It's important to get lost in the convolutions of the problem. You need to dive in if you hope to catch an insight.

However, if you also hope to survive and make something of the experience, it’s equally as important to set systems in place that drag you out of the whirlpool at a certain point.

A deadline is a great example of such a system, although not a very sophisticated one. Invest a little time in advance to define the area to sound out; you’ll be better able to estimate how long it will take you and less likely to get lost in the process.