I think it's poorly worded - there were a bunch of other aggregators at the time who said things like Clinton was 99% likely to win. IIRC 538 assumed that poll

I think it's poorly worded - there were a bunch of other aggregators at the time who said things like Clinton was 99% likely to win. IIRC 538 assumed that polling bias could be correlated between states while most analyses were assuming that each poll was completely independent, but that's probably an oversimplification given the vagaries of memory.