Altruism is the test case for Richard Dawkins' Selfish Gene hypothesis because he needs to explain a real-life behaviour in which individuals assist a sibling or cousin in raising young but, in doing so, fail to raise young of their own. The altruism gene has to be among the related offspring's genes, for which there are odds for and against, which means that, sometimes, it will fail, despite the effort. This makes it difficult to imagine how the initial mutation can propagate amongst the population, against dilution by the previous variant.
That said, evolutionary genetic changes sometimes do not come to the fore until the entire population is put under severe stress, by disease, famine, drought or other widespread crisis, resulting in a massive population crash. The family groups carrying the altruism gene will be the only ones able to gather enough food (plus mutual protection against predation) to rear one set of offspring, from the efforts of multiple adults. (eg with meerkats, only the dominant female breeds and any 'illicit' pregnancies result in ejection from the group).
Rather than propagating in competition with other genes, they end up in the whole population because only a small remnant population survived, thanks to carrying it. So evolution, to my mind, is at least partially dependent on catastrophe and mass death. (We never said it was a nice process).