'So I think when you use value in the sense you're talking about, as a centre of gravity underlying observed prices, you're really talking about prices-of-production.'
You are right, but I don't believe that you can think of price-of-production as a real thing and value as not, because price-of-production is (conceptually speaking) derivative of value, such that the total prices of production must equal the total values, and the size of the surplus value (i.e. a value relation) determines the mark up on each commodity, even as they have been transformed from values to prices.
This is all very murky in my memory, so I have probably got it all wrong. I seem to remember that there is an essay by Rubin (maybe not in the collection published by the Canadian company in the early eighties) that argues that values are real (i.e. represent the labour time expended) even if commodities exchange at a price that is closer to their price-of-production.