount of conscious communication and coordination. When my sons played soccer, the coaches always encouraged them to communicate more. On the other hand, a team that has trained closely together may be able to synchronize moves instinctively - unconscious communication may sometimes be faster, as well as harder for the opposition to pick up.

  3. Continuing these themes - and into a sport that even fewer people have experience with - racing sailing boats and the behaviours of crews. It is interesting to look at the evolution of a crew over time.

    First we can put together a crew where the members don't yet trust each other. Here the most experienced members will probably attempt to do too much. Leaving the least experienced members under utilized and possibly frustrated.

    Then as the crew bonds - interpersonal context and trust build and the crew starts to work well in unison - especially during the happy flow or standard maneuvers. So at this stage the knowledge is institutionalized. A tactician will make a decision to "tack" the crew knows what to do and there is no central orchestration. There may be an observer, however seeing if there are ways that the choreography can be improved. At this stage, if all hell breaks loose (equipment breaks, someone falls over the side, or something like that), the collective knowledge breaks down and again experience comes into play.

    In the next stage, the crew has built enough trust, enough confidence in each other, so you know that"someone has your back" if for some reason you are out of position. So if you can't do your task because you have made a decision to leave your post to attend to an emergency (saving someone's life, preventing a tear in an expensive sail, or whatever), you know that your job will be taken care of.

    So yes there are signals that are exchanged, but the "knowledge" is in shared confidence, shared beliefs, shared values, and Trust.

    Sounds a bit like VPEC-T thinking!

    So if we look at the notion of intelligence in systems, it might be helpful to look at that intelligence through the lenses of VPEC-T. Or maybe look at each lens and ask what properties viewed through that lens make a difference.

    The only one I will touch on is T - Trust. At some level, the components of an "intelligent" system must have some level of trust. Of course if we think about a complex adaptive system - like a swarm of bees, we need to ask ourselves what trust really means. Is it a blind following of the rules? Compare that kind of system with that of my racing crew above. In the racing situation, it is a confidence (trust) that each crew member will do what is needed.

    perhaps this is a little strained, because anthropomorphism of systems is somewhat flawed - but I do like to look at how things work in the physical world and look for analogues in the systems world.

  4. thanks Chris

    continuing the VPEC-T theme, let's look at V for value and P for policy

    I venture to suggest that everyone on the boat shares the belief that human life comes even higher than winning the race. But the cost of an expensive sail may not matter equally to all the crew members. Different values. So when a sail starts to tear at a critical point in the race, we can't necessarily trust that everyone will automatically follow the same policy.

    In this context, we can think of policy as a device for making intelligent decisions without having to work everything out from first principles in real-time - like a habitual response. The role of intelligence before the race includes designing, agreeing and instilling a set of instant responses to certain events; the role of intelligence during the race includes dealing with aspects of the situation that cannot be preprogrammed and recognizing situations that demand something different to the trained response.