The thing that killed micropayments back in the early 2000s (when Scott McCloud was calling for them) was transaction costs. Putting a paywall on
every click is a sales and user experience
nightmare. Keep in mind that the reason why everyone uses subscription billing nowadays is that you never have to bug the user to ask for more money. You just ask once and get paid indefinitely.
There are micropayments that do sort of work - for example, cellphone plans that bill per megabyte of "data". But even then that only works because mobile operating systems and apps do not provide a way to anticipate data costs. There is no cost displayed next to each link you tap or YouTube video you watch. If there was, nobody would use their phones. Usage-based billing is actually one of my pet peeves for exactly this reason.
What we really want is an all-you-can-eat model that just gives us what we had back in the Advertising Will Pay For It era of news. Except that will never exist because everyone knows how platform capitalism works. Once you condition users to pay a fixed amount per month, then all the creators on your platform - or, in this case, news organizations - are playing a zero-sum game. Because they are being paid out of one pot and the only way to make that pot bigger is to do the platform's advertising job for you. This is what happened with Spotify and Netflix, with the added intrigue of the platform playing artists off one another with original content and exclusives.
On a more technical level, transaction networks are expected to provide fraud insurance, and that bills at a tune of 30 cents per transaction. Any mechanism to shave pennies off your wallet will be abused to commit untold amounts of tiny fraudulent transactions. For example, imagine installing a browser extension that, in the background, starts loading up your news articles without the user's knowledge. That's the sort of thing that will kill a microtransaction system long-term.