Quote

I don't have the big picture what's going on behind the scene. For that it's better for me refrained from asking questions that have adverse implication and responsibility in the Riecoin Project. I won't be pursuing this matter of optimizing the superblock's difficulty over the demographic at ypool any further. I've made my points clear enough

Quote from: Ellieo on December 17, 2014, 09:03:45 AM

Quote from: fairglu on December 17, 2014, 08:27:50 AM

Quote from: Ellieo on December 17, 2014, 07:41:59 AM

One just needs to wonder how & what sort of parameters were used to estimate the first superblock difficulty before it took off. Had I realized the shortfall of the superblocks timing I'd have warned Gatra but correction is still aplenty. It's just a matter of choice, etc.

The timing is based on block intervals, rather than a set time of day, so the exact time is not set, and will depend on mining power, difficulty adjustement and a bit of random, so day and time will drift over the course of months/years.

First super-block was at 19:50, latest was at 22:07, and we had one at 22:30, so it's not a huge drift, but it's a drift.

So chances are the dry spell will last a few weeks, until we drift to the next peak.

The peaks don't drfit. Suffice to said the current superblock made a poor navigator (more stationary than dynamic motion) towards the peaks as its reference. As casues of the blocks drift versus their standard timing were the factors stated then there should be some algorithm placed (perhaps a self correcting drift) to pinpoint the correct timing for superblock at the peak of hashing power.

Since superblock helded as weekly event some database peaks pattern (weekly updated for 1 month of data) built into miners would help guides the drift correction, etc.

Just a thought.

If some suggestions are not implemented, it's not for lack of will. And some have technical issues that make them undesirable. Instead of refraining from asking questions, I'd like to encourage you to understand the technical details and ask more questions!

Hardcoding patters in the client is a security risk: patterns might change, and attackers could use them to inflate difficulty generating a DoS (long time without transaction processing). Difficulty adjustments should be based on block numbers and relatively large amounts of time, otherwise it would be too erratic. It's not an easy problem, and many others had issues with it (even BTC had an off-by-one bug, and don't get me started on KGW). Also remember that everything is probabilistic, no accurate predictions can be made on how long will a block take. For large sets of blocks it gets more accurate but still you can only get a confidence interval.

And even if we nailed it on the peak every week, we won't have a record each time: in fact it would be very similar to what is happening now (some weeks are record-breakers, some aren't, but all are above the previous record before Riecoin started). The only way to achieve a new record each week is by getting more miners each week.