the fine-tuning fallacy

clay shentrup
2 min readJan 1, 2023

there are two necessary checks a system must pass to demonstrate intelligence.

a computationally challenging goal

suppose alice hands a note to her son and he responds, “17,375”. at this point, her husband bob has no evidence of mathematical talent in his son. now suppose bob looks at the note and it says, “pick a random number”. he wouldn’t be particularly impressed at his son’s intelligence. but suppose the note says, “add 17,000 to 375”. this isn’t rocket science but it’s at least somewhat computationally challenging. but now let’s say the note says, “multiply 125 times 139”. now bob would be fairly impressed because that’s a much harder problem.

so a result isn’t enough to evince intelligence. evidence of intelligence requires a computationally challenging goal.

improbability

suppose alice asks her son a difficult yes or no question, such as, “is 17,353 the square root of 301,126,609?” although this is technically a much more difficult problem, it’s highly probable that he’ll get the right answer purely by guessing.

so evidence of intelligence requires improbability. that is, there must be a very low ratio of correct answers to incorrect answers.

the fine-tuning fallacy

the problem with the fine-tuning argument is that it is based on only one of these two criteria being true: improbability. that is, the specific conditions of this universe were were extremely improbable.

but no evidence is offered that “god” (or whatever you want to call it) had a computationally challenging goal. we obviously didn’t see god declare its intent to create this particular kind of universe before creating it. so far all we know, this universe was just randomly selected, and thus didn’t require any intelligence whatsoever.

--

--

clay shentrup

advocate of score voting and approval voting. software engineer.