This is a good point. People wildly overstate the inaccuracy of long range models because everyone is so focused specific outcomes in their own backyard. Long range outputs are still pretty useless for local forecasts, sure, but like you say, the general 500mb theme will usually not be unrecognizable from the model outputs 15 days beforehand. A ridge axis shifting 500 miles here or there will result in enormous differences for any given point in a local area, but on the scale of the global climate at large, it's not that big of a shift. Let's test this, I'll set a reminder to come back to this post on 15 days. I bet you the actual 500mb pattern will still be vaguely recognizable from these maps. They already look pretty similar on a global scale. Bet you when we check this in 15 days, there will be some kind of negative anomaly over NE North America and the SW US, and some kind of blocking over the North Pacific, Alaska, and/or far eastern Russia.

^ Really trended toward a +NAO in the medium range. I would be curious to see if there is still strong 10mb warming forecasted, with now a moderate +NAO happening.

10 day models have 0.85 correlation with accuracy and 15 day models have 0.80 correlation (500mb). You can tell by 500mb what the probable pattern will be, at 15 days.

00z GEFS - Moses Lake. Nice cold pool signature which might trend colder.

Eh, I would say we still can't predict anything past 7 days reliably.