Should you quit while you're ahead? Here's what the science says

Have you ever achieved a personal best and then relaxed a bit afterwards, not pushing yourself further? If so, what are the implications of such behavior on your future performance? That’s the subject of a new study by Etan Green, a Wharton professor of operations, information and decisions, and Ashton Anderson, an assistant professor in computer science at the University of Toronto.

They recently joined the Knowledge@Wharton show, which airs on SiriusXM channel 111, to talk about the findings of their paper, “Personal Bests as Reference Points,” which was published this year in the Proceedings of the National Academy of Sciences.

An edited version of that conversation follows.

Knowledge@Wharton: Tell us a little bit about the research that you did.

Etan Green: I’m somebody who really disdains self-help books, but I should admit that this is a paper about how to improve yourself. There are many aspects of my life where I want to be better: I want to be a better runner; I want to be a better biker; I want to be better in the classroom — get better ratings [from students] — I want to be a better dad.

In many situations, we have quantitative performance measures that tell us how good we are at something. … These quantitative ratings help us think about how to develop things like goals to push us to become better, to improve ourselves.

[But] it’s a really tricky process to figure out what the right goal is to set for yourself. Goals are this double-edged sword. They’re double edged in the sense that they can be motivating. If you’re just short of a goal, you’re going to put in some effort to surpass it. But once you surpass it, it becomes demotivating, so you tend to take your foot off the gas pedal.

So we propose a solution to this problem. We propose the idea of a personal-best goal as something that can be an appropriately calibrated goal. If a personal-best [goal is] too easy, you’re going to surpass it quickly. It will be reset. If it’s too difficult, well, it’s never really too difficult in the sense that you’ve achieved it before, so it’s definitely within the realm of possibility. And so what we do is we show that this operates as a goal in the manner that I’ve just described — it’s motivating when you’re just short of it and demotivating when you just pass it. And we do this in the context of online chess.

We have this really beautiful dataset of more than 100 million online chess games played over 16 years. It’s incredibly rich data. We actually have players in this dataset who have … played 178,000 games or something like that over 16 years. [That means] playing like 30 games a day on average for 16 years.

Knowledge@Wharton: Of course, a chess game isn’t exactly a five-minute venture that people are going to be doing.

Green: This is blitz chess. The games last between six and 30 minutes. And this person, presumably, is playing many of these games simultaneously. So in chess you have a rating, this rating tracks how good you are. When you win a game your rating goes up. When you lose a game it goes down. And so we’re able to look at people’s trajectory and their ratings and see when their ratings approach the highest-ever rating they’ve achieved before.

As the rating approaches their highest-ever rating, what we see is that it looks like they put in a little bit more effort than they usually do. So their performance, their win rate, exceeds what we would expect given the opponents that they’re choosing to play as they approach that marker. But once they exceed it, we see this really large [effect] — a five percentage-point or a 20% increase — in the rate at which they quit or step aside. They don’t play for at least an hour. [So we see] this motivating element [to push harder] when you’re just short of your personal best, and this demotivating element [of quitting] whenever you pass it.

Knowledge@Wharton: Ashton, when you’re talking about somebody who is trying to reach a personal goal, there’s that level of competition from people who want to succeed. So if somebody has reached that goal and they take that break, do they lose that level of competition?

Ashton Anderson: That’s a really good point. It’s definitely something like Etan said, and that we see repeatedly in the data: When people reach this goal — which is actually a very difficult goal to reach, because by definition you’ve only hit this level once before — they’re so excited and they don’t want to drop down again, so they stop playing.

One thing that people have surmised is that this is maybe the wrong thing to do because you’re at your best level. This is when you’re playing your best chess. This is exactly when you should be playing. So it’s kind of a counter-intuitive finding.

Knowledge@Wharton: Part of this is also for people who take a break. Once they reach that goal, they seemingly have a concern of not scoring as high as they would have that last time.

Green: You definitely want to be at this place where your current ability is at your peak, at least measured by these ratings. This happens in a lot of domains. In baseball, if you’re batting, say, .301 in last game of the season, maybe you try for a walk or you ask your manager to take you out.

Knowledge@Wharton: I will give you the most historic one from baseball. [Boston Red Sox’s] Ted Williams back in the 1940s, the last player to hit .400 in a season, had to play a double header on the last day of the season. He had the opportunity to sit out the game and not play it and secure the last .400 batting average season in baseball history. And he still played and he still kept it. So he had that competitive nature. Ashton, and that’s something that I think a lot of people really build off of, correct?

Anderson: Definitely. One thing that would be interesting to study, something that we haven’t been able to yet, is to understand which players keep pursuing their personal best even after they’ve gotten there, and which players stop. You point to Ted Williams, who was maybe one of these people who had that competitive nature and that’s why he was such a great player.

Knowledge@Wharton: For those people that do take that time off, what do you think they lose in the interim?

Anderson: In the context of chess, your chess ability fluctuates. Your chess rating is an average measure of your skill. But from day-to-day and from week-to-week and month-to-month, your skill goes up and down. And so if you are not playing when your ability is the highest, then you’re potentially missing out on striking while the iron is hot. In other domains, like business, maybe something similar would be going on.

Knowledge@Wharton: Can you start to formulate ideas as to how this not only impacts people that are playing chess, but in other areas of society?

Anderson: The biggest takeaway from this work is that people are very motivated by [achieving] their personal best. It’s something that people want to strive towards. And it’s something that you’re only able to strive towards when your performance is quantifiable. So as performance becomes quantifiable in different domains, I think we’ll see this kind of behavior in more and more domains.

Green: One thing that we’re able to do as well is look at multiple goals. We basically compare the motivating and demotivating effects of personal best with round numbered ratings. So we can look at what happens when you’re just short of, say, rating a 1,500 or 1,600, and what happens when you just surpass it. And we see the same demotivating element when you surpass a round number that you quit, basically, at similar rates. But we don’t see the same motivating element in anticipation of a round number. … What this really points to is the importance of setting the right goals for yourself, goals that are appropriately difficult and well calibrated.

Knowledge@Wharton: Etan, what do you think are the implications of this type of research moving forward?

Green: This helps us think about how to formulate goals when we’re thinking about improving ourselves. It’s not sufficient just to say, “I want to be better at X. I want to be a better runner or I want to be a better biker; I want to be a better father; I want to be a better teacher.” You have to set goals and a plan of action for yourself. Goals are important because they provide the right motivation. The right goals are obviously important, as we show, because some goals are more motivating than others.

One other thing that comes out of this data is if you look at some of these players who have played a long time and the trajectory of their ratings, for the vast majority there’s very little improvement. This player who’s played 178,000 games over 16 years, the rating is basically fluctuating in a very narrow band. So they’re practicing, but maybe you can say they’re not training. There is some element beyond just setting the appropriate motivation and just doing something. It’s not sufficient to play 10,000 hours; you have to play 10,000 hours in the right way.

Knowledge@Wharton: How much of this do you think is just the personal mindset of the individual involved? A lot of times those goals are being set by managers.

Anderson: That’s right. I think what’s really nice about this work is that we show that people can be motivated by numbers that are generated by themselves. So you do definitely have these external goals — goals set by the manager and so on. But it would be feasible to also make more salient [personal goals] — like your best sales in a month or something — because this work shows that people are definitely motivated by numbers like that.

Green: I think it’s tricky to set goals for other people. If you set a blanket goal for, say, a sales force, you’re going to have some sales people who are demotivated by that goal because it’s very easy to achieve; they’ll achieve it relatively quickly and then they’ll stop putting in more effort. For other people, it’ll be too difficult for them and so they’ll say, “Well, I will never get there.” So it’s important to basically calibrate these goals for the individual, and something like a personal best does that naturally.

… It’s interesting when you think about sports. I’m a sports fan. I think about people who retired on top like [Denver Broncos’] John Elway. So we praised John Elway because he knew when to get out. Right? But we don’t know the counterfactual. Maybe he had a number of good years left.

Knowledge@Wharton: [Broncos’] Peyton Manning, too.

Green: He was arguably on the downhill even though he won a Super Bowl at the end of his career. But you know, maybe people like John Elway had a lot of years left. They could have played into their 40s like Tom Brady and played well, and we’re missing out on that because something about them said they should go out on top. And so there’s this weird thing.

You’d think if you were competitive, when you set a personal best, you’d want to set another one. But certainly John Elway is one of the most competitive people in the world, I’m sure, and his version of competitiveness is to say, “no, I’ve reached the apex and I’m going to leave at this point.” [Chicago Bulls’] Michael Jordan, the same way.

Knowledge@Wharton: Is there a next step to your research?

Green: We’re looking at two things. The first thing we’re looking at is what happens when you win versus when you lose. Is that motivating or is that demotivating, aside from where your rating is relative to the best-ever rating? And how does it vary based on how good of a player you are?

What we find — and this is definitely preliminary since the study’s not published yet — is that the better players basically get demotivated by losing. They start a session, they lose a game, they’re out. It’s the worst players that when they lose a game at the start of a session, they’re like, ‘OK, I’m going to try going forward.’ That was a little surprising to us.