a) Is definitely part of the emerging science. The future would be satellites that can do soundings. We're already part of the way there with GOES-16/17. They have 3 different WV channels to sample moisture at roughly 700, 500, 300 mb.
b) Is definitely a source of potential errors. Convection is always a problem, but NWP has come a long way from having convective feedback that early in the cycle. It wasn't that long ago when by hour 6 a nocturnal MCS could already be running wild through the GFS.
It goes to show you how much work the GEFS need though. Realistically there shouldn't be such wild swings from 00z to 06z, given it's the same base set of data. I'll be curious to see if the EPS 06/18z runs show as much volatility, but as of yet I don't have access to them.