My current understanding is that, while the trend over hundreds of thousands of years has not (yet?) been significantly deviated from, the rate at which the temperature has been rising over the last 100 years has been unprecedented* (I concluded its not in fact unprecedented, see below).
I'm not entirely sure that the rate is in fact unprecedented, but I could certainly see it being that way. Look at the graphs at https://en.wikipedia.org/wiki/Temperature_record , it looks like:
- Between about 14 thousand years ago and 7 thousand years ago, the temperature rose by 8 degrees, which is 1.15 degrees per 1000 years on average.
- Between 132 thousand years ago and 124 thousand years ago, the temperature rose by about 14 degrees, which is a 1.75 degree per 1000 years on average.
- In the last 100 years, the temperature has increased on average about 1.1 degrees. So that's a good 11 degrees per thousand years, which is about 6 times what looks like one of the highest rates on the long term ice core record.
So because of the different lengths of times, and different scales, its a certainty that there was variance in the rate of temperature change between 132 thousand years ago and 124 thousand years ago. This means that there were definitely spans of 100 years in the those few thousand years in which the rate of increase was greater than the average.
So this lead me to looking at the actual data the graph on wikipedia was derived from. Here's the source: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/icecore/antarctica/epica_domec/ . Looking specifically at the spreadsheet called edc3deuttemp2007.xls , I see that indeed between between 132 thousand years ago and 124 thousand years ago, there are about 12 approximately hundred-year periods with greater than a 1.1 degree increase per hundred years. I took periods of three data points, which means they're not all the same exact length of time. In any case, this was from 76 3-data-point periods, so a full 15% of the time periods in that section of the data had greater temperature growth than we've seen in the last 100 years.
In fact, since 133 thousand years ago, there have been over 150 non-overlapping 100-year periods where the temperature has been rising faster than it has in the last 100 years. This is about 11% of the 100-year periods in the last 133 thousand years.
So in conclusion, based on this data, I have to conclude that the rate of temperature rise in the last 100 years is certainly not unprecedented, and thus my question still stands: why is this not compelling evidence that the temperature rise is largely not anthropogenic?
Here are the calculations I added to the spreadsheet to come to my conclusions: http://www.btetrud.com/files/edc3deuttemp2007%20-%20annotated%20with%20~100%20year%20temp%20diffs.xls . Note that the data is all in the second spreadsheet tab (the 'data' tab).