Someone once explained it to me like this:

Let's say 1% of the population commits crimes.

Let's also, for the sake of argument, say all crimes are committed by black people.

Of the general population, 10% is black.

If you profile all black people, you will indeed profile the 1% of the population that commits crimes, but you will also profile 9% of the population that does not commit crimes.

And before you say "not the entirety of the 10% black population is profiled": maybe ask some black people about this. It's not 100% of the black population that gets profiled but certainly a very significant percentage.

And even if it's just half or even a quarter of the black population that you're profiling, you're still profiling a lot of innocent people.

This is the reason why profiling is bad.

(and note that in reality the figures are even worse).

The conclusion doesn't really follow from the premise. If all crime were committed by one racial group it would be madness not to use that information to target crime-fighting measures.

If there are credible reports of a white man committing murders it doesn't make any sense to waste resources listing mainly female Asian suspects, for example. There are also other similar issues. For example, if a crime happens on my street I expect to be treated as more of a suspect based only on my proximity.

The issue of racism is (and needs to be) grounded in things other than rational resource allocation.

The majority of crime in America is committed by white people, in terms of absolute numbers of arrests, value of stolen goods and funds, etc. By your logic, it would be most effective to target white people. This is without even taking into account that white suspects tend to avoid arrest, prosecution, conviction, and long jail sentences at higher rates than other racial groups (that is to say, when brought to a particular level of the justice system, they are less likely than others to proceed further).

However, it's clear that the vast majority of resources are devoted to black areas and poor areas, and to responding to directly "violent" crime. A mugging is more sensational than white collar crime, but from a purely utilitarian standpoint, I'm not sure it's rational to devote more resources to catching the mugger when, for example, there is widespread wage theft depressing incomes in the mugger's city. Which causes more damage in the long run?

A dead reply to this post expressed incredulity at the disparate treatment of Bernie Madoff and countless black convicts when comparing the scale of the damage they respectively caused. They illustrated this with the erroneous "factoid" that Madoff stole more than the value of all robberies and burglaries by black people in the country's history. This is probably untrue, due in no small part to the nature of his scheme. That said, the trillions in value lost during the Great Recession, which was the result of completely unpunished malfeasance by executives and regulators across the financial industry, and which stripped black families of roughly half their wealth, almost surely eclipses whatever figure one could come up with for the proposed above. I agree with the spirit of his comment.

Yes it does, because this is about profiling.

If all crime is committed by South Asian women, you're right that it would be madness not to use that information to target crime-fighting measures such as, for example, only looking for suspects that are female and have a South Asian suspects.

Profiling, however, where you preventively search if not outright harass people, is something entirely different: if one in ten South Asian women engage in crime, every time you preventively search and harass one of them, you have a ninety percent chance of searching and harassing someone who is innocent.

So it better to target 99% of the population that don't commit crimes rather that 9%? You haven't convinced me. Interesting that you chose black people for your example, its almost like you are profiling them.

it's better not to target any fraction of the population until you have some reason to suspect that an individual has committed a crime.