A book which provides some good ideas for Software for Domain Experts is “Making the Most of Mess” by Emery Roe, a professor of catastrophic risk management in Berkeley.
The book presents the idea of a manager of any system (Dr Roe starts by talking about electricity control rooms, but it could be a school, police station, company, anything), by describing them as ‘mess managers’.
What he means is, in a role like this, you never get completely on top of anything. There’s no such thing as a perfect school on a perfect day, or a perfect company, or a perfect electricity control room.
But this isn’t the job of the professionals who run systems like this. Their job could be described better as keeping the overall system within acceptable parameters.
For example, for a school, it is manageable if a teacher turns up late once in a while, but if the problem gets bigger steps need to be taken to address it (I’m not sure what the right step is to address a problem of too many teachers arriving late in a school, but I’m pretty sure an expert school administrator or headteacher would know).
Dr Roe uses the analogy with electricity control room operators, who turn up to work every day not knowing if there will be a black-out. A blackout (cut off of electricity) will simply occur if the demand for electricity is more than a 5 per cent more than its supply. People can’t have electricity and the system breaks down.
There are various measures electricity operators can use to adjust supply of electricity (turning stations on or off, using stored supply) and in extreme situations they can adjust demand (asking people to use less, eg in a hot day when everyone wants air conditioning).
The control room operator can also keep track of times when they got close to blackouts. They can try to spot patterns (times when electricity consumption is likely to rise). Over time they build up expert knowledge of the system.
How could software potentially help here?
Automatic analytics tools could look for patterns in current consumption and compare them with previous patterns, and see what they led to
The software can use other data (eg outside temperature) together with electricity consumption data to help the expert manager build up a picture
The software can keep track of events where the system got close to a blackout, what the conditions were, and warn you if these systems are happening again.
The software could store ‘stories’ of what was happening on a certain day (eg a Royal Wedding so lots of people watching TV – or a major outage in a certain power plant) – to explain data in future – and give advice about what might happen when a similar event happens in future.
The software can’t replace the human expert but it can offer the expert a great deal of support.
I don’t know if anyone has done this in electricity control rooms – or for schools. It sounds interesting.