3 ms·

I think that the problem is even deeper, the idea of an actual concrete "universal expected moral utility" seems so shaky to me. I appreciate utilitarianism, an

I think that the problem is even deeper, the idea of an actual concrete "universal expected moral utility" seems so shaky to me. I appreciate utilitarianism, and think it provides important and useful guidance, but the EA approach takes it way too literally. If you're trying to decide whether to fund research of vaccine A or vaccine B, sure the outcome variable is clearly well defined, but if you're talking about the sum total human happiness 100+ years down the line....I mean the error bars on that are so wide any honest error propagation is going to render your conclusions useless. It's the same grift as prosperity gospel stuff: provide tissue thin justification for its rich participants to feel good about the horrible things they do to get rich.

kwhitefoot 3y ago
> .I mean the error bars on that are so wide any honest error propagation is going to render your conclusions useless. Are any of the leading players in the movement scientists of any kind? I have worked with many people who while being obviously very smart, including engineers of many kinds and software developers, who don't have any idea about errors and compounding of errors. When I studied physics in the mid-70s a compulsory feature of every experimental report was the expected error in the result which had to be justified with reference to the expected errors in all of the measurement equipment and any other sources of error. When I got into industry this discipline seemed to be pretty much absent. In my 40 year career in electronic and electrical engineering I don't think I ever saw an engineer include any information on the expected error of a set of measurements even though most were familiar with the concepts of statistical curve fitting and so on. They were not much better than social scientists and would plot the results of an experiment in Excel, click the button to get a polynomial fit and accept the coefficients as gospel totally disregarding the R value. And most of them didn't care that the polynomial form wasn't even applicable to the data.
- tech_ken 3y ago
  Don't even get me started on error bar usage in industry. I've seen Big 3 consultants charge several 0s for a regression analysis that I would give a C+ in a first year grad stats course. I can't imagine how any of EA's guiding lights could have used stats in a practical application and still want to hold it as the foundation of their entire moral outlook.
BlueTemplar 3y ago
Like the assumption that maximizing something is "good", even QALYs...

kwhitefoot 3y ago

> .I mean the error bars on that are so wide any honest error propagation is going to render your conclusions useless. Are any of the leading players in the movement scientists of any kind? I have worked with many people who while being obviously very smart, including engineers of many kinds and software developers, who don't have any idea about errors and compounding of errors. When I studied physics in the mid-70s a compulsory feature of every experimental report was the expected error in the result which had to be justified with reference to the expected errors in all of the measurement equipment and any other sources of error. When I got into industry this discipline seemed to be pretty much absent. In my 40 year career in electronic and electrical engineering I don't think I ever saw an engineer include any information on the expected error of a set of measurements even though most were familiar with the concepts of statistical curve fitting and so on. They were not much better than social scientists and would plot the results of an experiment in Excel, click the button to get a polynomial fit and accept the coefficients as gospel totally disregarding the R value. And most of them didn't care that the polynomial form wasn't even applicable to the data.

tech_ken 3y ago
Don't even get me started on error bar usage in industry. I've seen Big 3 consultants charge several 0s for a regression analysis that I would give a C+ in a first year grad stats course. I can't imagine how any of EA's guiding lights could have used stats in a practical application and still want to hold it as the foundation of their entire moral outlook.

BlueTemplar 3y ago

Like the assumption that maximizing something is "good", even QALYs...