Additionally, I don't believe bot detection is an inherently hard problem. And if every bot detected means the bot creator has to pay another $8 for their next bot, then bot detection becomes profitable for Twitter.
I doubt this ends up being the cheapest way for Russia to influence US elections.
If he didn't think he could improve on the status quo for Twitter, the thing to do would be to trim as much staff as possible to cut costs and then continue with the current ad revenue based model, without making any changes or introducing new features. Twitter was making on the order of $5B/year in ad revenue. Pushing costs below $5B/year should not be difficult given the nature of the service provided.
Alternatively, if what you say is true and he doesn't think he can make it work, he could try to "back out" by flipping Twitter to another buyer, or just take it public again. Revealed preferences suggest he thinks he can outperform those alternatives.
Additionally I think "verified users get priority in replies" has the potential to be absolutely killer. Everyone knows that clout chasing is huge on Twitter. I could easily see a spiral like so:
* Initially 5% of daily actives buy the check marks, mostly just to own the libs (Fox News loves Elon btw). But the newborn blue checks are everywhere in replies. They're accumulating followers like no tomorrow.
* Others begin to follow suit. $8/month is a small price to pay for clout. Remember, that's what they're all there for. "If anyone tells you they don't want more twitter followers, they're lying."
* 10% of the userbase has bought a blue check. 20%. The anti-Elon folks are holding out. But their voices are growing quieter and quieter.
* The entire tenor of the site begins to change. The loudest voices are the Elon fans who bought the check mark early. His politics become dominant on the site.
* Eventually the anti-Elon folks either ragequit or end up purchasing the checkmark. Petty tyranny and clout chasing are kind of their thing. It's a case of irresistible force meets immovable object. Endgame seems hard to predict.