I agree with rafaelenrike, I like the previous behaviour of The Dude much better. Why:
A device has a problem, it turns orange or red. You notice it, ack it so the scary orange/red is no longer shown and go on to fix the problem. Once fixed, the device goes back to green. Another problem? Please turn red again and do not 'auto-ack' to the non-warning blue. Nobody notices the blue! "Hey, that device turned blue! Is it a 'good' blue or a 'bad' blue?" Who knows?!
Alternatively, make it a choice (per device) if they should auto-ack or not. For unimportant systems (client PC's) auto-ack is great so you can see when they are online or offline, but they don't cause scary orange/red labels. For servers, I never want an auto-ack. Every time there's something wrong, I want to know about it with scary colors!
How am I to know which devices are on 'auto-ack' and which not? Should I have to unack every device once a week, just to be sure I am getting the 'reds' when an error occurs? I mean, who knows what others might have acked? I feel that is the biggest issue with this new behaviour: keeping track of the 'good' auto-acked and 'bad' auto-acked devices.