The post that started this thread was about thought experiment of Twitter killing off their API. So that third party clients wouldn't work is a given. And since they wouldn't work, there's no reason why Twitter couldn't show a captcha many more times after signup.
1) Allowing a delay between creation and posting e.g. Buffer.
2) Posting updates to Twitter whenever CMS content changes e.g. Wordpress.
3) Large companies who need tools that work across all social media sites e.g. Hootsuite.
4) RSS -> Twitter.
The NYTimes automating tweeting a link to an article after its published provides value. Yet that same functionality allows spammers to tweet links to their spam automatically as well.
Likewise with scheduling, etc.
Anecdotally, before 2010 (when scheduling tweets became popular) I remember Twitter being much more active and discussion oriented, hence I paid more attention to my feed. Now i only use Twitter to search for specific things (ie is AWS down)
Edit: Reminder that Twitter filters out posts that don't receive much interaction.
Much of the platform is dedicated to users wanting to follow products, brands, services etc. For them the services I listed are a critical necessity. You can argue all of it is spam but then by that definition so is most of the web.
If twitter banned all API write access, what would be the problem?
The vast majority of people would still use Twitter via the app or their website, and it would obviously solve a lot of spam problem. But do we really need automated emergency alerts that don't filter through a human?