All scientific theories start off as ad hoc hypotheses that are then
subjected to experimental verification. Even Ptolemy's epicycycles were modifications to a successful theory that kept on showing problems as it was subjected to greater demands and scrutiny. Eventually, it had to be abandoned altogether, but in the meantime, it was good enough for engineering work, you could navigate ships and predict eclipses with it, and we still use its calendar.

A consistent theory (even if it is dead wrong) is useful if it suggests strategies for further research, even if it needs a major ovrhaul down the line. Even Newtonian physics served its purpose, and even today, is good enough for engineering work.

Out of all this dark matter and energy, strings, branes, quarks and other nonsense will eventually come a new synthesis, one that can be backed up in the lab or at the telescope. Even if it looks like nothing we have today, our 'epicycles' will have suggested the experiment that will show the way to the next level.

Newton, and Einstein did not emerge like magic from some experimentum crucis, the mathematics had all been done by others, worked out in other blind alleys long before, for other reasons, ready to be picked up as useful tools as soon the conceptual breakthrough made them necessary.

Science is proven by observation and experiment. But the ideas, the concepts, come straight out of the human brain, as if by magic. Faraday understood electricity and magnetism instinctively and intuitively before Maxwell actually wrote down his Equations. The scientific process is much more like art than it is like engineering.