Its still up for debate whether multilevel selection occured, though biologists now are developing (for monetary gain) cooperative shrimp, cooperative chickens, etc. that when raised in tight quarters do better (in total weight) than the 'selfish chickens'. Mike Wade and David Sloan Wilson are writing on this topic.

But altruism (in the truest sociobiologic sense termed reciprocal altruism), fits very well within a selfish gene framework of a social species. People 'do' altruistic acts, especially small ones, because their mind does not immediately do the payoff calculus, but in ancient times, living in small communities where everyone knew one another, helping, sharing and playing tit-for-tat had evolutionary advantages -so that altruism was a form of cooperation that had mutual benefits.

The modern world where people can pick up and move half a continent away and start over, sometimes short-circuits these altruistic tendencies, as oil at 18 cents a cup allows alot more people to 'cheat' this neural system designed for social give and take.

Though I dont want to derail this thread with a long debate with Ron Patterson ...;), I actually agree with you - that we did have historical bottlenecks where group selection was active - but still with individual selection at the core of long term evolution. But the majority of modern biologists, after George Williams blasted VC Lynne Edwards group selection hypotheses out of the water 30 years ago- conclusively do not believe in group selection (though it is making a comeback.)