I think it would be beneficial to discourage the auto submission scripts from submitting random posts. Currently, it is too easy to run automated scripts to make random submissions across any permitted subverse. It's possible to just take a text file, organized by a line of title and message body, and make timely submissions without human interaction. A slightly more advanced user could simply scrape topics(also very easy to do) from mainstream media to feed the auto submission script.
Recently, @Plant_Boy has observed a potential bot that is spamming across multiple subverses(archived) with the title "We were unable to suggest a title", which is a strong indicator of an automated system that is diluting the contents on Voat. I have also noticed the possibility of anon subverses been run by automated scripts. A large number of /v/AskVoat may also exhibit this behavior where those submitters exercise zero interaction with the replies in the comment sections.
So I think it may be useful to discourage automation scripts from making random submissions. This could be done by pre-generating and periodically regenerating/flushing a collection of captcha images with noisy and distorted alphanumeric values. These captcha images should be reasonably easy to solve by a genuine user upon submission; it should also be somewhat computationally expensive to beat within certain time frame.