Quote from: thezerg on July 05, 2015, 02:06:05 AM

Quote from: Peter R on July 05, 2015, 01:46:56 AM

Am I being sensitive or is this an unnecessarily spiteful reply from Greg Maxwell?:

Quote

...
You've shown that you can throw a bunch of symbolic markup and mix in a lack of understanding and measurement and make a pseudo-scientific argument that will mislead a lot of people, and that you're willing to do so or too ignorant to even realize what you're doing.

Its a strategy (implemented unconsciously by many) to limit participation to the select few. Unfortunately it tends to create a situation where only similar personalities contribute which is where we are today with the core devs, Gavin excepted.

Thanks. That makes me feel better.

Quote

I read his 21ms validation number but its weird because I was wondering just weeks ago why it was taking so long to sync a measly week of blockchain data and came to the conclusion that either the P2P code is complete garbage (compared to bittorrent for example) OR the validation cost is high (given my fan speed, I assumed it was validation). And if validation is so fast, why would these pools have custom code to skip it?

This is the empirical fact that motivated me to perform the analysis. I was surprised how many "defensive blocks" we were seeing. Furthermore, based on these numbers:

Quote from: TheRealSteve on July 05, 2015, 12:23:37 AM

...in the last 27,027 blocks (basically since jan 1st 2015), f2pool-attributed blocks: 5241, of which coinbase-only: 139
For antpool, this is 4506 / 246.

it looks like the average time these pools are mining empty blocks is 16 seconds (F2Pool) and 35 seconds (AntPool) before switching to non-empty blocks. Like you said, why are these numbers so big if processing the blocks is so fast?