The Watchmaker: my fav argument

[polymath257]

if there is only one predictive theory, that is the one that is accepted [emphasis added], at least until another comes up. . . .
If there is no observational difference between the two, the one that makes more assumptions about non-observable entities is the one that is dis-favored [emphasis added].

Not to keep barraging your premises, but I see a lot of problems arising from this underlying Science as a Courtroom metaphor. So I would ask―accepted by who and for what purpose? Clearly, the diversity of competing theories that exists in science ought to show that nothing is ever accepted in such a manner, much less by everyone. Theories are just tools―they require our understanding not our credence.

As I said, if there is only one predictive theory, it is the only candidate. You are introducing a competitive theory, which I assume is also predictive.

Lastly, I would avoid evaluative terms like disfavorable, because why should anyone care what someone else doesn't find favorable? Useful theories can be inconvenient, complicated, and unfavorable.

A theory is useful if and only if it can make testable predictions and those predictions are verified by actual observations. That is the ultimate test of any scientific theory.

If you have *two* theories that manage to make predictions that are verified by observations, the simpler one is used *because* it is simpler. Having additional assumptions that cannot be tested makes a theory less useful because it cannot  be tested.