I'm sorry Bikewer, but you are incorrect. We have seen increases and decreases in crime over the last century. Wikipedia's first graph, is concurrent with the plethora of graphs I've seen charting crime rates
http://en.wikipedia.org/wiki/Crime_in_the_United_States Please note the major increases up to the early 90's, though we have seen a marked decline since then. Murder does stay pretty constant but violent crimes had a huge boom.
And as I said, these are useful for noting trends within short periods, but the objective numbers nobody knows. To say nothing of the issues with reporting the actual crimes to the police, the FBI doesn't gather statistics it relies on individual law enforcement agencies to submit their data. For a couple of years, due to bureaucratic issues, FL didn't have any murders according the UCR. Most researchers extrapolate what they should have been and use that for statistical purposes. The FBI doesn't have the authority to require them to report, they bribe them with grants if they submit the paperwork. Which many officers have admitted then get filed shoddily or inaccurately (the city of chicago got caught fudging the numbers badly in the 90's by independent researchers). There isn't the capacity or the will to double check in most instances. Again, these are useful data sources but have very large limitations as I drill into my students heads.
I really would need to see a source on this because the data, even for murder (one of our 'best' crimes for actually getting data) is difficult to obtain and interpret. Yes there are crime records but the question of accuracy is a huge one. To say nothing of issues once you begin comparing country to country data. Sticking with murder, do honor killings count as murder or justifiable homicide? More importantly, how does that country count it? I do not mean to be rude or disrespectful but based on my reading in the field these macro level declines are based on very shakey data and analyses. It is not wise to put too much stock in them.
I need to read Pinker's book, it's on my extensive to read list, but I imagine he and I are going to have issues with data and interpretation as well. To be perfectly blunt, a large amount of current research on crime is crap because of crappy data. The further back you the data doesn't get any better.