It's 5%. Not much more than my predictions on the "bike weight matters no more than rider weight" model would suggest - and there was a revolution in training methods between those two dates.

Otoh if rider weight didn't matter at all, then the change from steel bikes to CF ones should have dropped the time to less than 90 minutes!

So, no, that only confirms my logic: a kilo off the bike matters no more than a kilo off the rider. It's total weight that counts, so if you knock a lb of bike weight a combined rider bike weight of 200lb then you have increased hill climbing performance by - at most - 0.5%.