Now that preseason pro gridiron games have started up again, I’ve been thinking about whether or not I want to play in the game-picking pool I’ve participated in over the past few seasons. I’ve had some interesting results in the pool – I’ve done extremely well in the regular season, but horribly during the playoffs. I know exactly why this is so, but I don’t think I can fix it.
My secret during the regular season is that I’m the only person in the group (usually 15-20 people) who understands that no one can actually pick football games. So I’ve developed an algorithm. It only consists of two rules, which are very simple to apply. Now, instead of agonising over who to pick – looking at point spreads, and stats, and the results of previous match-ups, doing research on who’s hot and who’s not, checking on weather conditions, thinking about whether games are being played on natural grass or astro-turf, and so on – it takes about 2 minutes of research and 3 minutes to write out my picks. And I don’t have to worry about whether or not my logic is right, or if there’s some hidden factor that I’ve forgotten to take into account. It’s much less stressful.
Using my algorithm, I’ve won the regular season pool two years in a row, and I’ve outperformed guys that know a whole lot more about football than I do who pick games for outfits like yahoo sports and espn. Why does this system work? The outcomes of games are genuinely uncertain. When face with uncertainty, we usually like to do things that make us feel in control. That’s why most people trying to pick winners put so much effort into research and number crunching. The problem is that results are pretty random. On average, the better team usually wins, but it’s often difficult to figure out which team is actually better. Having a good algorithm is actually an excellent strategy when you’re facing genuine uncertainty. A lot of people try to have a perfect week, where they pick every game correctly. My system will probably never do that – but on the other hand, it also won’t blow up. I’m very confident that my algorithm will win out over the course of 300 or so games each season. The people that are convinced that they know the game inevitably pick a few too many plausible upsets that don’t occur, or pick their favourite team to win an improbably game – but one way or another they are usually misled by the details of individual games.
However, my algorithm is close to worthless when the playoffs roll around. There are only 12 games in the playoffs. Over that small a number, the strength of the algorithm becomes a weakness – there’s no value in not blowing up, and the winner is the one that actually gambles and gets things right the most. So in the playoffs I get killed. All the analysis that other people do might help here, but so do judgement and luck (the person picking their favourite team all the way through when that team happens to win it all, for example).
What’s this got to do with innovation? A lot, actually. Big firms (or granting agencies, or state governments) with a lot of available resources, need a good innovation algorithm. They don’t need to pick individual projects that will win – they need processes in place that generate enough variety, that can experiment with the ideas relatively cheaply to see what works and what doesn’t, and that can amplify the ideas that are most promising. The focus needs to be on the process – not the individual cases. For these larger economic actors, innovation is the regular season, and over time, the best algorithm will win out.
However, if you’re in charge of an individual innovation project, or if you’re a small firm trying to execute one big idea, then it’s more like the playoffs. You need to be lucky, and you need to be passionate about supporting your particular idea. This requires a different skill set, and a different way of picturing the innovation process.
Over the long run, the people and firms that win through innovation are the ones that can do both – they have a good overall management process in place, but they also have people that can champion and execute individual ideas. Another key skill is to be able to identify when you need an algorithm and when you need judgement and luck. One of the reasons that a lot of big firms and government agencies get in trouble when they try to stimulate innovation is that even though they need an algorithm, the try the judgement and luck approach. This often misfires – leading to the truism that picking winners is bad policy.
The good news in innovation is that we can actually do some things ourselves to improve the odds in our favour – so outcomes aren’t quite as random as they are for football games. Having good processes is one of these things. Especially if we’re in a situation that calls for an algorithm.
(photo from flickr/ladybugbkt)