|
Edited on Sun Nov-29-09 02:17 PM by bemildred
I have a bachelors degree in math and a masters in computer science. You don't know squat about what I know about "how scientific reviews work". You just seem to have a "faith based" position that anyone that disagrees with you must be irrational or something. That seems like a very "incomplete model" to me.
In order:
"that's not my fault. If any data point used in making a mathematical model is shown to be falsified or just wrong then the entire model can no longer be accepted as accurate. That's just the way it works."
All models are "inaccurate", as I already pointed out, the question is whether they are useful or not.
It isn't my argument because it isn't an argument. The data appears to have been falsified, there is no argument on that. Anymore than saying it is 74 degrees outside my house is "my" argument, or an argument at all. It is a statement of fact.
I didn't say it was an argument, I said it was "bullshit". Calling it an argument would be generous. It is more like a windstorm of emotional accusations.
That's why they have an error estimate (which thusfar has been ignored). But models also require accurate information to form them otherwise they are worthless, correct? You can certainly come up with a model to explain any data set, even a falsified one. That doesn't mean it is applicable to real life.
Error estimates apply to observations, not models. Whether models are useful of not depends on whether they are useful predictors of observations. Models in general are not based on any real world data inputs at all, real world information may be used to design and test the models, but it is not the basis on which a model operates. Models operate by applying algorithms and computations to an internal state description, which is a data set of sorts, but not necessarily the result of some observation, it can be completely made up. Models do not observe. Sometimes the entire point of a model is that you can play with things that you can't do in the real world, e.g. "testing" weapons or theorizing about what happens when a supernova goes off. Data inputs to a model are generally about setting initial conditions, not deciding what to do next. Data sets do not "explain" models, data sets are used to test models. If you test a model with a bad data set, what you get is an invalid test of the model, not an invalid model.
Have a nice day.
|