You just reinvented p2pool.
Oh, I was afraid it could be something like this...It isn't a technology problem but rather a human "problem" (or maybe human nature is simply a better term).
Great, that thing again. Im lovin' it (NOT). 
Topic: Yet another Pool Problem solution : Mining multiple blocks simultaneously ? (Read 902 times)
June 24, 2014, 12:34:09 PM
June 24, 2014, 11:32:41 AM
The p2pool sharechain has an expected interval of ~30 seconds per share and that is probably close to what is achievable. Maybe with a greenfield solution optimized to be as lightweight and low latency as possible you might get that down to 10 seconds
P2Pool was previously 10 seconds but had to be changed after asics started shipping because so many of them had very high latencies and 10 seconds was somewhat problematic. 
June 24, 2014, 11:04:15 AM
You just reinvented p2pool. Sadly it has roughly 1000 shares per block (so 10x your proposal) and is still barely used (~2% now and maybe ~5% at the peak a year ago). Could p2pool be tightly integrated into an altcoin (or even bitcoin)? It certainly could however it is very likely you would simply see the same major pools running off the sharechain rather than the full blockchain. There is a lower limit on the expected interval between events (call them shares, blocks, hashes it doesn't really matter). The network needs time to adjust. When the average time of propagation becomes close to the average time between events, a network will drift away from consensus. The p2pool sharechain has an expected interval of ~30 seconds per share and that is probably close to what is achievable. Maybe with a greenfield solution optimized to be as lightweight and low latency as possible you might get that down to 10 seconds but you aren't ever going to have a scenario where 100,000 miners will be able to directly share in a block in a decentralized manner.
It isn't a technology problem but rather a human "problem" (or maybe human nature is simply a better term). A miner's major ongoing cost is electricity and payment of that cost isn't due on an hourly or daily basis but monthly. Sub-month variance is not an economic issue. If a miner has sufficient confidence (mathematical term) of the expected return over a month long period lower variance doesn't improve the economics or reduce the risk. Simply put, in the long run with a mere 2% of the global hashrate a miner can have a 95% confidence of being within 5% (+/-) of expected returns after 30 days. That is more than sufficient to hedge the operating cost risk. There is no economic benefit to more hashrate but miners want those multi-hour updates. They just want it and that probably isn't going to change.
It isn't a technology problem but rather a human "problem" (or maybe human nature is simply a better term). A miner's major ongoing cost is electricity and payment of that cost isn't due on an hourly or daily basis but monthly. Sub-month variance is not an economic issue. If a miner has sufficient confidence (mathematical term) of the expected return over a month long period lower variance doesn't improve the economics or reduce the risk. Simply put, in the long run with a mere 2% of the global hashrate a miner can have a 95% confidence of being within 5% (+/-) of expected returns after 30 days. That is more than sufficient to hedge the operating cost risk. There is no economic benefit to more hashrate but miners want those multi-hour updates. They just want it and that probably isn't going to change.
June 24, 2014, 10:00:03 AM
Just had an idea.
What if instead of mining just single valid block per 10 minutes, miners could mine (for example) 100 blocks simultaenously with 100 - times lower difficulty, which would then be treated as one single block in the blockchain ?
What i mean in detail:
1. Miners can mine X blocks simultaneously, all of which can be valid and added to the blockchain, but
2. Difficulty is X times lower (or insert proper difficulty calculation algorithm here)
3. Block reward is split by X. Each of the small-block rewards goes to the miner who mined it.
4. X adjusts automatically with difficulty Y. The bigger Y is, the bigger X becomes and the more pieces are produced each 10 minutes.
5. After all the small blocks are mined, they are merged back into one normal-sized block, which then is inserted into the blockchain.
This way, mining pools are not really necessary, every miner could mine solo (or in tiny pools). The biggest problem i see now with it is deciding which mini-blocks get included in the final big block.
Am I making any sense, or is this one hell of a crazy idea ? Please bear with me.
What if instead of mining just single valid block per 10 minutes, miners could mine (for example) 100 blocks simultaenously with 100 - times lower difficulty, which would then be treated as one single block in the blockchain ?
What i mean in detail:
1. Miners can mine X blocks simultaneously, all of which can be valid and added to the blockchain, but
2. Difficulty is X times lower (or insert proper difficulty calculation algorithm here)
3. Block reward is split by X. Each of the small-block rewards goes to the miner who mined it.
4. X adjusts automatically with difficulty Y. The bigger Y is, the bigger X becomes and the more pieces are produced each 10 minutes.
5. After all the small blocks are mined, they are merged back into one normal-sized block, which then is inserted into the blockchain.
This way, mining pools are not really necessary, every miner could mine solo (or in tiny pools). The biggest problem i see now with it is deciding which mini-blocks get included in the final big block.
Am I making any sense, or is this one hell of a crazy idea ? Please bear with me.
Jump to: