What I Learned While Running the New York Marathon

When?I stated on this blog that I was hoping to run the NY marathon in under four hours, I was hoping that my public commitment would spur me on.? And it did.? Sort of.? I ran under four hours – 3:54:59 to be precise – which I’m thrilled by.? So score one for Ian Ayres and?the value of public commitments.

But I think I could have run faster, and my public commitment is to blame.? You see, at mile 20, I realized that I was on target to beat my goal – “all” I had to do was run six more miles at a moderate clip, and I would be a sub-four hour marathoner.? The only thing that could prevent me from getting there was if I “hit the wall.”? So rather than running the last few miles hard, I ran them cautiously, doing whatever I could to prevent my legs from blowing up.? If a runner could play defense, then I was playing it.

There’s an economic lesson here.? I respond to incentives, and the incentive I set up was to beat a specific goal.? But that’s different than committing to run my best race.? And this led to the perverse result that I slowed down over the final miles, rather than running hard.? By picking a specific target, I distorted my incentives, and those incentives led to a good outcome (a sub-four-hour marathon), but prevented me from striving for an even better marathon.

So today the runner in me is left wondering if I could have beaten 3:50. And the economist in me is left wondering how to set incentives that don’t distort effort.


Gary

You merely balanced risk and reward in this marathon. Probably a wise thing at mile 20 with the goal in sight. Next time, set up a reward system that has a better payoff than a bit more personal satisfaction for clipping 5 minutes off the time. Next race, get pledges for charitable donations that are tied to the minutes you come in under four hours.

Sean Cooper

"And the economist in me is left wondering how to set incentives that don't distort effort."

Call me simplistic, but, at least in this kind of scenario, you can't. The expending of the effort was predicated on an incentive. As the incentive grows, the effort to reach the incentive will grow until the maximum effort is being output. However, because we are finite creatures, there will come a point where increased incentive will not yield any more effort. You've hit an asymptote. I'd argue that increasing the incentive even further would lead to a defeatist attitude and yield a declining curve on the right side of the asymptote.
But, because the expenditure of effort is predicated on the reward, the incentive will always be the determining factor for the amount of effort expended.

Tristan

I may have missed it, but did no one else really suggest that maybe the 4 hour goal just wasn't the best target to set?

I can see why you set that target, it's a nice even number, and is probably where most people seperate people who are just trying to finish from people who are trying to go at least moderately fast, and it was in the general range of your abilities. But the goal wasn't targeted to what time you thought (or hoped) you would run with any accuracy.

For example, your actual time was about 8mins 55 seconds a mile on average, if you thought that a good average pace for you on that day would be 8m 50s, your target would've been 3:53 or so.

Who knows how things would have gone if you'd tried to run faster, as many people pointed out conserving some energy to avoid a complete failure is probably a smart idea. Although I would think that taking the time to set a more accurate goal would lead to better performance, or would do at least as much good as a public commitment.

Read more...

Michael

Competition against formidable opponents provides that incentive. That's why world records are set during the olympics, never during practice.

There are some downsides to competitive incentives, like the possibility of unethical activity.

But in general, fairly matched and transparent competition will bring out the A game.

Peter Vanderwaart

In mathematics, we know you can only optimize one thing at a time. The lesson carries over into life in general. But in life, we sometimes don't know what the real objective function, except in hindsight.

Was there a risk of being delayed by a pack of people all trying to beat the magic 4 hour mark?

Erik Brynjolfsson

The 4-hour target wasn't wrong, it was the concept of a specific target. When you can adjust your effort over time, ANY all-or-nothing target will be suboptimal as long as there's any randomness in the outcome. Under reasonable assumptions, a linear incentive scheme (a constant reward for each one second reduction in time) is optimal.

See Holmstrom-Milgrom "Aggregation and linearity in the provision of intertemporal incentives" http://bit.ly/bQdlW1

VB in NV

I suggest running with/against a group with comparable abilities. The elite runners run to win, and if you run to win your race-within-a-race, it might bring out your best.

Of course, keeping track of this with a zillion others in the race might be impossible, but maybe you can create a sophisticated tracking device.

Renchix

I guess it is related to focusing more on the process than an outcome, as it was already mentioned.
So what if you change your initial goals and incentives already being in the race. If you feel you are going really well - upgrade your goal to 3:50, if in the middle of race you feel that in no way you are going to make it 4:00 then downgrade it to 4:10.

Johan Urh

The answer to you question: 'how to set incentives that don't distort effort.'

Answer: impossible

Reason: The purpose of creating an incentive is to distort effort. If the incentive you creat does not distort effort then what does it do...nothing...so might as well not have bothered with the incentive. To put it in other words its like asking 'how can I employ a person out of a pool of candidates without discriminating'.

--------------------------------------------------------------------------------

Johan Urh

@ 7
Hehehe.. at least when you 'invest' in a casino you know what the edge you are facing is... the banks didn't even know that

Apache_Chief

I love to run, but don't like running with a watch or ipod anymore. It seems there are so many people running in marathons, but they're so plugged in they're all in their own universe. Your post shows that you're marathon left you in certain ways unfulfilled. The reward (an excellent time) did not justify the effort (or as you put it - the lack of effort.).

You're a social creature - so run your next race with a friend and make sure you stick and finish together. The best rewards are shared experiences. Think of it this way, would you play golf by yourself?

Hannah H

My guess is that the best incentive scheme would be to have layers of incentives, the higher layers (miles 20-26) being dependent on the outcome of the first layers (miles 0-20). In the case of being "great" vs "good", a one-and-done static incentive might not be the best way to go.

And to be the best -- I guess you need some guts too.

Good job!

Mike

The sales incentive analogies (to me) don't apply to marathoning, unless you have a scheme where you're putting bonus already earned for the month at risk.

If you're at 110% for the month, what's the risk involved in going for 115%? In most bonus schemes, very little.

Marathoning, on the other hand, you're putting what you've already accumulated at risk with every step. You can be ahead of your goal time at 25 miles and still not finish.

To REALLY succeed at road racing, you have to take chances including the chance that you'll DNF. I'm sure there's a roulette analogy in there somewhere, but its not coming to me right away.

Michael O'Neill

I think Roger Bannister said the same thing about the 4 minute mile.

Anonumous

In the Navy, we had a fitness test. If you do X situps and X pushups and run this mile and a half in 13 minutes and 45 seconds, you pass. Except for showoffs, most people will save effort for the run, the last item. There was no incentive to do significantly better. In my case, people would bet between themselves if I'd make it. I used a strategy to pass the run. I would use my fast walking for a quarter mile then run and walk the rest. The 7+mph walking at the start effectively shortened the running phase enough to make it. Barely in 13 minutes and 30 seconds!