I'm following the logic up to the point where the hopper leaves the pool. (if the pool finds a block before he leaves everyone wins). Now assuming he's left... hash rate decreases... hopper has generated a large amount of shares...
I'm still not seeing how it decreases the chance of finding a block for the pool... lets do some mental math here...
If he leaves, he'll get paid for whatever shares he did submit while he was mining. The shares he submitted 3 hours ago before leaving (if no block has been found yet) have the same value as someone else who was also mining at the same speed/difficulty/etc and stuck around and didn't leave. Your mining right now does not change the value of the shares you submitted 3 hours ago.
So an average round, you have 15% of the total shares going to the hopper, and 85% of them going to the normal miners.
3.75 btc to hoppers.
21.25 btc to miners.
In the average case miners lose 15% of the reward for the block to the hoppers.
No reward has been "lost". The hoppers put in 15% of the work and got 15% of the reward. That's totally fine.
Case of extremely lucky round:
2 blocks back to back (and we'll even say 10 mins before that second block is found).
Total shares for hoppers : 6,000,000,000 shares
Total shares for legit miners : 60,000,000 shares
so on this round the 1% of the shares are going to non-hoppers with 99% going to hoppers - who then probably stay at the pool until some other pool finds a block.
25.75 to hoppers.
0.25 btc to miners.
Again in this example they got their fair share. No problem.
Case of extremely bad luck:
Takes 9 days to find the block.
Total shares for hoppers at pool over 10 mins : 6,000,000,000 shares
Total shares for legit miners over the 4.5 days : 77,760,000,000 shares
7% of shares to hoppers, 93% to miners.
23.25 btc to miner
1.75 btc to hoppers
What you are ignoring here is that the hoppers leave when the round goes long, not before. On average, a full time miner who never leaves will see short and long term blocks and it'll all average out so they get block reward * 1/D reward per share submitted. But a hopper rocks the boat by only mining until a round goes long. So on short rounds he gets paid more than B * 1/D since the total of total shares was < D. In long rounds he gets his proportional share except he abandoned ship when total shares submitted so far = D, so the miners left are mining with a block round going down on each new share submitted but the hopper is over on some other short pool mining at a per share value that is high.
There are threads dedicated on "how to hop" most profitably. Worth a read if you aren't seeing why hoppers in proportional pools hurt loyal miners who never leave.
My only point with all this is that it doesn't really hurt the legitimate miners in any real way, given the differences between their actually hashing and the inflated estimation of hoppers (2 orders of magnitude). Yes they lose in average round 15% of income - which they could just as well lose if the pool grows to 115gh/s --- while the pool loses the added income from more short rounds.
Nothing is lost in the average round. The hoppers put in X% work and got paid X% of the reward. As it should be. Hoppers don't care about the average case, they want to benefit from the short blocks while leaving loyal miners to work off the less profitable final shares of a long block. This is why some types of PPS systems (SMPPS, RBPPS, etc) end up collapsing. Too few people will work on the shares needed to finish the current block when their reward for doing so is less than mining someplace else. CPPSRB is one method of solving that specific issue by paying most recent miners first, so if you keep mining you'll get paid the full PPS value and there's no incentive to abandon the pool.
So all you're really doing is limiting the potential benefits to the proportional pool of the occasional botnet or high hash rate hoppers hitting it.
The only "benefit" is finding blocks faster, which doesn't increase earnings to loyal miners because they are the ones mining alone in long rounds when the expected reward of each new share they submit is less than B * 1/D since the total number of shares submitted is already over D.
If you prefer pools that can be hopped or you prefer to hop yourself that's fine. You have the freedom to do so. It just bugs me to see people misrepresent systems like DGM by claiming it does the opposite of what it really does.
