The problem with saying that science is a purely cultural construct is that it doesn’t hold up.
While the ball-and-stick diagrams I use to demonstrate how a beta-lactam block bacterial cell wall formation is a cultural construct, the action of the beta-lactam is not. We could originally observe (objectively) the bacterial population being reduced by these compounds. Then, with physical chemistry we could elucidate the structure of the beta-lactam (which we then represented through a cultural construct). Then, with X-ray crystallography, we could observe that in the presence of beta-lactams, the crosslinking of the mucosaccharides did not occur. Next, because of our cultural representations, we could observe that the beta-lactam structure resembled the shape of the polypeptide used in the crosslinking reaction. Then, once again, through X-ray crystallography, we could observe that the enzyme responsible for the crosslinking reaction did, indeed, bind the beta-lactams and through various enzyme kinetics experiments, we could observe that the enzyme bound the beta-lactams with a higher degree of affinity than the original polypeptide.
Now, you can say that my ball-and-stick drawings are not the real molecule, I agree. You can say that my ball-and-stick drawings aren’t even a true representation of the real molecule. But you can’t say that my ball-and-stick drawings don’t let me accurately predict the mechanism of action of the molecule as observed by direct means. You cannot say that the molecule does not work. That is the point of science.
The sematics at this point become everything. When I say it is a scientific fact, I mean that it has 1) been observed repeatedly by the scientific community and 2) it works. Now, this is distinct from a ‘fact’ if you want to define ‘fact’ as an absolute representation of Truth. Science isn’t about Truth, it is about what works. Now, as a scientist, I believe that if something works it is a model of the Truth, but if you mistake the model for the actuality, well, you’ve got a problem.
To say the Earth is flat is an adequate model for short distances. If I throw a ball (or a spear) at you, over those distances the curvature of the Earth doesn’t play any role in my calculations. To say the Earth is a sphere is adequate for longer distances and it greatly simplifies calculations like rotational speed. To say the Earth is geodesic it allows for calculations over distances further still, as well as more accurate calculations wrt rotational speed and it introduces the ability to calculate ‘wobble’.
Now, all of these are models that reflect the actual shape of the Earth and for their task, they are sufficient.
Also, I take issue with the assertion that all facts are human constructs. Even if there were no observers, a beta-lactam would still inhibit mucosaccharide crosslinking. Before anyone observed Mars, it had the shape it does now (within reason, discounting things like erosion, ect.).
Now, that is an article of faith, but I see no reason not to believe it. There have been no instances where induction has been conclusively shown (models have changed, what the models reflect has not), so in my mind it is as foolish to believe in the possibility that reality is contingent upon an observer as it is to believe in a Divinity. Sure, there is no strong evidence against it, but at the same time, it is absurd and, above all, not useful.