Interesting question. It's probably not too fair to pair people off at the beginning of the year and tell them their review score will be the same. But depending on how many people you have on the team, you could always try the approach they use on reality shows: constantly divide people up into groups and base some part of the score on being in the winning group. Say there are half a dozen big projects that can reasonably be divided up into different pieces of roughly equal difficulty with some objective basis for comparison (finishing date, overall quality, sales generated, etc). You're not going to kick people off the show if they are on the losing team -- even on these shows I think no more than one or two people are lost per episode anyway -- and many good people will be on the losing team and should not be penalized. But maybe you can structure it so that certain members of the losing team know their stock is going down, and the others are not penalized. Or maybe everybody on the winning team, and only certain people on the losing team, get "bonus" points for that round.
Again, I dislike reality shows and am not sure this would be a great working environment but maybe some ideas from it would be useful if you choose the right ones.

The potential benefit to something like this is that people would get constant, say at least bimonthly, feedback about how they are doing, and if you split up the teams a lot it might not be too unfair. The potential complaint is that people will blame the team rather than themselves, so there would need to be a strong individual component, and this would need to be a wildcard until the very end. The same reason students want to believe acing the final will let them ace the class is valid here too, psychologically. The last thing you want is people dragging their weight because they think it's over.

Hi Ohai, Solveig,

I'm really not a workplace person, and frankly Solveig, your suggested workplace environment scares me, so I'm not the person to give advice about this, but I thought I'd throw in a sports metaphor.
This grading on a curve is not done company wide, I assume. That is, if the company is at all big, it is probably divided to certain groups. Maybe you're the manager of one group but there are many such groups, or teams shall we call them.
What you need to convince them is that even if they are graded on a curve inside the team, that the teams are also graded on a curve, and to be the lower end of a winning team is better than the upper end of a losing one. Or mostly so. (I'm thinking of a basketball league).
Now as long as they feel that the major ranking is the group and the personal ranking is of second order, you're relatively safe.

There is a danger to this though. If you have a star player, and they feel the team is a losing one, they might want a trade so they can be on a winning team. That is, if you are not a sports fan, they will ask to be moved to a different team because of personal issues, or something like that. Here is where you need to convince them that first it is not completely team grade and only then personal grade, but that the top person can still be graded tops no matter which team they're on. And moreover, that a great player on a bad team can earn even more because they are more important to the team, blah blah blah. Second, to turn to their mentoring feelings and suggest how they can help the other people by being a good influence etc.

So my answer is essentially, as long as it is a big enough company, to move the question from your own team to a league issue.
Hope this helps.
 

When you say you want people to cooperate, I'm guessing that a lot of the problems will be with issues like who gets credit for something. If the final evaluation of people is based on their productivity, then spending an afternoon helping an officemate get a spreadsheet done or close a deal (or whatever the equivalent is in your office) is a wasted afternoon since only the officemate gets credit.

One solution is to explicitly reward people for helpful behavior, for instance, by making one of the categories on the yearly evaluation be "How much did this person contribute to a cooperative office environment?" If your team is too big for you to realistically be aware of every nice thing someone does or fails to do, you might even consider distributing worksheets once a month. List on the sheets the major criteria on which you will be evaluating your employees at the annual review, and ask them to list any major evidence in their favor in each category which has occurred in the past month. This will do two things. First, it reminds them what you think is important and encourages them to do things in each category, like helping their officemate for an afternoon. Second, it lets you correct misperceptions on the employees' part with a relatively short time delay. And of course, it gives you a paper trail.

It is common in the academic world to require that any grant application have a section on "benefit to society". These usually get worried over or laughed at by everyone but engineers and chemists, but the principle of asking people to answer this question is not a bad one.