Status as of Saturday, October 11

[Steve Sokolowski]

Here is today's status:

• Chris states that the next release is 80% certified. He still has to test 20%. If we cannot get merge mining working before he finishes testing, then we will release without it and simply disable the dogecoin daemon. The major issue with this release is the infrastructure changes required to support multiple exchanges and merge mining; the actual implementation of Poloniex and dogecoin mining can be turned on without a restart.
• Poolpicker.eu is listing profitability delayed by an entire day, such that the October 10 figure goes in the October 9 column. Now I understand why some people were saying that the top bar payouts didn't match up with poolpicker's statistics. I apologize for that. From now on, always prefer the top bar if the two figures are in conflict. In the next release, I will make the data we send to poolpicker.eu "incorrect," so that it shows up correctly on their system.
• The algorithm is displaying behavior that we did not expect. One example comes from the Elacoin market yesterday. We were selling Elacoins all morning, and when the price was depressed enough that Elacoins were no longer profitable, we left. Then, exactly at noon, someone dumped 12k Elacoins, crashing the market by 80%. We had been wondering why no other pools had been onto Elacoins before this, and our theory is that other pools actually were mining Elacoins, but they are not trading in real-time. The pool that mined Elacoins lost 3/4 of their investment, and therefore, selected a less profitable coin to mine. If correct, that implies that the larger this pool grows, the more we will decrease the profitability of any pool that does not have realtime capabilities and multiple exchange support. While we were confident that our algorithm is profitable, we had not anticipated that other pools would decline as a direct result.
• I'm going to post an analysis of this next week, but it's interesting that of the payouts that were paid yesterday, 27 of the 29 ($300 out of $320) were in bitcoins, despite the ability to select any coin. At this point, I think this statistic is more a result of not having advertised in forums supporting altcoins and most of the people having come from a site that compares profitability in bitcoins rather than a true preference for bitcoins, although if these numbers hold, they make a strong statement that bitcoin is the currency of the future and altcoins don't have a chance. Again, don't make that conclusion until we have advertised everywhere.
• Looking at referer logs, there were 1672 people who visited the site on Thursday. Of those, 24 people visited Chris's facebook profile. It's an unimportant but interesting fact; he didn't get any friend requests, so I'm not sure why so many people would go to his profile (which he never updates).
• In the next release, I changed the top bar's profitability to the past hour, at dysan's suggestion. But because less data is included, that has the potential to go way up to ridiculous values when coins like Elacoins are being mined. Feel free to disagree in the comments. The original reason we changed it from instantaneous to 24-hour average was become it would mislead people into thinking that current profitability would stay that high indefinitely, but with a 24-hour average, current profitability can also be much higher or lower than the displayed figure. Perhaps 1 hour will fix that.
• When this release is issued, we need to take the entire system down to issue 47 database upgrade scripts, to perform security patches on various equipment, and to upgrade code. While one should usually have a backout plan, there is no rollback available for this release since a rollback would take 3-6 hours and it would cause a loss of funds. I'll notify people when the testing is complete, which probably won't be this weekend. In the meantime, if you're reading this and have software that supports it, please add a "backup pool" to your mining software so that your miners can continue mining somewhere else while we are performing this massive upgrade.
• This morning, I found this shocker: https://stackoverflow.com/questions/970 ... imary-keys. Apparently, postgreSQL doesn't automatically create indexes on foreign key columns like every other database in the world does. This should allow me to immediately increase performance by a factor of 10 or 100, which I'll do today. I can't believe that the developers wouldn't do that automatically.

[jimlite]

You should have TWO stats, the last hour profitability (so people can see in real time) and 24 hour profitability like you have. Do both.