Merely thinking about second or third order consequences is not sufficient if the predictions can't be reality checked.
The mental models we use are rarely good enough for that, unless you're playing a complete information game like Chess.
This is why "gradient descent" is so powerful and widely used intuitively.
Merely thinking about second or third order consequences is not sufficient if the predictions can't be reality checked.
The mental models we use are rarely good enough for that, unless you're playing a complete information game like Chess.
This is why "gradient descent" is so powerful and widely used intuitively.