Yeah, not for these tasks. This concerns learning something new and analyzing an algorithm. It's new ground for me. And the algorithm looks deceptively simple, but some of the tooling fails its analysis procedure on what is clearly bug. So I can't "rely on a successful analysis".
Also, you only find out how many details there are to consider, until you start writing it down. Not that it's difficult, but takes time to be structured and thorough. 

 To clarify, add some nuance: *this kind of analysis* is new ground for me. 

 Sometimes writing down helps you find the defect - and I often start any relatively complex automation by writing it all out anyway. Gives me some perspective, and I usually find ways to either speed it up or make it less complex as I go. I'm motivated to do so, as writing it out isn't all that fun. 

 Oh sure, I know, same experience here. I'm writing out an idea that seems to be quite dense with subtleties.