In addition to the reasons Glorfindel mentioned, this would increase the surface area for abusing the reputation system. Let's say you have one main account and five 'puppets' that you use to support the main account.
The main account asks a question. The five puppets all up-vote that question and answer it, and then the main account up-votes all answers by the puppets.
- Increases the rep each puppet account can give to the main account (by voting for the question and answering it)
- Conceals puppet shenanigans to a broader extent; folks would begin seeing duplicative, often unnecessary answers as more commonplace
Currently, we stop this sort of behavior through automated scripts and users paying close attention to the quality of posts that gain a significant positive or negative score. Adding another layer of how reputation points can be earned would just make detecting attempts at gaming the system more complicated.
From an engagement perspective, finding ways to help new users earn more points early-on is a big goal; loss-aversion alone is a major reason for people to commit to maintaining and advancing a single account in good standing. But, we can't do things that risk people getting privileges before they've shown the breadth of experience needed to use them. So, the quest becomes more looking for small units of work new users could perform more regularly, and then incentivize that with a small amount of reputation points for each (such as the suggested edit system).
Increasing the reputation points earned for something we've already got, or adding multiple ways of getting reputation points for the same thing tends to be problematic, both from a 'game' and engineering standpoint.