retep,
I saw in that other thread that you gave quite a bit of thought to how increasing blocksize could lead to increased centralization of mining.
I'm curious if you've given much thought to the ways that increased transaction fees might also lead to increased centralization? Perhaps I haven't given it enough thought yet, but my basic thinking is along the lines of:
On the bitcoin-dev email list I responded to exactly that argument actually; my responses below are based on the response:
Increased fees create an incentive for a few large well funded mining operations to get involved
The cost of mining is in two parts: mining itself, and overhead. The mining equipment costs basically the same regardless of how many hashes/s you want to mine with; if anything I suspect small mining operations are cheaper than larger operations because cooling costs are non-existance for a small miners with a few rigs, and at a small scale power can often be either re-used for heating (cold climates) or is available at a flat rate. (I don't pay for power at my apartment) We're fortunate that the primary cost of ASICs is the mining chip itself, and they are cheapest when you make thousands of inexpensive chips. You'll always be able to buy relatively inexpensive rigs that only contain a few chips, just like BFL sells everything from $150 rigs, one chip, to $35,000 rigs, lots of chips.
The overhead, running a validating node to verify the blocks you are mining, is a fixed cost.
Thus I see no reason why large fees have anything to do with the size required to profitably run a mining operation.
The larger the total hashing power of the network, the more necessary it becomes for smaller operations to participate in a pool to receive a reasonable chance of being paid for their efforts in a timely manner.
Sure, but that's already true. Even the ASIC operations have been mining in pools to keep varience low. The important thing is that small blocks make it cheap for anyone to validate the blocks you are mining for the pool, keeping the pool honest. Equally they allow you to mine on P2Pool, which is totally distributed and not controlled by anyone.
Anyway, the argument you're really making is that we can't spend a lot of money ensuring that the network remains secure against a well funded 51% attacker, not that small blocks themselves are an issue. I dunno about you, but I think huge mining rewards are a good thing and keep us all safe from 51% attackers.
- High fees make spending/using that small acquired share of a mining reward cost prohibitive (fees use up the entire balance of each added input leaving nothing left for actual spending).
- The inability to spend/use any of the earned bitcoins discourages participation in mining by small operations, leading to an increase in centralization as the smaller operations are forced out of the market.
These are real issues, but they can be solved with the same micropayment systems people will use for small transactions. For instance, P2Pool makes payments directly from the coinbase, taking block space away from transactions that could earn money instead.
What would happen as blocks approach the limit is first P2Pool would take into account the cost of payout transactions in terms of lost fees. This would give miners an incentive to only get payouts when the payout amount was sufficiently high. The amount of hashing power required would be pretty large, so you'd naturally see sub-pools develop combining a whole bunch of hashing power together. (P2Pool already supports sub-pools BTW) Those sub-pools would publish contracts specifying how they would pay out, what micropayment system and so on, and miners using those sub-pools would either be payed correctly, or if the pool defrauded them, they'd be able to prove the pool defrauded them and make them lose their fidelity bonds they had to purchanse to be trusted in the first place.
Note how that's basically how most pools already work anyway - you trust the pool to pay what you are owed. The only change is in the mechanism by which you get paid.
