One thing that the below assumption doesn't appear to take into account is user demand for quick confirmations. I haven't fully thought out the game theory on this but here goes:

Example: if 75% of hashing power accepts 'medium' fee transactions while 25% is willing to accept low (or any) fee transactions, then a user paying a lower fee wishing to get their transaction included in the next block runs a ~75% chance that their transaction won't be included.

Users desiring the most reliably fast confirmations are better off by paying the minimum fee that a majority of miners will accept. Miners breaking ranks and accepting lower fees only affects users who aren't sufficiently interested in quicker confirmations. As long as a majority of miners 'collude', they will likely be able to keep the average fees higher than miners with the lower costs of operations might be willing to accept.

On Thu, Jul 30, 2015 at 12:00 AM, Adam Back via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> wrote:
On 29 July 2015 at 20:41, Ryan Butler via bitcoin-dev
<bitcoin-dev@lists.linuxfoundation.org> wrote:
> Does an unlimited blocksize imply the lack of a fee market?  Isn't every
> miner able to set their minimum accepted fee or transaction acceptance
> algorithm?

The assumption is that wont work because any miner can break ranks and
do so profitably, so to expect otherwise is to expect oligopoly
behaviour which is the sort of antithesis of a decentralised mining
system.  It's in fact a similar argument as to why decentralisation of
mining provides policy neutrality: some miner somewhere with some
hashrate will process your transaction even if some other miners are
by policy deciding not to mine it.  It is also similar reason why free
transactions are processed today - policies vary and this is good for
ensuring many types of transaction get processed.

Adam
_______________________________________________
bitcoin-dev mailing list
bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev