You shouldn't take the simplest. You should take the best one which already proved to work in other coins.
Times for alt coins won't become simpler once the scrypt asics get into this game.
A moveing average is something that is always delayed. With or without limit you will pay in difficulty everytime a pool-hope jumps in.
I rather liked if you considered to fix it right once and for all.
Certainly retargeting after each block and averaging is already a lot better than what it is now. But will that hold for the future?
I have my doubts.
Note that Mav's comment was 'simpler' not 'simplest' and that's important. I'm happy to report that my proposal, the 100 and 500 block SMA with limit as developed by the Feathercoin devs and used in Phoenixcoin, was not selected as it is more complicated, designed for a coin much faster than ours, and would likely only move us 10-20% closer to long-term stability. Likewise, we did not choose to completely replace the guts of the retarget process as has been heavily promoted by one large cat speculator. Mav and Hozer found that we can get the performance we need - today and tomorrow - with the 1/36/12% process we've been testing.
The 1 block retarget, 36 block SMA, with a 12% up/down limit has been tested on the testnet and has reacted very smoothly and quickly to rapid increases in hash rate (we started with instantly adding 20 times more hash than base, then increased that last night. This fairly closely simulates the current low hash rate and the injection of computer power by switching pools we've seen to date. The difficulty and block times are responding very well. Most importantly, this eliminates the massive difficulty swings that have brought coinhopping pools to us in the first place.
Rest assured that the number one goal is to fix it right and stabilize the coin for the very long term - for miners, investors, speculators, and retailers.
Here's a look at the diff response from last night's 30x hash rate 'attack':
http://waywardlife.net/blak/catcoin/Thanks,
Andy