There’s something to keep in mind when trying to create a disruptive change: Most such attempts fail, and when one does work, things usually get worse before they get better. The old way has had some time to mature, and is probably now as good as it is going to get; the new way is going to require lots of learning and tweaking before it can match, and perhaps surpass, the performance of the old way.
So you’re not going to see much disruptive thinking in an impatient organization that punishes failure or one that looks only at short-term results. And it’s important to find ways to explore the new idea, without betting more than the organization can afford to lose.
One of the things that new students learn in artificial intelligence courses (or used to learn) is the idea of “hill-climbing”. Suppose you’re in unmapped terrain and want to get to the highest ground possible. All you can see is your immediate surroundings. So you figure out which direction is (locally) uphill, and then you take a step in that direction. Keep doing this and you will get to the top of some hill. At this point, a step in any direction makes things worse.
But this is only the top of some particular hill. There may be much higher hills around, but you can’t see them. The hill you have climbed depends on the place where you happen to have started. If somehow you can get a more global view of the terrain, you may be able to spot a mountain off to the southwest, and you may be able to see the path from here to there. If you can do that, it is worth going downhill for a while in order to reach this much better place.
But if you’re in a group, you have to be able to persuade the others that you really have seen that mountain — that it’s not a mirage or hallucination — and that there really is a viable path from here to there. Your bosses and colleagues may not be fully convinced, but perhaps have enough confidence in your vision to take a gamble on exploring in that direction. But are they prepared to stay the course if things get much worse before they get better? Are they even willing to fund a small exploratory expedition in that direction?
So disruption (or changing mountains) is a tricky business. Proceed with caution! But sometimes there is a big payoff for your organization, and maybe for you, and it’s worth the risk.. A well-run organization will find some way — some incentive structure — that can encourage people to look for big risks that are worth taking, and sometimes to take those risks
Scott E Fahlman is a Research Professor at Carnegie Mellon University, Pittsburgh, PA, USA. Scott is interested in Artificial Intelligence and its applications. ([email protected])
Would you like to comment?