If someone had enough hashing power to build off their own chain (aka, 51% of the network; or using selfish-mining principles, as little as 30% of the network), they could predict the nTime required on the next block that would result in the following block being an anomaly. Using this system, and by only building on your own chain, you could cause EVERY block to be an anomaly. A would-be attacker (say, multi-pool), could mine the network, orphaning everyone else's blocks, and get a payout on every block at the same time.
Furthermore, there is absolutely no reason to mine worthless blocks (non-anomaly) - if you wanted to mine CGA, it makes the most sense to mine litecoin (for example) until the next block is an anomaly, and then switch to CGA, switching back as soon as a block is found by the network and the next block is not an anomaly. The diff is low enough that solo-mining would likely find this fairly effective and profitable.
If pools, or someone solo-mining felt like it, they could also pick nTimes selectively and cause the following block to always be an anomaly (wouldn't necessarily benefit them, but would significantly increase the number of anomalies that occur).
If you have this code I would like to see it. What you explained still requires someone to solve the 0 blocks... The probability of a block EVER generating an anomaly is at the most 1/3. So someone has to solve those 2 other blocks before you can even think about generating an anomaly. Once the difficulty is over 3 then the probability is 1/diff so you would still need to solve the 0 blocks. You can't just have everyone skip over them and make "more" anomalies... that is impossible.
True, at diff < 3, only every third block can be an anomaly. This in and of itself is pretty amusingly exploitable by miners - again, why would you mine the obviously worthless blocks? Let someone else (a fool it seems) waste their hashing power on the 0 reward blocks. A smart pool operator would know this too - and would be redirecting their hashing power to another coin (with or without telling their miners). If you can't find a valuable block anyway, then why bother - you can simply _say_ you found a worthless block once and a while, meanwhile keeping the litecoin or dogecoin or whatever you are also mining for yourself.
If diff > 3, then every block can be an anomaly, and therefore all an attacker needs to do is raise the difficulty above 3 while holding 51% of the hashing power (setting the difficulty as desired would be fairly easy - again by abusing KGW and nTime), to ensure every block is an anomaly.
Anyway, I would honestly prefer this coin succeed. I have made it pretty clear how to selectively mine this coin at this point, and I decided I would prefer not to release the few lines of code required to predict the next block. CGA seems like a neat idea - what brought me here is that I was considering mining it, but as I usually do, I dug thru the source code first. Which brought me to this conversation now.
I do hope this coin could succeed, but unless a huge change is made to the protocol, at some point, someone definitely will exploit this huge flaw.