rah

You've asked me another question which in fact doesn't seem to be related to the issue at hand. You haven't answered the question I asked. I am asking you why you think one would "not want to have a theory that contradicts what is well established". I can't see how you've gone from that statement to asking me about "theories" that cannot make predictions. When you originally wrote "you would not want", you saw some badness as a consequence of having "a theory that contradicts what is well established". To you, there was some negative thing that would result from having such a theory and this negative thing is to be avoided. What I want to know is: what is the negative thing? What is the badness?

There's some telling words here. There are a great number of non-trivial assumptions littered through this post and many of them seem to be unconscious. However, I won't go into them because I'm still waiting for an answer from ajb.

The problem is that there are people here, namely swansont and Klaynos, who do not entertain this.

I'm not entirely clear about what you're saying here. What things would agree within their error bounds?

Actually, this is exactly what swansont claims: That's your interpretation; you're assuming that you have a correct understanding of what ajb meant. I can interpret his words and assume that my interpretation is correct as well. However, I think it's best for ajb to say in his own words what his reasoning is.

That isn't what the thread begins with, that's the title of the thread. It's also a mischaracterisation of the issue at hand. I don't know who split this thread from its parent and gave it this title but they haven't titled it well. A better title would be "Why is it better to learn accepted ideas before anything else?" The main issue is learning "accepted" ideas, and the reasons why it is "better" to do that than something else. The something else could be learning about alternative theories but might not be.

ajb, you said before, and I asked You haven't answered that question. If you answer my question first, I'll be glad to answer yours.

So then you acknowledge that there is always room for a new model which has the same domain as an existing model, that disagrees with the existing model, but which agrees with observations?

You've used the word "probability" meaning that it isn't a certainty. You're saying you aren't certain that disagreement with Newtonian mechanics within the same domain also means disagreement with observations. The lack of certainty is a tacit admission that there exists the possibility of a model that has the same domain as Newtonian mechanics, disagrees with Newtonian mechanics but does not disagree with observations.

What do you mean by "safe"?

Here you are essentially saying "for all behaviours of nature correctly described by the model, the model is a correct description of nature". This is a tautology; it actually says nothing. Your argument does not involve comparing one model to another. The essence of your argument is "if a model is not logically consistent, then it cannot be a correct description of nature". (As an aside, it's worth noting that this is actually an assumption.)

What do you mean by "adequately"? swansont has stated that if a new model contradicts an existing model, the new model cannot be an accurate description of nature. This implies that the existing model was a complete description of nature. If the existing model is an incomplete description of nature then there is always the possibility of new observations that contradict the model. Hence, there is the possibility of a new model that contradicts the existing model but which accurately describes nature. To state that any new model which contradicts an existing model, cannot describe nature is the same as stating that the existing model is a complete description of nature. There is no complete description of nature. It is not possible to judge the correctness of a model's description of nature by comparing it to another model, unless the other model is a complete description of nature. It is only possible to judge the correctness of a model by comparing it to observations.

What has been observed is not necessarily everything that can be observed. The fact that a model can be used to make predictions does not mean that the model is necessarily a complete description of nature.

ajb used the phrase "what is well established", not "what has been observed". You've interpreted ajb's phrase to mean "what has been observed" but that's not necessarily correct. There are other interpretations of the phrase "what is well established" in this context.

I don't understand what you mean; suggestions about what?

Why not?

Why?

This argument is based on the assumption that the behaviour of nature is completely descibed by standard accepted physics. This is not the case. It is not possible to determine, a priori, whether any particular model will agree with how nature behaves. It doesn't make sense to state that a particular model "cannot possibly agree" with how nature behaves before testing it. It only makes sense to state that a particular model "cannot possibly agree" with standard accepted physics.

Why?

I must say I find this to be quite an extraordinary statement. Can I ask what, precisely, you meant by "better off"?