If we assume that the events are largely uncorrelated+ then we are multiplying the probabilities and our chance of wipe out are far lower.

+I would suggest that for situations where the probability of my machine and github's/bitbucket's servers being down due to the same event would be events of such magnitude that I would not be worried about my project anymore being more focused on basic survival...