Extreme variations are more likely the smaller your sample size is, considering America has a two party system they have a very small sample size causing it to be prone to extreme variation. Examples: No women candidate, no candidate that is not a politician, no candidate that favors alternate economic models, etc etc. This makes no "good" candidates according to a particular individual's view highly likely.
National primary turnout (as a whole or for any one party) is neither really meaningfully comparable across elections (because of changes in methodology of particular primaries, changes in timing between primaries, differences in the course of elections which change the perceived value of voting in later primaries, etc.) nor predictive of general election turnout (so pointing to primary turnout as evidence of a party's or its nominee's strength in the general election is meaningless.)
The targeted precision really is what's different. Your candidate can be N different single-issue candidates in a way that's just not possible without precise targeting.
The "echo chambers" explanation, in particular, seems wrong. Echo chambers always existed.
Also note that's "the typical (republican running in San Francisco)", as distinct from "(the typical republican) running in San Francisco", who would likely get even fewer votes.
That presumes candidates would campaign the same way and voters would vote the same way.
Look at the way primaries are run. Candidates put extraordinary effort into small states because that's where the party has decided it wants to have its candidates slug it out. They could instead choose larger states but they don't. Or the candidates could campaign in larger states but it would sink their candidacy --they campaign in order to extract maximum value. It's similar with presidential campaigns.
The alternative view is that CA, MA, and NY already get a Democrat candidate 95/270 or 35% of the way to the Presidency with just 20% of the population. Do those states really need a greater impact?
And there's "no difference" for the average candidate, not for outliers.
reply