Regarding a prison assuming a partner is trustworthy, etc.: This is not entirely true.
In the prisoner's dilemma and in all police matters, the police routinely lie to the prisoners. If the prisoners know this, it changes the dynamic.
The article suggests only a cynical view to the dilemma. If both prisoners trust each other and refuse to talk, the premise put forth must allow that in the case of weak evidence they both will walk.
In real life situations, a fair amount of the time the police don't have enough to convict either prisoner, which is why they use the prisoner's dilemma in the first place. Therefore, it is not necessarily a zero sum game.
Mutual shared goals and a strong bond of trust can result in a "win win" situation for the prisoners. The police are relying on the prisoners base instincts of fear, coupled with self preservation.
The real life percentage of the option I posit occurring may be low. This, however, can be explained by the fact that generally, the probability of people who commit crimes to be strongly trusting of each other is also low. They are likely to have trust issues in general. Big ones.
I propose an alternative situation: prisoners of war. If unit cohesion, training, and the soldiers' personal sense of honor are high, it is more likely that no one will talk i.e. "name, rank and serial number". There is no reason to believe giving truthful information will benefit any single individual since there is no reason to trust the captors in the first place, and soldiers fully understand that it will certainly harm the common goal: defeating the enemy. This flips the dilemma on its head.
In the final graph, cooperation in the automobile gridlock situation is put forth as the metaphor. This example posits many people who don't know or care about each other. In the prisoner's dilemma the issue of trust and mutual goals must be accepted as a viable strategy.