* Note: this is a backdated post from something I took notes on in the past
I like the concepts reblogged on Farnam Street regarding James March here, particularly the following points:
Most original ideas are bad ones. Those that are good, moreover, are only seen as such after a long learning period; they rarely are impressive when first tried out. As a result, an organization is likely to discourage both experimentation with deviant ideas and the people who come up with them, thereby depriving itself, in the name of efficient operation, of its main source of innovation.
I think the oft-cited examples of Xerox, Kodak, Blockbuster etc serve to illustrate the point. Xerox famously had prototypes of the mouse, graphic user interface, and the PC, but didn’t want to pursue them for fear of cannibalising its photocopying business. Kodak similarly invented the digital camera and was then ignored it. It’s hard to incentivise innovation and to know what to pursue.
Truthfully, genius is always recognized in hindsight, with the benefit of positive results in mind. We “cherrypick” the good results of divergent thinkers, but forget that we use the results to decide who’s a genius and who isn’t. Thus, tolerating divergent, genius-level thinking requires an ability to tolerate failure, loss, and change if it’s to be applied prospectively.
The difference between being outcome vs process focused.
In a simple model, a tortoise advances with a constant speed of 1 mile/hour while a hare runs at 5 miles/hour, but in each given 5-minute period a hare has a 90 percent chance of sleeping rather than running.
A tortoise will cover the mile of the test in one hour exactly and a hare will have only about an 11 percent chance of arriving faster (the probability that he will be awake for at least three of the 5-minute periods [since he needs 12 minutes to run a mile])
If there is a race between the tortoise and one hare, the probability that the hare will win is only 0.11. However, if there are 100 tortoises and 100 hares in the race, the probability that at least one hare will arrive before any tortoise (and thus the race will be won by a hare) is 1– ((0.89)^100), or greater than 0.9999.
I like the analogy here, but it implies that you need a large tolerance for failure. That may not be palatable in most organisations, so how should we structure incentives correctly?
2019-02-03 update: This twitter thread on the problems of having a ‘startup’ within a big company is worth reading through as a supplement to the above point.