As we learn a game, we construct a mental model of how it works. Sometimes our model turns out to be inaccurate, and we have to modify it to take into account new information. A game should not refrain from presenting evidence that will disprove false models of it.
In a multiplayer game, some of the responsibility for disproving our models belongs to our opponents. If I believe that Rock is a superior choice and always pick it, you can select Paper to confront me with the falsity of this belief. But our opponents will have inaccuracies in their own models as well; Frank Lantz describes the process of finding and exploiting inaccuracies in your opponent's understanding of a game while they're simultaneously trying to exploit your own as maneuvering in donkeyspace.
Situations can arise, especially with board games where there are typically small local groups of players, where all players share some of the same false beliefs and thus are never confronted with their falsity - "groupthink". We all believe that X is the most efficient, so we all always pick X, so we never learn we're wrong - unless perhaps someone from a different group joins us and shows us how X can be beaten.
Now, this isn't necessarily a problem. If a group of players settles into something that's interesting to them, it doesn't matter if they're playing poorly on an absolute scale. But if groups falsely conclude that a game is unbalanced or uninteresting and give up on it before they really understand it, if groupthink is reducing what they get out of the game, then that's something for the designer to be concerned about.
A groupthink problem has come up in a game I'm working on. It's a two-player local ipad game, like Glitch Tank, so there are the same risks as with board games for groups of isolated players. Some people are getting it fine, others aren't - they settle on a style of play that mostly involves trying to prevent their opponent doing anything, while only accumulating points very slowly themselves. This is an unstable equilibrium; if one player focused more on scoring they could get ahead, but as long as both players are doing the same groupthink prevails. It's also a less fun way to play it. One tester has reported that the rounds last a bit too long - and he's right, I agree with him, with the way he's playing it does take too long. The game ends when one player reaches a target score, so if you're accumulating points more slowly than usual the game takes longer. But tuning the target score to make those games an appropriate length would make it far too short for my preferred style of play.
Not sure what I'll do. It might be solved by the measures I'm taking to try to communicate the scoring mechanics more clearly, or it might not; once someone believes they know something it's quite hard to change their mind. I've considered different ending conditions (time limit, number of balls collected), but nothing else really seems satisfactory. Possibly I'll just relax and ignore it, and try not to mind if some people decide they don't like the game based on false assumptions about it - I should be getting used to that by now.