A while back (probably years now) Steve introduced me to the concept of ‘service entropy’. By this he was referring to this definition of entropy: ‘The tendency for all matter and energy in the universe to evolve toward a state of inert uniformity’. Both Steve and I are from an engineering background where entropy is evident in real things. Steve’s theory was that IT services operate the same way and drive down towards the ‘lowest common denominator’. If you heat something up, it doesn’t matter how hot, it will always cool down to the level of everything around it, unless you keep heating it. If we view an IT project as a heating process and the resulting service or application as the thing being heated up, what happens after the project has finished is the entropy process.
As I have pondered this and how it impacts IT infrastructures I have seen it at work all over the place. As with the heating process, it starts before the project has even finished. As soon as the project is exposed to the outside world it is already giving off heat. As soon as a project starts to be exposed to the cold light of day the ‘heat’ is leaving it. I’m using ‘heat’ here as a metaphor for the changes that the project is seeking to make in the environment in which it is being delivered.
As an architect who has a vision for a certain amount of ‘heat’ (change) to be produced by a project it is a huge challenge to consider how you can insulate the results of the project from entropy. Because entropy dissipates the energy within the ‘closed system’ the larger the ‘closed system’ the more energy you have to put in. In the context of an IT ‘closed system’ the extent of the change (heat) may be the whole organisation or may be a smaller group. In a ‘heat’ system it’s the insulation that makes it ‘closed’, for an IT system it’s probably the organisational construct that replaces the insulation and is the thing that keeps the ‘heat’ in (you have to remember though, that there is no perfect insulation)
Entropy explains why it is easier to impact a small organisation in a big way and to make a permanent change. Entropy explains why it is very difficult to make a small change in a large organisation and to make it stick. Entropy also explains why a large change in a large organisation is practically impossible without massive amounts of ‘heat’. I tend to be involved in making medium sized changes in large companies where organisations presume that because the change is ‘not huge’ that it should be easy. Most of the time these projects result in deploying technology and some process change, but never make the productivity increases that were expected. At this point it is generally the technology that is blamed rather than the lack of ‘heat’. It’s not normally the technology itself that generates the ‘heat’ though, it’s more to do with the way that the technology generates change and community.
The other thing with large systems is that it is impossible to apply the ‘heat’ uniformly. Some people will get direct ‘heat’ from the project, others will get it transmitted to them from another. Some people are conductors and some people are insulators. In some ways this lack of uniformity is worse than having a small amount of ‘heat’ everywhere. The reason for this is that when ‘hot’ meets ‘cold’ you generally get a reaction that cools down the ‘hot’ as much as it warms the ‘cold’. Take the example of calendaring and scheduling functions. These functions have been available for years (decades even) in corporate email services. Some organisation have managed to make a transition to making them uniform, but many organisations have not. As soon as there is any doubt (cold) about whether someone uses their calendar the value of all calendars is reduced.
There is a great danger in taking this metaphor to far, but I just want to take it one step further. You can undertake the heating process in one of two ways. You can either heat a small area to a very high temperature and hope that it will reach the edges of the system; or you can heat everything in a uniform way to the same temperature. The Internet has shown us that for large systems heating up a small area to a very high temperature works better than trying to heat up the whole. Take the example of Google Earth, this is a very ‘hot’ piece of technology and has generated a huge community. A few connected with that ‘heat’ and bit by bit the ‘heat’ is distributed. The trick is to keep the ‘heat’ going at the centre, which is what they have just done with the integration of National Geographic data. As a system becomes older it is more difficult to keep the heat going at the centre, and that is the challenge for Microsoft with Office 12 and Vista.
So how as architects do we resolve the entropy problem. For starters we spend as much time and attention on the insulation surrounding our heat source as we do on the heat source itself. In other words we try to understand where the heat will be taken out of the project and insulate the project from it. The other thing we do it to try and generate as much heat as possible knowing that some it will be dissipated.
Graham Chastney 