I've never had so much fun planning for coding before. Maybe because I've never made such a... right plan before. Accurate? No, it's the lack of false precision that I'm enjoying so much. The proof will be weeks from now of course, but I think I'm overcoming the planning fallacy this time. And explicitly writing down the bits where my variance is really high makes it very easy to see which parts of the plan could do with further thinking.
Lara gave her crash course in machine learning for NLP today. It was probably valuable? I don't know, it still seems like few points delivered well just works better. If people follow up on her talk then it's the best thing ever, but realistically who's going to? Unless we push... okay, fine. I'll push.
I'm nearly done getting together my own presentation. Tomorrow I'll finish it up so I can get the LW one written out with time to spare rather than in a crunch.
I would totally purchase extra time. But not, I think, an extra year for a hundred million dollars. :D