...
This is a dramatic overstatement of the truth. The models have numerous parameters, most of which are only approximately known. Look at the error bars on total climate sensitivity -- they're enormous. The models are tuned over many runs to make them consistent with both science and history.
You are grossly exaggerating and have little to no direct experience with GCM's based on your statements. GCM's are PHYSICAL SIMULATIONS. You make it sound like it's all a bunch of random parameters fed to some simple equation, which is very very far from the truth. GCM's are typically comprised of a million+ lines of code of the densest physical equations you'll ever come across.
SOME things are parameterized because they have to be, not because they are not known. Our technology is limited. We can't run full scale GCM's at 1 meter resolution, so some shortcuts are needed.
Tuning is done to IMPROVE model output, not to CREATE model output. A plain untuned vanilla GCM will get you a good answer if you feed it good data. A tuned GCM will get you an even better answer if you feed it good data.
The IPCC has a good summary on GCMs, how they work, error analysis etc. But in short, a good portion of the errors happen well before the climate models are ever run. We don't have perfect information, either today or 200 years ago. That uncertainty by itself guarantees a respectable error margin even before the first equation is run. The rest of the errors come from the physics itself. Again, we don't have the technology to run a perfect simulation, even if we did have perfect information to give the models. So like every single other model out there, from chemistry to aerodynamics, there are errors.
The error margins on climate models aren't huge when considered against the backdrop of what it is that's trying to be modeled.
I'm not suggesting that the models are junk or that they should be thrown out with the bathwater. The GCMs are quite good, and have accurately predicted a number of important features of climate change. But the idea that they are boxes of pure applied physics is not remotely true. Like all models, they are approximations of reality. And where approximations are made, uncertainty creeps in. No new physics or magic is required. Simply not properly accounting for changes in parameters or relationships is enough to send one's model wildly astray. In a rapidly changing system, it is very difficult to remain on track as you model into the future. Much harder, in point of fact, than fitting a curve to some data.
And again, I suggest you seriously read up on GCM's. They ARE big boxes of applied physics. Yes, they are approximations. Every model is. But those approximations are driven by physics, not some arbitrary relationships between parameters. In fact, parameter relationships have been discovered by examining the model output and some scientist somewhere saying "Well now that's interesting."
GCMs are not a bunch of Excel calculations making graphs. The GEOS 5 model, for example, is 2 million+ lines of code of some of the most convoluted physics you'll ever see.
