Looking for Global Warming.

NASA at the Goddard Institute of Space Studies (GISS)  has posted their temperature data base.  It goes back to the invention of the thermometer back in the later 1600's.  I downloaded the whole thing some years ago.  The records were fixed length, 80 bytes long, no separators like comma's.  Clearly showing their origin on IBM punch cards.  I last used punch cards on a Raytheon job in back 1972.
  I wrote a C program to convert the ancient data format into something modern that Excel could read.  I plotted number of records vs date.  As expected, there are few records from the late 1600's.  The number grows over the years to a million or so.  Then in the early 1980's, a great weeding out happened, and the number of records per year drops to a third of its peak in 1980.  You have to wonder which reporting stations were dropped, with no explanation.  Where I live, it is 5 degrees cooler in summer and 5 degrees warmer in winter than it is down at the bottom of three mile hill in the village.  If a Franconia Notch reporting station was axed, it would increase global warming.  If a Franconia village reporting station, only three miles away, was axed it lower global warming.  When they axed two thirds of the reporting stations I wonder which ones got the axe.  The warmer stations or the colder stations? 
  Next I plotted the reported temperature data going back to the beginning in the late 1600's.  GISS furnished two data sets, a raw data set and a "corrected" data set.  The raw data set plotted out properly, a smooth line, obviously real data.  The "corrected" data set had a problem starting around 1860.  Data before 1860 was obviously bad, it had vertical jumps, bumps and discontinuities.  Just looking at the plot I could tell that something in the "corrected" data was wrong.
  So,  working with just the raw data, I  subtracted the average temperature from each year's temperature, yielding temperature rise or fall going all the way back the the late 1600s.  Temperature rise peaked back in 1990 and has been flat ever since.
   I believe in things you can measure, far more than I do computer models.