I propose the following rule to determine the block size limit, once the block reward is low
The block size limit would increase (or decrease), by X%, if total transaction fees in the last N blocks is Y Bitcoin or more (or less).
......
I am aware miners could also manipulate fees by including transactions with large fees and not broadcasting these to the network. However why would miners in this scenario want to manipulate the limit upwards?
The fear is that a cartel of big, centralized, have-huge-data-pipes miners would drive out smaller miners by forcing up the block size high enough so the smaller miners have to drop out.
With the current mining dynamics my proposal would not be suitable for the reasons you suggest. I merely suggest it as an eventual objective for when the block reward becomes low and hopefully mining becomes more decentralised, competitive and fee driven. If mining doesnt develop this way then Bitcoin may not be sustainable in the long run anyway. We could still keep another maximum of maximums block size limit based on bandwidth considerations and then this transaction fee targeting based limit system could operate within this.
Whatever happens to the hash rate total mining revenue represents the economic value of network security. For example currently the security of the network can now be considered as 25 bitcoin per block, regardless of the large hash rate increases as in theory 25 bitcoin per 10 minutes is the cost of mining. In the future the value of the total transaction fees will represent the network security and therefore the dynamics which determine the fees will be vital. Having supply potentially grow exponentially forever may not be appropriate.
The above proposal could be a good framework for a discussion on how the dynamics for the transaction fees could be determined in the future. The system is kind of an aggregate transaction fee targeting scheme. For example a target of 1 bitcoin per block is around 50,000 bitcoin per annum or 0.24% of the eventual total supply per annum. Deciding if this is a suitable level would be difficult. Is 0.24% high enough to secure the network or should it be 1%? What if the number is too high, we create an arbitrarily high amount or environmental damage?