Inspired by you, I attempted to answer “Where do we derive values?” and put it in a chart.
via oracle
genetic (inborn predilection toward altruism)
revelation (direct)
prior authority (ignore original vehicle)
via search
historical evidence
counterfactuals (simulated behavior)
behavioral reward/punishment (randomized search with pruning, habits)
via algorithmic computation
consequentialism (same as counterfactuals?)
virtue heuristic (anything optimizing on fixed small rule set)
identity/hero based (intersection of virtue and historical?)
as consequence of lower level (emergence)
arise out of selfish desires (e.g. need for food, safety)
arise out of tribal desires (we do group optimize)
Not sure whether your section 4 represents greater subtlety that I’m missing or whether it is straying from the question/answering a different question. Thanks for offering a framework that will allow me to better target my learning.
The community consensus seems to argue that computation is the only legitimate source of values and that we should isolate the influence of other sources. I expect that most people use search methods. Provided a large enough historical record, it would be easier to ask “What has worked well for others facing similar choices?” than to reconstruct a correct choice using a heuristic based method.
Inspired by you, I attempted to answer “Where do we derive values?” and put it in a chart.
via oracle
genetic (inborn predilection toward altruism)
revelation (direct)
prior authority (ignore original vehicle)
via search
historical evidence
counterfactuals (simulated behavior)
behavioral reward/punishment (randomized search with pruning, habits)
via algorithmic computation
consequentialism (same as counterfactuals?)
virtue heuristic (anything optimizing on fixed small rule set)
identity/hero based (intersection of virtue and historical?)
as consequence of lower level (emergence)
arise out of selfish desires (e.g. need for food, safety)
arise out of tribal desires (we do group optimize)
Not sure whether your section 4 represents greater subtlety that I’m missing or whether it is straying from the question/answering a different question. Thanks for offering a framework that will allow me to better target my learning.
The community consensus seems to argue that computation is the only legitimate source of values and that we should isolate the influence of other sources. I expect that most people use search methods. Provided a large enough historical record, it would be easier to ask “What has worked well for others facing similar choices?” than to reconstruct a correct choice using a heuristic based method.