@theprofileth, the problem of equal share value was first exposed in the yPool forum in this thread
, in which someone wrote:
v6 has stayed pretty close in xpm income to v4 but with a lot more solved blocks. based on the total xpm income for the past 36 hours, i have to say v4 makes the miner more primecoins and v6 solves a LOT more blocks for the pool.
there is an equal amount of processing power allocated to each version. both versions of the miner are running with default settings.
to compare numbers:
v4 has 41 blocks found and 409 primecoins credited to the account.
--basically even with solo mining if i add the donation to the total
v6 has 51 blocks found and 370 credited to the account
--obviously this version is MUCH better at finding blocks and better for the pool over all. unfortunately, the share handling is killing this miner's profit potential.
You can see that the advantage of v4 versus v6 (versions of jhPrimeminer, by Mumus) was just 409/370 = 10.5 %. Also, it was counting several lengths of chains, say 6,7,8 and 9. So, the difference in primecoins if we count just consecutive lengths could be about 1.105^(1/3) = 3.4%.
In any case, it is less than 9/8 = 12.5%, and less than 8/7, 7/6 or 6/5. So, with just a share value proportional to the chain length
the problem will be solved.
And a proportional value is a better measure of the computing work contributed by each participant, whilst an exponential valuation will depend on the luck each time, mainly for the small workers. Also, exponential valuation will give huge variance to the pool (yPool cuts its exponential valuation for lengths greater than 9 to avoid an unaffordable variance for the pool operator).