simple maximizers vs reason-based agents

I guess I’m into this talk. In short (b/c it’s 1hr long and a little new-agey):
You can approach decision-making by maximizing some number, conceptually. Food A is taste 2 + healthiness 5 = 7, food B is taste 3 + healthiness 3 = 6, so you eat food A.

But you can instead look at it by looking at the values you have, and compare them to the values that each food helps you fulfill. My values are (taste, healthiness, x, y, z) and food A fulfills x and y, and food B fulfills z and not w. Plus, my values can shift after I see the options.

I think you can see them as the same thing, like I think they’re mathematically identical at some level. But as any designer or any computer scientist who’s written assembly and javascript knows, looking at it a different way (or at a different level of abstraction) really matters.

Half formed thought and I gotta go write more thesis now. Eh, hitting publish!


blog 2024 2023 2022 2021 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010