Conversation

Top 5 cost (or value) functions 1. Time 2. Space (2d or 3d) 3. Energy or mass (literal) 4. Bits 5. Money Random multi-dimensional “utility” is haram. Do not take rms of a bunch of variables 🤬 The very concept of abstract “utility” is some sort of communist plot
16
63
Replying to
I will allow densities composed of the top 5 as sort of acceptable. Mass/volume, bits/mass, energy/time etc. Space/time (speed). But avoid if possible. Edit: for space, 1d is also ok
2
8
Densities are somewhat corrupt because they allow you to be unacceptably vague/abstract about extensive properties and scopes/boundaries, so only he who is without sin may use them.
1
7
The 5 is really 7, since 1, 2, and 3d are distinct. Those pure enough to use densities actually have access to a whopping 42 densities (!)… just divide any of the 7 by any of the other 6.
2
3
Dividing one of the 7 by itself is a dimensionless cost function, and is allowed by special indulgence, since shady shit can happen
1
2
I mean really what are you doing when you put in dimensioned weight coefficients and taking an rms? You’re fooling yourself that an arbitrary Pareto defined by the weight ratios is meaningful rather than just, like, your opinion that apples matter more than oranges.
1
2