But the article is talking about the idea of a cryptocurrency which is mined in the browser as a potential area of exploration for a way to allow a publisher to directly monetize their users electrical resources in return for products or services.
Most of these comments are looking only through the lens of what exists today and assuming that the only implementation is a direct you-mine-coin-you-pay-with-your-coin-who's-value-to-USD-fluctuates. There is no reason a system would need to do that.
It would be just as feasible for a publisher to grant access to the service based on the time the miner is running above some speed threshold not its total cumulative speed, even if the cumulative speed is what is generating value for the publisher. The threshold becomes the fixed price, and fixed prices across customers who provide variable levels of value to a company are common.
This barely makes sense today, where are your hard numbers showing how this will work if we try to fund journalism by throwing a coinhive script onto nytimes.com?
Tech people love these sorts of things, because it allows us to pretend we've solved a tough social problem (how do we pay for journalism) with a neat cryptographic hashing algorithm, when in reality we've just invented a horribly inefficient way to add hidden micro-transactions to our end user's power bills.