I really really think doing a "hard fork" or changing the formula on this coin is a bad idea, because through a little bit of coordinated effort, we can make this work as-is, and once it stabilizes, it would be the only coin out there that is an exact copy of Bitcoin, which is a status you only get to lose once, and never get back.

If if we are going to change the code, let us at least introduce some innovation that no other coin has done, and keep enough of the parameters intact so arguably it is still a Bitcoin clone, and address the actual problem (destructive oscillation of difficulty levels), not the spilt second problem (high difficulty level). Changing the difficulty adjustment time only speeds up the destructive oscillation in difficulty levels, and this only makes things worse (just look at what is going on with GalaxyCoin).

If we are changing the code (and I still think it's better not to), it is possible to do that keep the following parameters intact, and address the underlying problem:

1) Block time = keep at 10 minutes
2) Difficulty adjustment = keep at 2016 blocks
3) Max supply = keep at 21 million
4) reward halving interval = keep at 210,000 blocks
5) Block reward = keep the same at 50, but with just one minor modification*

* You solve a block. If the current difficulty period is significantly easier than the last difficulty period, your block reward will depend on the results of performing a simple loyalty check. If you can point to a block you solved during the last difficulty era which you did not point to before (i.e., comes from the same address), you get the full 50 coins. Otherwise, you get half (25 coins). The balance of coins are carried over to the next block reward or destroyed.

This would generally keep away miner-dumpers, and reserve full rewards for loyal miners, regardless of what price swings may come in the future. Pool operators have incentive to enforce the same rule on those who pool mine, so the general trend is people will commit to mine or not mine the coin, based on longer term considerations (which is how people decided to mine or not mine Bitcoins orignially!), and not on split second "profitability" considerations. This code would be so Bitcoin friendly that they could back-port the code into the Bitcoin source code and it would make no difference in its operation, unless in some extreme future circumstances there was a wild swing in difficulty levels of Bitcoins (not something likely to happen), so we could lobby for this code to be back-ported into Bitcoin, thus making the coin again an exact clone of Bitcoin again.

I was actually going to use this idea for some future altcoin, but I am making it available for use here, because some of the other ideas floating around are so abhorrent to me. Please comment.