How much does candidate quality matter?
By 
The Georgia senate election suggests the answer is “very little, but sometimes just enough.”
If we want to be pessimistic about things, we can focus on the fact that Herschel Walker — an almost indescribably terrible candidate by all conventional metrics regarding qualifications, such as being able to read, knowing the difference between the House and the Senate, not coercing the women he was beating up into abortions etc. — got almost exactly the same percentage of the vote in the runoff (48.6%) as the two other most recent Republican senate candidates (Kelly Loeffler, 49%; David Perdue, 49.4%).
More optimistically, this was 2022 not 2020, and the other Republican statewide candidates for governor, attorney general, and secretary of state all got between 52% and 53% of the vote, suggesting strongly that Walker’s absurd buffoonery of a campaign cost the Republicans about 4% of the vote, i.e., a deplorably small but ultimately decisive margin. Note that these other candidates were standard issue reactionary middle aged white guys, who kept the MAGA rhetoric mostly tamped down, didn’t seem to be severely brain damaged, and didn’t have a long line of women claiming that they had forced them to have abortions after beating them up and so forth.
So how did Walker get the GOP nomination? The answer of course is that Donald Trump thought it was a great idea for Herschel Walker, ignorant idiot populist and serial women abuser — the key difference between the two men being that Trump has not actually paid for any of the multiple abortions he’s coerced — to be the Republican nominee. Walker got 68% of the vote in the Republican primary, with the extremely distant runner-up being exactly the kind of boring middle-aged white reactionary that won all the other statewide elections in Georgia last month.
So thank you Donald Trump.
Speaking of which, Trump’s demand this weekend to terminate the Constitution has led to a particularly hilarious, for certain values of black humor, instantiation of the classic pattern in which Trump first says X, then is praised by his cult for doing so, then immediately turns around and blames the media for falsely reporting he said X, which leads to his cult pointing out that he didn’t say X, which then leads to Trump saying X again in slightly different words immediately afterwards.
How long this mashup of Kafka, Orwell, and Neil Postman is going to keep going is another one of those interesting questions, for certain values of interesting.
ETA: Andrew Gelman addresses the same question from a statistical point of view.