How to fix the City to Surf: self-seeding incentive realignment

One of my two super-sisters (here’s the other one, for your YouTube enjoyment) has started her City to Surf training already, and has taken to it with a vengeance. Currently I hold the family record for that particular race, but as with most things, it probably won’t be long before one of my tougher, better-looking siblings wrests it from my grasp.

Self-seeding into the hottest group!Anyway, hearing about her exploits got me thinking about a problem I keep runni encountering in any race context, be it C2S, Parkrun, or whatever – misalignment of the incentive to self-seed accurately. Self-seeding refers to the pre-race ordering of participants relative to the starting line. Ideally, the fastest runners would start at the front, then the next fastest, and so on until those who are more socially- than athletically-inclined end up at the back, where they are usually happiest. If everyone self-seeded perfectly, then no-one would ever be stuck running behind a slower runner, and the whole pack would move with heart-warming smoothness.

Here’s the kicker, though: there’s (currently – stay tuned) no disadvantage in trying to get closer to the front of the pack than someone of your speed should be. Being overtaken by faster runners is no skin off your nose (apart from some potential ego damage, perhaps), and is certainly better than being stuck too far back and having to dodge slower-moving human obstacles. This creates the incentive to game the system, and leads to an inefficient outcome where a non-trivial number of runners will be dodging people regularly during the race.* To combat this, race participants who want to secure a decent spot in the line-up must either break social conventions by shoving and/or ‘subtly’ maneuvering around people, or by turning up earlier than they would otherwise.

So how do we fix this problem? For C2S with its high-tech timing technology, I have a solution that would be fairly easy to implement, although it would require a solid effort to convince participants of its utility. Runners are a fairly intransigent lot, have you noticed?

The solution: penalise people for the number of runners who started behind them (i.e. whom they self-seeded ahead of) who finish in front. There’s always plenty of uncertainty in how people will perform in a specific situation, so a generous penalty-free threshold could be allowed – say, 20 people who started behind you could finish in front of you before you were penalised. After that, though, start adding on time for each person you’ve slowed down by poor self-seeding. The penalty should start light but increase at an increasing rate, so that the most egregious offenders are punished the harshest. That way you bear some of the cost from your choices, and hence will have an incentive to err on the side of caution when choosing where in the pack you’ll start the race.

What do you think? Have an alternative solution to the problem? Tell us about it in the comments!


* I am aware of the existing C2S seeding system, which seems to work quite well overall. The problems I’ve described still exist within each of the starting packs, however, and can make a pretty big difference – take my word for it, speaking as someone who was stuck at the back of the red group in 2013…