"Science strives to explain things and make predictions."
Given.
"Usefulness is the main criterion for determining whether an idea will be adopted. "
Also given.
The problem with your argument is that you seem to misunderstand what scientists mean by useful. The explanation of an eclipse as being due to "magic" is not a useful explanation. What can you predict based on this "magic" hypothesis? To what further research will this lead? The "moon blocks out the sun" hypothesis, on the other hand is useful. If you know the orbit of the moon and earth well enough, you can predict when future eclipses will occur. It leads to the question of "why does the moon follow a regular path" which leads directly to a study of gravity.
However, to this day, we have no solid, absolute proof that the "moon blocks out the sun" hypothesis is true and the "magic" hypothesis is false. There is no observation that we have made to date that could possibly disprove the "magic" hypothesis. Furthermore, there exists no possible observation that would disprove the "magic" hypothesis, given a sufficiently powerful magician. You might say that the regularity of eclipses rules out the "magic" hypothesis, but that's untrue. Maybe the magician only performs his trick at certain times, and these times follow a regular pattern. Even a direct observation of the moon lying between the earth and the sun doesn't disprove the "magic" hypothesis. This could just be an illusion conjured by the magician.
This points out a further reason why the "magic" hypothesis is less useful than the "moon blocks out the sun" hypothesis. There's no way to test it. If the result of any "test" must always be that we can't reject the hypothesis, then we don't have a meaningful test. By contrast, the "sun blocks out the moon" hypothesis could, in principle, be shown to be false. All you would have to do is call an observer on the opposite side of the world at the time of the eclipse. If he were to look up and see a full moon, the "moon blocks out the sun" hypothesis would be falsified, since the moon cannot be in two locations simultaneously.
Hence we don't know for sure that our hypothesis that "the moon blocks out the sun" is true and the "magic" hypothesis is false. Since the magic hypothesis could be equally well applied to any phenomenon that is studied in science, we don't know that any of our scientific explanations are true. Hence truth is not the criterion for judging scientific ideas, but rather usefullness.