So instead of prepping for the 10000 for the date they ran their 10000 PR, they were actually prepping for the marathon (including getting a prior marathon or two under their belts if that would optimize them). I'm thinking the WR would have to be at least 30 seconds better than as it stands right now. KB ran 4:13/mile for for 10000 when he set the WR. Even a 26:45 guy was going 4:18/mile. 4:40/mile gets you a 2:02:21 Marathon. Sounds very reasonable, maybe even conservative. I know it's highly unlikely that a top athlete these days is going to shoot for optimizing at the marathon at the expense of track in their prime years, but if someone did, maybe it would be worthwhile for them.