Why do they keep changing the temperature databases?

Long time readers will recall the many rants here about how historical temperature records keep getting changed.  It's like the old joke from Soviet days: The future is always know; it's the past that keeps changing.  Well here we go again:

The surprise (when I plotted the source data myself rather than use NCDC’s tool) was how flat it was in the dust bowl heat of the 1930s.  I know that on the NWS NYC web site, they have archived raw monthly means back well into the 1800s. So I downloaded that and compared.


It was dramatically cooler in the NCDC v2.5 than the original data. This plot shows the differences between the original recorded temperature data at Central Park and the final adjusted data that NCDC presents to the public:



As is clearly evident, adjustments made the dust bowl period cooler, while post 1995 had no adjustments applied. This results in a temperature trend that is steeper because the past is cooler than the present. The only problem is that it isn’t what the data actually recorded then.

I don't know of any scientific field where it is not just considered acceptable, but considered normal to change the data after it has been recorded.  This is perhaps the strongest argument against paying any attention to what climate scientists say.

Note that all of the temperature readings before 1995 have been artificially reduced, generally by a degree to a degree and a quarter.  Remember, this is the rough amount that we've been told that the world has warmed over the course of the entire 20th Century.  Given this, it seems reasonable to question whether there has been any measurable warming over the last 100 years.  I'm not sure that I believe that, but the raw (unadjusted) data doesn't rule this out.