When working on the Calabrio WFM solution lately we have had a lot of discussions on the complicated topic of user intervention and psychology of the user when running systems.
There is always a risk associated with doing things completely in the background without any user intervention. With software being able to make better decisions and quicker analysis we generally need less user intervention. But there is a problem with that. Many users today feel really confident with the work they’re doing. If you have read “Thinking, fast and slow” by Daniel Kahneman you’re familiar with the example of wine pricing. In several studies it showed that computer based prediction on only a few variables (weather, type of grape and soil) worked out better than experienced people working with pricing of wine by actually tasting it.
Even though the studies proved correct, the wine community was very reluctant to this fact. The general feeling was “you cannot put a true value on a wine you didn’t taste”.
Our discussions have so far been related to that example. We have a supported manual way of working with forecasts in Calabrio WFM right now. But with adding proved statistical forecast models and machine learning algorithms we could probably create even better forecasts – without the user’s direct input.
Maybe we’ll get there in the long run. But for now we will add more relevant information to the user. Compare it with a car. Twenty years ago we didn’t have all the assisting systems we have today. We need to go the same way with our software. Cars of today can park in a really tight parking spot for you. They can spot people or animals on the road ahead. They can see that you’re on your way to the wrong lane. But it takes a lot of trust to let the car drive you around on its own, even though it is possible today already.
What do you think could be the assisting systems in our software?”