Are computer models scientific?

Science is a way of thinking about the world.  The ancients (Greeks and Romans) didn't have it.  It was invented during the middle ages, and put into words by Roger Bacon.  The key notion in science is that facts come from observation and experiment in the real world, and theories have to agree with facts.  For instance the ancient Greeks developed geometry, but it was entirely a construction of the mind.  Theorems were proved by mustering postulates and previously proven theorems into a neat array, never by actually measuring the geometric figure under consideration with a ruler.  Geometry, useful as it is, is not a science, it is pure mathematics, it's truth comes from thought alone, unburdened by any connection with the real world. 
   Science has been incredably powerful since its invention.  Science brought Western civilization from the dark ages to it's current position in the world.  You have to respect the scientific method, it has been enormously successful. 
   And now, we have people, claiming to be scientific, putting forth theories based on nothing but computer models.  These people are active in the "global warming" controversy, claiming that "global warming" is happening because they have programmed a "computer model" that predicts "global warming".  They imply that computer model results have the strength of real world observations and experiment. 
   A computer model is merely an ordinary computer program that calculates what might happen, given rules that relate the effect of inputs to the model to the output.  I have myself written computer models to predict the action of electronic circuits and mechanical systems.  All a model does is calculate answers based upon formulas and equations from the model writer. 
   If the model writer's formulas and equations are wrong, or incomplete, the models predictions will be wrong.  Surely everyone in these days has had some experience with the unreliability of computer programs.
   One test for completeness and accuracy of computer models is to set the inputs equal to some time in the past, run the model forward thru time, and see if the model's predictions match the actual real world situation. 
   The global warmer's models all pretty much fail that test.  None of them come any where close to matching today's weather when started in the past.  And, none of them predicted the past 19 years cessation of global warming.
   So when someone starts in on "global warming" ask them where they get their data.  When they say "computer models", beware.