I just read the whitepaper's section on fees. I like the idea that presenting a reward can avert a tragedy of the commons because miners can roughly predict the cost to them to process the transaction and the probability they will succeed in mining the next block so they can claim the reward fee. It is then advantageous to them to take up the task if
reward > cost / probability of successfully mining the next block
But this does not solve the scale problem: The probability anyone can successfully mine the next block is inversely proportional to the number of nodes in the network, so as the network scales, so will transaction fees!
Is there a solution to this? I have a hard time believing people will really use a cryptocurrency long-term if it gets more expensive to use it all the time. At least in a centralized star-graph network, like we have with credit cards and banks, the costs are flat no matter how many points the star has.
Aside: What about this related problem: Say I am a miner who controls a bigger part of the network than anyone else. Can't I now offer lower fees because my probability of successfully mining the next block is higher? Could I drive other miners out of business because they can't process those transactions cost-effectively and thus can't generate the next block? Could I slowly take over more and more of the network this way until I'm a monopoly? Or does the problem solve itself because other miners can still beat me to the punch by creating valid blocks that exclude those low-fee transactions/I would never take on low-fee transactions when there are higher ones around?
Note: My first question is not about basic blockchain scalability; it is specifically about how scaling the network will increase fees. The second question is less pressing.