The status of the project is that we are currently engrossed in trying to resolve profitability issues. I'll review each of the issues below and describe what we believe to be resolved so that customers can report them if there is still an issue there.
- Last week, we increased the minimum difficulty of the coins we mine to prevent networks with hashrate spikes from causing an increase in work restarts. The change appears to have been successful, with fewer work restarts being issued to most miners.
- On Sunday, we released a change that modified the way that difficulty is assigned to miners. Previously, if the most profitable coin had a lower difficulty than that of a dynamic difficulty miner, then the miner's difficulty was set to be below that of the network. Now, we try to assign miners who are already mining at a lower difficulty to those coins and we never reduce miners' assigned difficulties below a certain divisor. Currently, the divisor is set to 1, which means that miners will never be assigned to coins lower than their current target difficulty, but we may change this to 2, meaning that dynamic difficulty miners could have their difficulties halved when a period of exceptional profitability arises. It was not known to us previously that miners' output is affected by share difficulty and this change appears to have been successful at resolving many complaints about reduced hashrate on dynamic difficulty miners.
- The "low luck miners" problem appears to have been (strangely) resolved. The miners who had low luck continued to mine for several days while forfeiting their entire earnings of more than $100 each day, and then they suddenly disappeared and haven't returned. Chris found it strange that people would lose hundreds of dollars and, rather than getting angry and contacting him about it, simply vanish. Whatever the reason for that, such miners are easy to detect and their departure improved profitability by 10%. Because there are no remaining miners with the issue, we decided against devoting any further effort to creating the feature that assigns those miners to easy coins.
- After the resolution of these problems, we determined that profitability is still only at 89% of what it should be. Chris continued his investigation and found that slippage has recently become responsible for significant losses. He isn't sure what changed recently to cause that, but he thinks that shallow markets may be responsible. It's possible that the prices determined at the time of mining in those markets aren't available at the time of coin maturity. To resolve the problem, Chris will assume that a certain value of orders will be filled before we sell any of our immature coins. The result will be lower sell prices on shallow markets, which will cause the pool to favor coins with deeper markets where it is less likely to encounter slippage. Chris should be able to add the feature today and get it released overnight. While slippage will always be a concern, the effect of this change will probably be to mine coins that are, for example, 1% less profitable but result in 10% less slippage.
If you think that one of the issues we think is resolved is still present, or if you have another issue that isn't listed here, please let us know and we'll add it to the things to investigate over the weekend.
