Sunday, May 30, 2010

Under-estimating the impact of incentives

When all you have is a hammer, everything looks like a nail. As far as organizational structures are concerned, I'm a man with a hammer. That hammer is none other than #1 on Charlie Munger's list of causes of human misjudgment : Under-recognition of the power of incentives.

Take for instance, Jean Louis Gassee's criticism of Microsoft's Steve Ballmer. Setting aside that Gassee failed to sell to Apple at a good price, and BeOS never did very well in the market, it's not clear that any of the things Gassee would have Ballmer do was really actionable.

Microsoft is a 60,000 person company. There's a very strict limit to how much one man, even a CEO can do to move a 60,000 person company. The reality is, when you're at the stratosphere at such a company, the only thing you can do is to really set up incentives so that people do what's good for the company by doing what's good for them.

Take Vista, for example. Vista broke one of the most important rules of Microsoft Windows development: it broke backwards compatibility. Now you can rationalize that Windows' code base is better as a result. But the whole rationale behind Windows was that you can buy any $25 piece of hardware at Fry's and it would work. Windows XP, for instance did that marvelously, and I still have Windows XP boxes attached to various pieces of hardware that won't work on any other operating system. The minute Vista broke that compatibility, a customer would have to buy all new hardware for his new computer. At which point, Apple could (and did!) come along and say, "Hey, why don't you buy my shiny machine? It looks cool, it scores points with members of the opposite sex, and it can also run Windows if you have to."

But presumably Microsoft knew all that! Why despite knowing that Vista's lack of compatibility would screw with Microsoft's revenue and dominance, did it do so? I asked a current Googler who was an ex-Microsoftie this question in 2006. His response was: "The new driver model? The one that broke all your devices? Well, you don't get your promotion to Staff Engineer for being someone who keeps it compatible with the old cruft. You get your promotion for designing a whole new piece of infrastructure that has huge impact on the world. Well, whoever did that got his promotion, and who cares if it tanked the company!" Ouch. People have argued that the new model is indeed more stable, but other techniques such as MicroReboots were also available. There really was no reason for Microsoft to take the risk of defection of customers to other operating systems.

One would think that such perverse incentive systems that can cause companies billions would be fixed, but my guess is that these incentive systems lie deep in the heart of the corporate culture: inventing new things will always be better rewarded than either making existing things run faster, or keeping things compatible, despite the latter two jobs usually being far harder than inventing a new subsystem out of whole cloth. And executives, even C level executives frequently still under-estimate the power of such incentive systems. As an example in late 2008, I had a conversation with a top executive at a well-known Silicon Valley company about what these perverse incentives were doing to his company. His response? "I don't believe it all comes down to incentives. After all, if you do good work and do good things, when you leave and work for other companies or when your friends leave and work for other companies, they'll remember you and bring you new opportunities, so you always have an incentive to do good work." When I heard that response I did not know what to say.

Months later, the company unveiled a "good citizenship award" internally. It was driven by the same popularity-contest-based incentive system that had already failed to promote good behavior. Not surprisingly, things have not changed as a result. When you see repeated examples of such behavior, it becomes much less of a surprise that startups without an incentive system other than handing out stock to everyone will continue to outperform the large organizations. Which again begs the question: Why the rush to get big?
Post a Comment