The problem is not his accuracy but the frequency.
Talking for the sake of talking just to generate news. In most cases, analysts are rated not by their accuracy because they can always phrase their estimates to cover their asses but by how often they are quoted, by constantly changing their estimates by wide margins, they artificially generate extra quotes and because there are so many estimates made, it is easy to fall back on an earlier estimate and say, I got it right.
I actually disagree with you 100%.
Munster was rightfully criticized when he put in a million "what if's" when he predicted the 45 million iphone a year number --- 3 years later and we still haven't seen the mythical nano iphone. China didn't start selling the iphone 2 years later. He put the cart in front of the horse when he was making his estimates.
Now Munster is doing the right thing. When number sales numbers come in, Munster changes his estimate models. When actual news come in (like Apple actually announces international sales in this and that country), Munster changes his estimate models.
Do you want Munster to make his estimates on known facts or on some mythical rumors? We have no idea when China is going to sell the ipad. It may be next quarter or it may be in 2 years. I rather prefer Munster to make his estimates right now --- without looking at the China rumor. When Apple announces the China launch date, then Munster can make a new estimate by that time.