Using paleontology and molecular clocks
This image from Myers' post about vaginal evolution in mammals intrigued me enormously. It has two trees with the same topology but using "hard" data - the fossil record - for the black version and "soft" data - molecular clock estimates - for the red.
I imagine a cone of decreasing probability from the paleontological record back to the molecular point, so that the likelihood that the lineage existed within that range decreases (actually, it would decrease to the lower of the error interval of the molecular date). Time erases information - it's a fact of history. We don't know within a strong confidence interval when many evolutionary events took place - at best we know that they took place. I think there is something deep about the philosophy of history in this.
Just thought I'd share.