>>How is consensus achieved? Deliberating until there's a statement to which all can agree. That's politics, perhaps democracy, but certainly not science.
>
>That is not how I view it. A scientific consensus should be representing a hypothesis on a scientific basis, free from politics.

But it frequently is politics, and that's the problem. I wouldn't mind if the phrase was used to mean "the most plausible hypothesis so far" or "mostly proven but not definitely". Unfortunately, it's mostly used to discredit the dissenters, leave them short of research funds and grants, depict them as quacks and producers of junk science.

Just look at the series of fiascos at East Anglia university and their methodology - yet anyone disproving them is labeled as "denier" and all the flaws in their methods don't matter, they claim to have the scientific consensus, which is in turn used to justify carbon tax, the greatest fraud since papal indulgences.

>If we look at the topic of discussion, there is an overwhelming set of evidence, observation, peer reviews, that point into the direction of the "consensus" that IMO you need VERY good and solid arguments arguing otherwise.

Just like religion shouldn't be a sacred cow, so the scientific consensus should always be taken with some doubt. The internal politics of the academia is also politics, and anyone claiming consensus is suspect of political reasons to stake that claim - that's the moment when I'd gladly take a look into what their opponents' theories are.

>Again, the value of scientific consensus is that it acts as a truth until proven otherwise, a basis which can be used for further investigation or elsewhere in science.

Exactly - it's a working hypothesis, not yet a discovered law of nature or a proven fact. I'm just against the consensus being taken as a fact, when it's still not one.

>Example: In construction, lots of empirical numbers are being used to calculate whether constructions are strong enough. Those numbers have never been proven to be scientifically correct nor accurate, but they are being used in practise without proven to be incorrect.

That's not science, that's empirically established boundaries. And it won't be science for a long time, simply because we can't calculate the behavior at a molecular level. It's too complex to establish any kind of causality, i.e. predict the exact behavior of materials.

I've found somewhere the classification of problems addressed by science by level of complexity. The simple stuff, involving a single-digit number of actors, was mostly solved before XIX century - solid mechanics, lenses, hydrostatics. The XX century brought the ability to solve for hundreds or thousands of actors, so we got computers, airplanes, drones, CERN. We still can't calculate the behavior of anything involving millions of actors, specially if they aren't a homogenous mass, so we deal with them statistically, i.e. reducing them to a number of actors we know how to deal with, so we follow a few variables, take a few dozen kinds of measurements. It does give some results - we have decent weather forecast, our roads aren't always congested, we even have medicine which manages to cure, but in most of these areas we don't even know what exactly is the whole set of inputs, what are all the factors that influence the outcome. And I guess it's mostly impossible to achieve, as you can't model anything at the atomic level, if the object modeled has more atoms than you got computing power. The computer to model snapping of fingers would either have to be as large as the known universe, or would take about six forevers to complete the calculation.