The prisoner's dilemma is a common building block of game theory, which is often used in business strategy to describe potential competitive responses or explain company actions. It's so named because of the story that was used to illustrate its problem. Imagine two fugitives that have robbed a bank and were just picked up by the police. Each man is taken into a separate room, and the police make the following offer to each: if they confess to the crime and help to implicate their partner, they'll receive an extremely lenient sentence. However, if they stay quiet and their partner implicates them, they'll receive the maximum penalty - while their partner gets off scot-free. Of course, if both confess, the evidence they provide isn't particularly useful, and they'll each go to trial, ending up with a sentence somewhere in the middle. What should each man do?
The most logical way for a prisoner to look at this problem is to decide what their best response is, given the response that their counterpart might take. Typically, for simplicity, this is presented as a 2-by-2 matrix:
The numbers in parentheses represent the number of years that each prisoner would receive in each scenario, in the order (Prisoner 2, Prisoner 1). So, for example, if Prisoner 1 opted for a "Don't Tell" strategy while Prisoner 2 decides to tell, the payoff is (0, -20) – Prisoner 2 receives no punishment, while Prisoner 1 goes to prison for 20 years. The specific numbers aren't important, but modeling the scenario described in the story is.
The "trick" comes when you think about exactly what each prisoner should do. Say you're prisoner 1, and you think your partner is likely to snitch; your best response, then, is to tell yourself (payoff of -10 instead of -20). On the other hand, if you think he's likely to stay silent, you should still tell, since you'll get away free rather than taking a year in prison if you stay quiet yourself. Your "dominant strategy" is to tell. Unfortunately for our prisoners, though, your partner is facing the same payoffs. As a result, you both end up telling and end up with a ten year sentence apiece – even though by cooperating, you'd both be better off. (The game falls apart if you can actually speak with your counterpart throughout the game, and have some way of punishing them for poor performance... for example, if you play the game repeatedly.)
So what does this have to do with Guitar Hero and Rock Band? Well, both Activision and Harmonix are facing essentially the same problem.
"In theory, they could cooperate: consumers would be able to buy both games and freely swap between them as desired." |
The payoffs, then, are still from the companies' perspective. Presumably, sharing is going to expand the overall market for music games and allow each of the firms to gain some additional customers and revenue. Similarly, if one shares and the other doesn't, the more generous firm is going to be at a disadvantage as customers flock to the more feature-rich game. And as a result, the equilibrium is for neither to share – they're paralyzed by the lack of trust they have relative to one another. And gamers are generally worse off, even though the real "game" is being played by Activision and Harmonix.
"There's little point in conceding the market prematurely to Harmonix." |
As co-editors of A Link To The Future, Geoff and Jeff like to discuss, among many other topics, the business aspects of gaming. Game companies often make decisions that on their face appear baffling, or even infuriating, to many gamers. Yet when you think hard about them from the company's perspective, many other decisions are eminently sensible, or at least appeared to be so based on the conditions at the time those choices were made. Our goal with this column is to start a conversation about just those topics. While neither Geoff nor Jeff are employed in the game industry, they do have professional backgrounds that are relevant to the discussion. More to the point, they don't claim to have all the answers -- but this is a conversation worth having. You can reach them at
