I suspect "children" here may be a distraction: under the same artificial constraints, adult players might show the exact same behavior. From the article (and

I suspect "children" here may be a distraction: under the same artificial constraints, adult players might show the exact same behavior. From the article (and abstract), it's hard to imagine a game where the 'bad' advice is really 'bad' if (1) the player always trusts it; and (2) they then always succeed. In such a case, the forced-success may retroactively make the bad-advice 'good'. (At best, maybe the bad advice forced a slower completion? But that requires assuming some more complicated game.) From playing the game of interpreting this article, I have concluded that neither the article author nor the researchers (via their original abstract) can be trusted to summarize such research in sufficient detail for unambiguous interpretation.