Watching election polling: small changes in assumptions produce meaningfully different probabilities. That leaves me at 72% until we get better evidence.