@jen
I follow Mark McGeoghegan - who noted this on Scotland yesterday
"The Ipsos team have done an excellent job communicating uncertainty & marginality in their MRP outputs - 12 toss-ups in 🏴 & a further 14 seats decided by <10pts (i.e. marginals). Nearly half the Scottish seats could be swung by small shifts in national vote %"
I'm also learning that even the top independent analysts can't follow MRP variation on seat count, as the modelling assumptions and data sets are proprietary so it's a bit of a "black box" (tonnes of chatter on this last night)