Fatman3001
Legendary
*
Offline Offline

Activity: 1274
Merit: 1003


Make Bitcoin glow with ENIAC


View Profile Personal Message (Offline)

Ignore
January 02, 2016, 11:44:53 PM

I'm going to stop including 1 transaction blocks in the full-block calculations though. There were three out of six (all F2Pool) when I looked yesterday. That's skewing things way down.

If the current formula accurately reflects the percentage of total potential used block space that had actually been filled, I would advocate no change. To do otherwise turns it into a meaningless statistic. Better to have the simple unvarnished truth, rather than some manipulated figure meant to illuminate some vague outcome.

+1  The existence of empty blocks is more a result of the gaming inherent in bitcoin than any explicit aim on the part of miners. As Richy said, empty blocks are allowed, but they will only succeed where no better block is found in the interim. A chain with a non empty block at its tip will always have more work than one with an empty block.  A chain with the most valid work is by definition the longest chain. But if the fuller block isnt found quickly enough, the empty one will be built upon next.

So I see no issue with including them.

I support Richys conclusion that including empty blocks skews the results. Certain pools intentionally mine empty blocks even when there's tens of mb in the mempool and all other blocks are full. There doesn't seem to be any fix for this other than to wait until fees are a larger part of the miner reward, so it's irrelevant for the blocksize problem in that respect. However, to have as many as three empty blocks in an hour sounds extraordinary, so I'm wondering whether the numbers will be significantly affected over time.

Ideally I'd like to see both versions plus a snapshot of the mempool size. Or a version without empty blocks, number of empty blocks for that period, and a snapshot of the mempool size... but that sounds like a lot of work.