Miners play a zero sum game within a positive sum economy. They compete with each other, not the economy. Rising utility is the reflection of a positive sum and a natural consequence of trade.
It has been argued that blocks mined in a period of rising price produce outsized returns for miners, at least until the difficulty adjustment. This idea is based on the common failure to understand that market prices are not predictable. Wagers on price change are speculative. There is no reason to assume that Bitcoin speculation is any more or less effective than any other. To the extent that a rising price is generally predictable by miners, competition predict it, invalidating the idea of any inherent outsized return.
Bitcoin mining investment on the other hand is based on the predictable relationship between profit and competition over time. That relationship predicts that the average of all mining approaches the market rate of interest. As with all markets, shorter time periods are unpredictable in price and longer periods approach market returns. Ultimately time preference controls the market rate of investment return.
So how does a miner achieve outsized returns? It cannot be done with side fee agreements. There is only one way to make a higher-than-market rate of return, which is to have a below average cost of hash power for the coin. This is achieved by either taking advantage of pooling pressures or through superior operational efficiency. Because of the zero sum property, these are offset by lower-than-market rates of return by other miners. The premium therefore declines for an honest miner above 50% hash power, to zero at 100%.
However other miners will eventually exit as their capital seeks market returns. This would leave one miner, bound to market returns. In other words, making outsized returns requires others from whom to capture those returns. The highest return that can be sustained is a function of the greatest opportunity cost others are willing to sustain. This is a function of differential reward utility, as discussed in Threat Level Paradox.
By limiting dividends to market rates of return and reinvesting all remaining reward, a miner can maintain a constant hash power and thereby obtain market returns against a capital base proportional to Bitcoin capitalization. Reinvesting dividends increases hash power and liquidation decreases it. Grinds are liquidated by taking each device offline as it becomes a net negative producer, or discounting those future returns by selling the device.
Mining rate of return on capital is dependent on time preference alone. The relationship between the economy and miners is further explored in Balance of Power Fallacy.