[XorNot]

My understanding is there's a very - superficially weird - sort of logic at play with this though.

Basically if we're going to take a child not presently abandoned or in danger, and place them with someone, we need to know damn well that we're not worsening the situation for the child.

But if you have a child who was already abandoned and in danger, and you start looking after them unprompted, the situation for the child has already improved and almost any other action will worsen it - i.e. it's generally accepted that children being wards of the state is a worse outcome in almost all circumstances compared to a dedicated parent.

A comparable example I suppose would be the question of what's the best strategy for seeking help if you're lost: basically, statistically, it's approach the first person you see and ask for help. Because the occurrence rate of predators in the population is low, so the first person you see is unlikely to be one. But if you stand around for a while looking like you need help, well now you're obviously a target and the chances of someone who approaches you intending ill-intent rises.

[Digit-Al]

I guess, if you really think about it, a lot of the safety checking is to make sure that the person trying to adopt is not a predator who will abuse the child. In this case, someone who was like that would not have called the police and handed the child in, they would have just taken the baby to abuse; so his handing in the child to the authorities automatically ticked the "not a child predator" box. Thanks for helping me to think about this in a different way :-)

> A comparable example I suppose would be the question of what's the best strategy for seeking help if you're lost: basically, statistically, it's approach the first person you see and ask for help.

Ah yes, from the Paul Graham article on security. I bring that one up myself from time to time :-)