Unfortunately, the problem with “fuck the symbols” is that the universe is more complex than any participant can calculate, so abstractions and symbols are usually absolutely necessary to actually think or act on any topic. There is no “the thing” that fits in a brain—only different dimensions of compression and lossy abstractions.
Certainly “be sceptical of symbols”, and “try multiple symbols for the same state of things”. Being aware that over-focus on a simplified model is usually wrong in important ways is key here. But you can’t do away with them entirely, you just don’t have enough processing power.
But you can’t do away with them entirely, you just don’t have enough processing power.
I don’t think that’s what johnswentworth is advocating.
This isn’t always the best choice, but it’s usually worth at least thinking about how to do it—because the process of thinking about it forces you to recognize that the Symbol does not necessarily give the thing, and consider what’s actually needed.
Related to https://www.lesswrong.com/tag/goodhart-s-law, which has had a lot written about it, though in somewhat different contexts.
Unfortunately, the problem with “fuck the symbols” is that the universe is more complex than any participant can calculate, so abstractions and symbols are usually absolutely necessary to actually think or act on any topic. There is no “the thing” that fits in a brain—only different dimensions of compression and lossy abstractions.
Certainly “be sceptical of symbols”, and “try multiple symbols for the same state of things”. Being aware that over-focus on a simplified model is usually wrong in important ways is key here. But you can’t do away with them entirely, you just don’t have enough processing power.
I don’t think that’s what johnswentworth is advocating.