I think modelling uncertainty is related to the subject that includes uncertain issues. There is some info (can be described as knowledge) about the subject but it is uncertain. It is not easily defined by conventional approach, so some uncertainty modelling approaches can be used.

the short answer to your question is it depends on how and what you used to model it.

Yes, certainly it is not the same modelling about a mineral property that about the history of a people. Because it is not ethically acceptable all the technically possible.

Usually in statistics, the model is a supposition about the distribution of a (random or of unknown frequencies) quantitative variable.

I think that call knowledge to uncertainties is a contradiction itself.

I think "knowledge" is to be informed of the truth. Different models about the same fact is uncertainty because there is only a true truth of the facts or the realities. A fit is a way of approximation, not a truth as to have or to discover a perfect knowledge about a thing.

Hi Mariano,

you talked about truth but what is the truth?

In agriculture (and more general in science) is better thinking about models. Models should be realistic (with uncertainty) to be able to generate knowledge.

A truth is, for example, a well measured datum. But an idea without rigour is not a truth.

Roberto, the model is not able of generating knowledge, but the data yes. The model allows to treate data, but I believe it is not knowledge by itself, but it is a supposition without rigour.

"Truth" is a religious concept, not a scientific one.

The way we see and understand the world, that is: the world itself (whatever it means to us) is the result of models. At the vary basis things like "time", "space", "energy" are concepts derived from models. There is nothing within science that justifies these concepts being "true" or "the truth". These are just concepts in (very!) useful models.

In some contexts, otherwise well-established models these three fundametals are not adequate, useful, or required (see quantum mechanics).

At this time we have (at least) two models about the nature of energy (being represented as matter or field) that are both very useful for different purposes, depending on the context - but which cannot be thought to be "true" at the same time: the particle and wave description. Again: the "truth" is not the question of science.

I find it (scientifically) funny when we discuss about things like "what is the true weight of this thing" (just an example of a "well-measured datum"; take the mass of an electron, or the weight of an apple, or the light absoption of a protein etc.) when we actually have no idea what this "thing" is, and what it is what we measure. Everything has a meaning only within the models we used to define these things we measure, and by how we measure them ("What is time? - Time is what we measure with a clock." [I think Einstein said this]).

The only connection I see between "truth" and science is that we may call a "truth" what is accepted by all subjects. This is somthing (inter-)subjective, but never objective. And so the "truth" is dynamic, it can change over time, change with new data and observations, change with modified or new models.

And "knowledge", I think, is a collection of beliefs. In a religios interpretation, these "beliefs" may be about a "truth". But in science, beliefs can be more useful of less useful. They can be based on more or less data, and on more or less expertise or intuition.

Jochen, if you do not believe in the truth in the science, of what do you like to convince us? You contradict you.

I do not talk about the concepts we are unable to measure.

In religion, we believe in truths, we do not make supposed truths of our beliefs.

Mariano,

I don't want to convince you. I wanted to post my poiint of view. For discussion.

I think that "truth" is a concept that is not helpful in scientific discussions. I think that science is about "usefulness" rather than about "truth". What "useful" measn is also matter of discussion and agreement in the scientific community. Practical applicability is one aspect, elegance and generalizability may be other aspects.

Jochen, the "truth" is not a mere concept, it is the way in the science and in the life. Without truth the "usefulness" would be a falsity.

How will you define "truth" based on evidence (empirical data) with a meaning beyond the data itself?

Doing so is metaphysics, philosophy, or religion. But not science.

In science we only have data. We can invent ideas or models that make observations more or less predictable (and thus useful), but all this is within the frame of data, not beyond. As I said one can see "truth" as a state of inter-subjective agreement on a model. For instance, electrons "truely exist" as particles only in the frame in which we measure data that relates these data to what we call "electron" in our particle model. There is overwhelming data and we all agree there, but still is the whole "existence" of electrons only a concept. Actually, there is no scientific question about the truth of electrons or their existence. It's just a model (a very useful one, though). The inter-subjective "truth" of their existence is based on the fact that we all know what we mean when we talk about them.

Jochen, if you say me that a man, as you talk, is only a scientist without religion, philosophy, etc. I believe that you have invented a monster, you have not recognised to a man.

The meaning beyond the data itself could be a known or unknown cause, the evidence is a true manifestation of a fact. A datum evidence is a manifestation of a fact which can have cause.

As unknowledge because to create a model from where there is not anything secure, it is uncertainty, that it is to say, unknowledge.

Random modelization substitutes an unknown reality for a supposed idea. And all idea can come from supposed realities or from nothing.

Modelisation of the unknown is to imagine what one does not know, this is unknowledgement.