Corrupt mayors aren't the only reasons those cities languished. You lose banks, big box retailers at a time when people still shopped at them and have a rise in crime due to all of the jobs being shipped to the suburbs and your cities would go to hell in a handbasket too.

Corrupt mayors didn't help, but let's not act as if this was some sort of problem that was borne solely out of poor management.

People left the cities because of the rise in crime. Crime can always be prevented by a reasonably effective government. Read this book to get a sense of what happened in Brooklyn: http://www.amazon.com/Canarsie-Italians-Brooklyn-Against-Lib... The same story was repeating itself across the country.

oops, I meant to say presidential elections, sorry. You are correct though.

Interestingly, if there are more racist white voters than racist black voters, the racist black voters actually improve the quality of the decision because each racist black vote cancels out a racist white vote. This makes the best candidate more likely to win the election, because the best candidate has fewer votes to make up. (Unless the racism helps them, in which case it doesn't matter.)

What does "quality of the decision" mean? Does that concept even apply to a democracy?

If you are going to define the best decision as the decision made by the majority, then no. But let's say for the sake of argument that one leader can be better than another independent of the decision made by the majority.

Let's create a model to look at how this works. Say there are 30 voters. 6 of those voters are racist; 4 only vote for whites, 2 only vote for blacks. Of the remaining 24, all try to vote for the best candidate, but are only able to successfully determine the best candidate 52% of the time.

So there are 24 voters trying to choose the best candidate.

Without the racist black voters, the black candidate starts out at -4 votes.

With the racist black voters, the black candidate starts out at -2 votes.

With or without the racist black voters, each candidate needs 16 votes total to win. (A simple majority of the original 30.) Now let's assume the black and white candidates are equally likely to be the "best" candidate. Given that a voter trying to choose rationally is likely to choose correctly 52% of the time, we can see that the system with the racist black voters would choose the correct candidate a much higher percentage of the time if we were to run the simulation 10,000 times.

edit: I actually tried this in excel, and the results were quite interesting. 52% is way too low though, rather try 60% instead. What we see is that without racist black people, the best candidate essentially never wins.

Using your reasoning, wouldn't it also make sense for non-racists (of any race) to vote for the minority candidate not because he/she is better but rather to combat the racism of the majority?

My reasoning was descriptive, not prescriptive.

So a white racist vote = 'bad decision' but a black racist vote = 'less bad decision'?

Nah, white racist vote = bad decision in one direction, black racist vote = bad decision in the opposite direction. If – in a two party system – there are about the same on both sides those bad decisions cancel each other out.

Bad decision in the sense that race played a role where it shouldn’t. This works only in two-party systems like in the US.