 |
February 09, 2014, 09:31:13 PM
Last edit: February 09, 2014, 09:41:38 PM by bfire |
|
There are near-term scalability issues with the block size, primarily because there is a disincentive to mine large blocks due to the transmission time. The effect is small but real enough to create an economic problem that is already beginning to manifest (a miner called Discus Fish is submitting pathetically small blocks, presumably to overcome issues with transmission time).
Why not create a 4% penalty for small blocks that would negate any benefit from "cheating" with artificially small blocks?
Basically I am proposing this logic get inserted into the protocol
If (CURRENT_TRANSACTION_COUNT < MEDIAN_TRANSACTION_COUNT / 2)
{
BLOCK_REWARD = 24;
ACCUMULATED_PENALTIES++;
}
else
{
BLOCK_REWARD = 25 + ACCUMULATED_PENALTIES;
ACCUMULATED_PENALTIES = 0;
}
Where MEDIAN_TRANSACTION_COUNT is the average over the last N blocks
This way the extra 1 BTC gets thrown into the next valid fullsize block.
Even if a block is found very quickly, the miner can include a bunch of free transactions to get to the 1/2 size threshold. If not enough free transactions are available, oh well it's only 1 BTC lost occasionally. Everyone is on the same playing field, so there is no net loss to the mining community.
One criticism is that a miner who solves a fast block might create fake (wasteful) free transactions to get to the threshold, but there is no advantage to this vs. accepting available free transactions. In the worst case, it would represent a very small bloat (< 3%) in the blockchain size growth rate.
You could get more fancy by prorating the penalty vs. the size of the block (i.e. smaller blocks = larger penalty)
This tweak in the block reward system would also have the effect of increasing the theoretical max block size, by removing the disincentive to mine large blocks. In 10 years (3 more halvings), this tweak will become less relevant as the ratio of transaction fees vs. block reward grows.
I realize there is a formal path for suggested enhancements to Bitcoin, but I'd like to get feedback first in case it's a dumb idea.
|
