Larger block sizes allow more transactions, so although per-tx fees will be lower, there will be more transactions.

The computational costs of larger block sizes are borne on the entire network, however the effects are most severe for miners as larger blocks tend to increase the profitability of large geographically-close pools in comparison to small pools [1]. This is why segwit is only an increase to ~2.1MB.

[1] http://bitfury.com/content/5-white-papers-research/block-siz...

But the relationship between block size and per-tx fee is based upon the amount of competition between txs, not absolute tx volume. So if the block size is raised high enough to accommodate the current rate of txs with no competition, the fees don't just fall proportionally: they fall to near-0.

The higher the tx backpressure, the higher the fees. The larger the block size, the lower the pressure regardless of overall tx volume.

Miners could also simply decide to not include any transactions with a fee smaller than X. Transactions can't "shop around" for a miner, the first miner to mine a block wins.

Yup, it's not an exact tradeoff, and near-0 it breaks down entirely. At zero it's worse than that, because of the diminishing block subsidy schedule, having zero fee pressure will reduce the incentive for mining, and therefore security.

However, miners can't simply apply the limit to themselves without a consensus rule. There is code in Bitcoin Core to apply a soft limit, but as long as there is any fee pressure, ignoring transactions leads to a disadvantage compared to other miners, so in practice all miners tend to always use the maximum permitted block size. Most proposed schemes for increased or variable block size still enforce the same size across all miners for this reason.

> however the effects are most severe for miners

Incorrect. The most severe effects are on the nodes.

The utility of bitcoin is damaged by the limited maximum transaction rate. It's no help to miners if bitcoin is replaced by another cryptocurrency, which is looking more plausible now than it ever has. In the end they could be collecting fees off a lot more transactions if the block size was larger anyway.

Is it possible a majority of miners sell their block rewards immediately and have little investment in the long term success of Bitcoin? I know they have hardware investments to recoup, so I'm sure there are many sensitive variables in their financial equations.

If their hardware is viable for an altcoin then it seems like increasing the Bitcoin block size would not be in their best interest. Why not maximize shortterm Bitcoin profits and hedge long term profits with altcoins?

Asics can't be redirected to different alt coins that have a different proof of work. They will run to recoup the costs. That is why bitcoin will be more secure even if prices and transactions drop from a move to an alt.

Miners are playing a game of chicken to milk the fees while hoping users don't swerve to another chain.

Some Asics can mine both Bitcoin and Lightcoin though not at the same time. From a cost perspective they are slightly more expencive, but miners could swap over 2 years to a new coin with minimal issues limiting their long term risks.

I don't see an altcoin beating out bitcoin soon, but if one started gaining traction it would probably have a proof of work system compatible with most miners' asics.

All major miners have spent loads of money purchasing mining hardware, and they all need to run that hardware for a long time just to break even. Higher future Bitcoin prices is critical to their ROI.