Using something from the block you are making as well as from the previous block means you can mine harder to try for a winning block.
Just keep changing something to get a different merkle root until you get one that makes your block a winning block. Spend enough computing power on it and maybe all blocks will be winning blocks!
-MarkM-
Just keep changing something to get a different merkle root until you get one that makes your block a winning block. Spend enough computing power on it and maybe all blocks will be winning blocks!
-MarkM-
Yeah, that's basically mining.
But the process would delay your block too much to hope it being the first accepted block.
That could be a nice race to watch.
Well no it would be trivial to exploit. Say the odds of hitting a lucky/super/gold/handOfGod block are 1 in 10,000. You could just keep constructing merkle trees (by changing the coinbase) until you found one which was a "winner". It would take on average 10,000 attempts. A good CPU can do that in a fraction of a second.
Essentially someone would release an optimized miner and all blocks (as in 100%) would get the super rate. The advantage of using the prior block is that it is something you can't control. Even that however does allow a large pool (say one with 20%+ of hashing power) to "cheat" unless the odds vs the "bonus" are high enough that it really has no material increase in overall reward.
So yes of course you could keep constructing other merkletrees but finding one which makes the block hash (i) below the target and (ii) complying with the super block rule would take 10,000 more times than just (i). So by the time you found the super block you would be 10,000 blocks behind the blockchain.
Um you are forgetting that one can use the same coinbase multiple times. You wouldn't find a "winning" coinbase/merkle tree and then try one hash and throw it away. It takes a lot more than 10,000 hashes to find a block on average. Say at 1,000 difficulty it would require 4,294,967,296,000 (2^32 * 1,000) hashes to find a block and that block has a 99.999% chance of being the base coins. Now why wouldn't you just calculate an additional 10,000 merkle trees (a million hashes at most), throw away the ones which don't produce a super block and only use the "super" merkle trees. It will still take you 4,294,967,296,000 hashes to find a block but when you do it will be worth 250x as much.
You are adding <0.001% additional work and getting 250x as many coins. Nobody would mine any other way. If it didn't happen initially it would as difficulty rises. The number of coinbase calculations required to find a "super" merkle tree remains the same however the number of total hashes necessary to solve a block rise linearly with difficulty. So the profit from the exploit would rise with difficulty.
