Why do 'moderates' do better in general elections? The most plausible theory consistent with the empirical research is that elites in business and media are very effective at tanking non-moderates (and yes, they use money to do it) 
The Hall (2015) paper comes up a lot in these debates. What never comes up is the part of the paper on mechanisms. The most plausible mechanism is that moderates get more PAC (corporate, trade assn) money.
andrewbenjaminhall.com/Hall_APSR.pdf
andrewbenjaminhall.com/Hall_APSR.pdf
This theory is consistent with what we know about voter attitudes, too. On their own, few people vote based on, say, Medicare-for-All vs a public option. But elites (on all sides) use resources to frame policies, including in ways that interact with strongly held identities.
So it's curious that (especially around POTUS elections) social scientists sometimes characterize voter preferences as some immovable constraint on policy, and don't focus much on elite framing or the huge swings in mass policy support it generates over short periods of time
