Quote from: LeoC on January 08, 2014, 11:04:18 PM

There needs to be a cap of 20% on difficulty increases, and 30% on difficulty decreases.
I can't think of anything else that needs to be changed.
Retarget time is good. Block rewards and such are good. Max coins is good.
Literally the only thing fucking with us are the large hashrate coin jumpers.

As I understand the proposed mechanism, this would result in a rigid structure of difficulty levels changing very gradually regardless of what happens in the coin markets. So if the coin goes way up in value in the markets (press coverage or some big investor decides to jump in), it would cause the coin to almost immediately land at the top of the profitability charts. In turn, lots and lots of coin hoppers will show up to take advantage of the ability to instantly mine and dump on the exchages for instant riches. This process would inherently also involve a lot of coin inflation, with block solutions maybe every 30 seconds. Inevitably, this sudden influx of rapidly created coins dumped on the exchanges will sooner or later cause a crash or correction in the price of the coin - before the difficulty has had the chance to adjust - at which time the profitability will suddenly be worse than LTC, and all the coin hoppers will leave as if in a concerted coordinated massive action. Now blocks are coming once every few hours, and difficulty adjustments are not allowing things to return to realistic levels quickly enough. So in the aftermath of a crash in the coin price, we have the added burden of block solutions being hours apart (from the recent high) during which period the coin would be criticized as 1) unprofitable and 2) unusable. More miners leave. These things together aggravate the recent market crash, further reducing the coin's exchange value and reputation. Then inevitably, there would be talk of yet another hard fork. Around this time, Cryptsy and other exchanges get tired of it and start internally referring to it as yet another not well thought out junk coin requiring a lot of chores of them in seemingly endless hard forks - and drop the coin. More miners leave. There is no longer any clear market price for these coins anymore without exchanges to trade them at. People openly call the coin dead. The QT clients start having a hard time connecting to the network. I apologize in advance if you consider this to be an unfair characterization or extrapolation of how your formula would respond to the hypothetical scenario I have presented - if you have a better way of describing how the scenario would play out, please let describe how you see it playing out.

I am holding out hope that the next release of the coin if it involves a hard fork, will specifically address coin hoppers collecting easy profits - and by that means establish stability in the difficulty level relative to market prices - and not have the network hashrate or difficulty level respond so much to temporary cherry pickers trying to show up to cash in. Please keep in mind also, if there are any questions about credibility of my models of the coin hopper's psychology, difficulty swings, and market prices - that I used this same model to predict - before the hard fork took effect - that we would suffer wild difficulty swings afterwards. My publicly declared prediction based on this model has come true - so if the same model predicts problems with this proposed solution - it may be worth taking heed and adjust it or do whatever it takes so the solution we eventually implement will be successful in light of the proven model of how difficulty levels, coin exchange markets, coin hopper psychology interact.

Quote from: hozer on January 08, 2014, 11:09:32 PM

quit bitchin' and https://github.com/kR105/catcoin/fork show me some code, or at least send KR105 some catcoins.

I've proposed something non-simple ( https://bitbucket.org/dahozer/catcoin/wiki/Stake-mine ), but what a very simple change: retarget *every* block based on the last 36 blocks.

Fair? Say yes and give me a block number and I'll run it on my p2pool node and post the code.

I think if we retarget at every block, we will see difficulty changes would be hard to predict, and the profitability of the coin, if it could be put on a graph, would look very erratic. There are other altcoins known to have very erratic profitability graphs, and they seem to suffer negative reputations for this reason - and they show up randomly at the top of profitability charts, because being erratic inherently involves landing at the top of the profitability charts (even if for a few minutes at a time), which could contribute to the erratic behavior, as major hashpower is thrown at the coin for a few minutes at a time to try to take advantage of the apparent split-second profitability. I do not know if this is a formula for long-term success - would you see back-porting this difficulty adjustment formula to LTC or BTC and having them do as well as they are doing now - or better? After all, BTC and LTC are the gold standard of stable coins with high market cap. If anything, the retarget should be lengthened, while simultaneously taking away all economic incentives to coin-hop.

Also, I think it makes sense to study other altcoins for strategies that have already been tried, and see if they were successful. From the GalayCoin information thread at cryptocointalk.com:

Quote

GALAXYCOIN GLX
Galxycoin is a descedant of Novacoin, it uses both Proof of Work and Proof of Stake. It is effectively resistent to 51% attack. It provides steady coin supply at 64 coins per block, for 8 years. The difficulty retargeted each block, with the last 10 blocks average, this ensures a fast adjustment to the network hashrate, while maintaining a stable transition.

I have seen GLX appear intermittently at the top of the profitability charts at coinwarz.com, but usually near the bottom of the chart, and sometimes in the middle. This appears to be the characteric to expect from continuously varying the difficulty of a coin. Also, I am not certain that GLX has avoided the attention of coin hoppers entirely, or I would think despite the continuous difficulty adjustment, it would have found some degree of stability from a core pool of dedicated miners. Maybe I missed something and there is a fundamental difference between what GLX implemented and your proposal, but in any case it seems good to look to earlier efforts for any potentially useful lessons.

Thank you for your consideration.

Etblvu1