Math is only as good as its premises, which are the foundation assumption behind the math. For example, video games use math. Some action or shooter games using the premise of infinite lives to make game play more enjoyable. Once you accept that premise then the math can be structured to fit the curve and make it happen, even if it is not real. Basically you run experiments, plot the data, draw the best curve and then find the math formula that expresses the curve. Now it is tool with predictive value. In the case of infinite lives we add that premise to the math and do beta testing, to see if the game play is better
If your premises are sound and you make a math model, this becomes a powerful and compact tool. But if your premises have no basis in reality we can still create math, and it will still be a loyal follower, but now will support a game or magic trick.
If you look at Casino math; probability and statistics, what is the natural source of its foundation of uncertainty? Is it natural, spiritual or man made? The math does not answer that question, but rather works under the assumption; premise, and then runs with it; experimental simulations, like infinite lives. Due to the utility of the math for helping to solve some problems, it creates its own reality; cart before the horse. This is why logic has to come first, so the basic premises are not unproven, or have no explanation, but rely on faith in the gods of dice and cards and lottery winners.
In statistical models the best curve for the experiments do not always touch all the data points; see below left. This is where margin of error and level of confidence are added. This stretches the data points into fuzzy dice, so we can pretend we have a genuine curve for the math model. To me the curve better fit into a 2-D oval than a 1-D line. What I see with that curve is that the experimental premises was nonsense, since it was not based on any sound reasoning, but from the POV of blindness in a black box. A more rational theory, leading to rational experiments will see trends, earlier than the experiments, and the data will become much more compact for a tighter line and full level of rational confidence; below right.
My favorite application of that tool is called risk. Risk fixates on the exceptions to the rule, and then that become the center of attention. All the curves or theories, only have to touch those few stray data points, while the preponderance of the data is ignored. Look what happened during COVID, where half the economy was shut down by ignoring the majority of the data due to the fear manipulation from partial bogeyman data. This is also why natural climate change is called being a denier; only partial data is needed for the magic trick.
This shows the low quality of the foundation premises that often support that math; man made climate change. I would get rid of that math model in science, since it allows for scamming and bad science. It can be used for exploratory uses, but in the end, when time to publish, if you cannot reason with the data, running a con under that fuzzy dice umbrella should be forbidden or subject to liability. Sometime you need to pull off the bandaid so we can let the wound have some air, instead of stay in the dark. The Golden Age of Science peaked about the 1920's. What we have today are mostly derivatives. That was time blind man's prophesy started to replace well thought out logic.
Galileo was connected to the Age of Reason Movement and was breaking away from the whims of the Gods. I agree with his statement but not with fuzzy dice or whims of the gods math that he never endorsed. People of his time ran the wrong experiments since they used faith in one size fits all and not reason.