So maybe some racist white people might rate any area with black or Hispanic people as unsafe. Hopefully, if the area is actually not bad, some non-racist black or Hispanic people will rate it as such. The positive and negative ratings should cancel each other out in such cases. The areas that end up with negative ratings would have to be rated such by users of all ethnicities, which more or less rules out too strong…
Not subject to as much subjectivity. Drawback is that it may not be realtime up to date, but should be good enough.