First part of the process is making an email. Seems simple enough EXCEPT all major free email addresses require you to setup a secondary email to verify your primary email and if you don't, chances are you can't use this email to verify your account for account creations.
This is obviously done in hopes that you link an email that is linked to you in your verification step. This has nothing to do to ensure protection or get rid of automated accounts because it's easy for automation to do this secondary step.
So, you're forced to create two new emails from two different email providers and cross-link them.
Once that's done, you've got to setup your X account which is a nightmare in itself because their new "confirm you aren't a bot" thing is garbage with 10 questions you have to answer. It took me two tries to get it but since I took so long, it didn't properly load when I got it and I was forced to do it two more times to get it. That's answer 40 questions to prove you aren't a robot.
The irony is my X account got banned for being a bot even though I've never used a botting program ever and I don't even use AI to post.
The first platform to figure out a way to skip over all of this and just charge people $5 (payable in XMR) to make an account will do great. No subscription, just a single flat fee. Layer it over the original system and advertise it as a permanent guarantee against shadowbanning and any kind of deboosting. Elon could probably juice Twitter's earnings by a very significant margin by going the 2000's MMO route and having one-off "Twitter Basic" passes sold in pharmacies and WalMarts nationwide, which give you the anti-shadowban benefits of being a verified user without any of the extra stuff.
The "anti-botting" measures are all solved within a week by people whose full time job it is to circumvent them, so the only thing they do is constrain legitimate users who aren't willing to link their IDs (which is often the intent, but ensures that only boomers and idiots with nothing to say end up joining).
That amounts to, essentially, a cryptographic captcha. Brave Search uses those, and they're definitely better than the conventional ones.
You can't really use the compute for anything practical, both for technical and PR reasons (nobody wants to burn up their GPU mining crypto for someone else, and it's incredibly inefficient without a dedicated setup - there's a reason only malware considers it worth the tradeoff), but it's still not a terrible idea. That said, the CP spammers are human (barely) antifags with too much free time, so this wouldn't stop them.
We implemented crypto captcha on 8chan clear back in 2021. Its set at a mid difficulty that takes a low end PC or IOT about 5 minutes to pass where the average user gets through in about a minute. It stopped about 90% of botspam, with only the hand-guided turbo autist spammers still getting through.
I get the feeling that this is going to level the playing field significantly WRT Cloudflare-style monopolies. While conventional captchas are (by design) a treadmill that new ML models will be able to solve in a year, proof of work captchas don't go obsolete until Moore's Law eats them, and even then all it takes is a parameter change to get them back to the expected level of performance.