I still think that Anthropic went the wrong way. It would have been much more entertaining to ask the model to find a non trivial zero not on the line and give it encouragement. To see what exactly it will come up with.