It seems to me that the correct reasoning behind Occam's razor is that the more assumptions that a hypothesis must make the lower the prior probability must be. Likewise, the more specific a hypothesis is, the lower the prior probability. For example, the prior probability that "a red F150 will pass by my house within the next five minutes" is lower than the prior probability that "a motor vehicle of some sort will pass by my house within the next five minutes" for reasons that I think are fairly self-explanatory.