III. All of us are often in situations that look like prisoner's dilemma games--where either party can gain by betraying the other. Yet most of the time, neither party betrays. Why do you think this is? You might consider some real world interaction you have experience of.
One possible answer is that we have structured the situation as a repeated prisoner's dilemma with an unknown number of plays, thus avoiding the unravelling problem described in the text. Obvious examples include business interactions, where one party can get a short term gain by (for example) delivering goods of lower quality than expected, and social interactions.
For a simple example of the latter, suppose two couples alternate inviting each other over for dinner each week. To make the problem interesting, suppose the trouble of hosting the dinner is larger than the pleasure, so each couple experiences a small net loss when it is host but a larger gain when it is guest. If everyone knew that one couple was leaving town in exactly seven weeks, and if there were no other considerations (such as reputation to third parties, or mechanisms for signing binding contracts), the game would unravel and the dinners would cease.
Another possible answer is that we have structured the situation so that even a single play is not really a prisoner's dilemma. One way is by commitment strategies, such as training outselves into having consciences. Another is by threats of retaliation--most obviously, against a criminal who testifies against his confederate.