I simply don't believe "what is good for people" is sufficient for a system of morality. You need to be able to order things in terms of "goodness". And on top of that, many moral and ethical choices will involve figuring out how to weight each individual's "well-being" in relation to the whole. These two points in the model are simply points where you basically make shit up while only partially constrained by empiricism or reason. And these are the points that I would consider closest to what morality ought to handle. And in some sense, when we do science, a pretty good chunk of our models are "shit we made up" and was chosen that way partly in order to make things easy enough to work with and partly because we just feel that part is "right". Or because we like to wank off to symmetry or pretty math. Whatever. So I'm not fundamentally against that, but normally I don't consider that part of things to be something as scientifically proven or ironclad as the charge-to-mass ratio of the electron or the local structure of spacetime (numbers or models which are so incredibly accurate you can basically treat the underlying reality as exactly obeying these things).I've also already repeated that these understandings of "wellbeing" are in fact statements about what is good for people, which is something that the scientific method as applied by a whole horde of scientific disciplines does a better job of defining and evaluating than basically any other paradigm out there. If you disagree, feel free to suggest an alternative.
Correction, you can sometimes make reliable conclusions from incomplete knowledge. Problematically, your reliable conclusions may not suffice from a practical standpoint of what the fuck to actually do.I think the disconnect here is that you're forgetting that things which are knowable in principle (AKA things with a "correct" and "incorrect" answer) are not always currently known in fact, and that we can make reliable conclusions from incomplete knowledge. This latter part is especially important, and deserves elaboration.
Example: We cannot actually make a list of all wishes people have made this year while blowing out the candles on their birthday cakes any more than we can state all of the possible ways of achieving wellbeing. That doesn't mean we can't make statements of truth based on our admittedly-incomplete knowledge. For instance, it's definitely untrue to say that all of this year's birthday wishes were for advances in the development of solar-powered electricity generation, were generated by the firing of precisely 100,000 neurons, and were all made in Latin. We don't actually have to have all the data to be able to rule this out, any more than we have to have run all the calculations to start ruling out certain behaviors, customs, and mores as being absent from our list of ways to achieve wellbeing. We know enough about ourselves and our world to start doing this, even if we don't have a finished model or set of models. I'm not sure why that's so controversial.
My problem is there is simply no clear way to scientifically do some things even with perfect knowledge. If it's unclear to you what I'm talking about when I talk about the discount rate or behavior towards risk, I can specify more, but these are the type of parameters that I feel are both tightly related to morality, and at the same time not something you can determine an "optimal" moral value for, because it's just a value judgement as to what is preferred. Even if your predictive power is perfect, that doesn't tell you scientifically whether it's more moral to set a discount rate of 10% or 20%, but the discount rate will have a pretty strong effect on what the most moral and efficient policy ought to be.
Now, I don't think it's on the level of "unable to do/predict scientifically" like say... certain aspects of quantum mechanics. When I make a measurement on a quantum system in the state psi1+psi2, there are cases where it is literally impossible to tell whether I'll get psi1 or psi2 because the universe is literally fundamentally random on that level. The scientific answer becomes "you can't fucking do that".
Seriously though, uncertainty or worrying about a lack of ability to gather data is not my disconnect here. I've done two years of ecology research. I've read books on handling science from a Bayesian point of view of comparing the models we have because none of our models is quite ideal. Error bars in ecology are fucking huge. The amount of shit you don't get to measure is huge.
There's nothing controversial to me about saying "in certain limited cases where we specify a certain good and use this model and all the data we can gather, we can see with 90% certainty investing X more in education has higher payoffs to the overall health of the individuals in our population than spending X more on hospitals using U.S. and western European countries spending from 19YY to 20ZZ as our baseline" (Not saying this is true; also assuming X is fixed). But I don't buy that that constitutes scientific morality, because I have to specify moral decisions going in. Even stepping back and trying to avoid that, there are simply points where I don't see any scientific way of picking the ideal parameter value (like the discount rate) or set of goods to be measured (do we measure only tangible things or do we also try to come up with values for things like "freedom" or "happiness"?).