Sheep, Wolves, and Employee Motivation
What a game in which virtual wolves chase virtual sheep can teach us about organizational culture
One of the better articles I’ve read in the past few years on the topic of motivation didn’t have much to do with motivating people. I recommend that you read the article (and, of course, give it some applause and good feedback) but I’ll summarize its main points here. It reports on a computer program that a couple of Chinese students wrote as an artificial intelligence project. The students developed a “sheep and wolves” game, in which they set up some rules for the wolves to chase and catch the sheep and the sheep to avoid capture. They also programmed in a few rocks to serve as hindrances to the wolves.
The Sheep and Wolves Computer Game
The programmers set up an incentive system for the wolves; a wolf got 10 points for each sheep it caught and lost one point each time it ran into a boulder. The programmers sought to motivate the wolves to catch the sheep quickly, so they deducted a tenth of a point for every second that went by as each wolf chased the sheep. The students then ran the “game” hundreds of thousands of times. Their intent was to see what the wolves learned. They were eager to see if the wolves moved up the “sheep catching” performance curve over time.
Before we check to see what the students found, let’s pause for a moment to notice the approach to motivation that the programmers deployed. They didn’t put a great deal of thought into their approach to motivation (as we’ll soon see). And why should they? After all, motivation is just common sense, right? We give prizes when the “subjects” do what we want and administer consequences when they don’t. After a while, the subjects will learn to do more of what we want them to and less of what we don’t.
Most managers seem to think all that’s needed to motivate their own employees would be to throw in some benefits and free coffee. The theories that most managers have about motivating employees are no more sophisticated than that developed by a couple of students designing an artificial intelligence program in which hungry wolves chase frightened sheep.
Let’s go back to the experiment to see just what the wolves learned while being well rewarded for catching sheep and suffering the consequences of not catching sheep. After 200,000 trials, the wolves learned…to commit suicide immediately by bashing into the rocks. It turned out that the sheep were difficult enough to catch such that the “logical” choice for the wolves was to limit the points they lost…by killing themselves.
You might think that my comparison is silly because no employee is going to commit suicide, virtual or otherwise, as a result of a poorly designed approach to motivation. And I’d agree, though one wonders if a poorly designed approach might lead to less dramatic methods of “opting out”….burn out, or even leaving the company. My broader point is that managers, like the student programmers, often approach motivation in a way that gets surprising and often undesired results.
“Will Work for Jackets”
I once worked for a coal company down in eastern Kentucky. Management told the miners that anyone who missed no workdays during the year would receive a jacket with the company logo on it. At the end of the year, jackets were handed out to the “deserving recipients”. They soon became aware, though, that one of their co-workers, who had missed only the contractually allowed bereavement days when his mother died, hadn’t received a jacket. As a group, the miners who had been awarded jackets laid them at the mine superintendent’s office door. A program intended to improve morale achieved exactly the opposite.
Steel Town Blues
A few years later, I worked for a steel-making firm in Cleveland. I trained and facilitated “quality circles”, teams of union employees that identified and developed solutions for problems in the plant. Members volunteered for the teams and a team member could “volunteer off” the team at any time for any reason without consequence. We never had trouble getting employees to volunteer for new teams and most stayed. Team members were paid for their time at meetings, but no other rewards were given for their ideas.
With good intentions, the industrial engineering department proposed that teams share in any cost savings that their ideas provided. My fellow trainers and I were against the idea, but management and labor liked it and it wasn’t a battle we were ready to fight. A program was implemented wherein a team could “earn” up to $60,000 to be distributed among team members.
The team awards very nearly destroyed what was a positive and productive program. Whereas the teams had once happily worked on safety and quality of work-life issues, they began to ignore those altogether because, of course, there was no “payout” for those ideas. Teams wouldn’t work on production improvement or cost-savings ideas unless they were able to determine, in advance, that their work would lead to a maximum payout. Why work on a problem that will net $10,000 for the team when another problem might yield a $60,000 payout? Teams that were awarded large payouts split up as arguments ensued over how the reward should be distributed among the members.
As teams spent more time searching for the big payout, they quit working on the many smaller problems that affected them and the organization. Soon, no more ideas were coming from the teams. We eventually redesigned the program to provide much smaller rewards.
Roads Paved with Good Intentions
The managers in these cases did not intend to create bad morale or threaten the existence of a successful quality circles initiative. Just as the student programmers did not intend to get their virtual wolves to bash their heads into the boulders more quickly. But that’s what happened.
The programmers, perhaps, can be forgiven for their failed “motivation project”. (They kept tweaking the algorithms and running the game until the wolves learned to catch the sheep fairly quickly. It took five generations of “tweaking” and running the game over two million times each generation, but, hey, those rocks and sheep can be tricky!)
I’m not as forgiving of managers who apply the same thinking to their employees. It’s especially frustrating when one considers that managers respond to their failures at developing motivational approaches by blaming the employees: “We reward them every which way we can think of but they still won’t do what we want them to! They must be stupid or lazy or some combination of the two.”
Keep Two Things In Mind
Managers need to keep two things in mind as they think about how to motivate employees:
- We’re all more strongly motivated by intrinsic rather than extrinsic rewards.
- That doesn’t mean extrinsic rewards and recognition aren’t important.
Employees will join teams, solve problems, participate in making decisions, provide information, work collaboratively with others, seek teaching and provide learning, and improve their own performance just because it’s fun and pleasing and rewarding to do so. Sadly, not many organizations provide a culture in which it's expected that employees are willing and eager to do those things.
If every supervisor and manager in your company went to one of their reports and asked, “How’s the work going? What gets in the way of your doing your best work? How can I help get that one of those things out of your way?”, your company would have more process improvement, waste reduction, cost-cutting ideas than it would know what to do with. Your employees are intrinsically motivated to talk about their jobs, their frustrations, and how to get rid of them. If all your company’s managers and supervisors went to an employee each week and asked that question and if the company followed up on the ideas and suggestions obtained, your firm would realize a strategic advantage in the marketplace.
Now is when the extrinsic rewards come in. You would thank each employee for his or her ideas. You would mention the ideas brought up by employees in department and organization meetings. You might put an employee’s picture along with a description of their idea on the department bulletin board.
Motivation isn’t created by managers who come up with prizes for “good behavior”. Your employees are already motivated to perform well. As managers, you just need to provide a culture in which that motivation can bloom and grow.
Sheep and Wolves Redux
Back to the “Sheep and Wolves” game…how should it have been designed? Here’s my modest proposal:
1. Don’t give or deduct any points to the wolves for anything at the start.
2. Set the big, bad wolves after the frightened sheep with instructions to catch the sheep as quickly as possible without bashing into the rocks.
3. Provide the wolves with measures of their team’s performance.
4. Between chases, have the wolves discuss their team’s performance. Which wolf caught the most sheep? What can that wolf teach the others? Which wolf caught her sheep most quickly? What can that wolf teach the others? Which wolf ran into rocks the least? What can that wolf teach the others? How can they, as a team, catch more sheep more quickly?
5. As the trials continue, give points to the team of wolves for its collective performance. Give points to individual wolves who collaborate well, teach other wolves, and come up with effective ideas.
Now that would be an interesting experiment, wouldn’t it?