)I'm going to chime in with my opinion, if I may...
Captchya's quite simply don't work how anyone wants them to. All of the time and effort spent engineering them is easily reversed to defeat them. They are a backward way of doing user authentication and verification, they are difficult for users with visual disabilities to see, and obviously the audio method doesn't work with deaf or hearing impaired users. Anything that I have seen in use thusfar other than the afore mentioned is far too complex for your "typical" user to understand, with click throughs or multiple transactions required.
The least defeated solution here involves human intervention. I believe it would be easier to auto-moderate a forum based on karma/censors/word comparison to the root topic, than it would be to auto-descriminate a registration robot versus an old man with glaucoma who has failed the captchya test 5 times (for example.)
As much as the best way to ensure this is to monitor each registration with human eyes and allow them to make the judgement call, I also understand that the reasoning behind a Captchya is to remove an administrator from the process almost entirely. Having not installed Beta2 myself, I am unaware if there is a maximum number of times a potential registrar can fail before a moderator or administrator is notified.
Realistically, keeping it simple through automation is the best way to provide registration. I forsee a more intuitive system to more adequately prevent the registration of robots or masses of unwanted users. Using a trifecta of logging a registration I.P., creating a minimum threshold of allowed registrations from this I.P., and monitoring the introductory post(s) of a user to establish forum credibility.
I know that karma has been discussed to death, however I think that semi-automated/mostly moderated user credibility ratings are the future. With forums being as popular as they are, and getting more and more commonplace on even the smallest websites, I forsee the only true way to catalog known and recurring spam registrations to be globally and universally accessable, similar to using an internet database to get track titles off of a CD.
What I am proposing, is a centrally hosted database where spam user names, I.P.'s, and any other pertinent information are saved through the administration portion of a forum. If companies can spend thousands on picture hosting websites so people can look at pretty things, surely someone has the capital to host a website that's providing a functional and vitally important service.
Now of course, the downfalls are that without adequite follow through, anyone could add anyone to this database. My only split second way around this is to only allow reporting of spam users from administrators/moderators of the originating forum. The central database would be linked to the moderator/administrator control panels, and allow a send/receive type of functionality. Basically a mod would be logged in, having priveledges to report a spam bot, send trigger that report which is sent to the central DB. The moderators name, I.P, and the website submitting the bot would be logged. In the ACP a threshold can be set as to how many times a name or I.P. has been reported for you as the administrator to decide to automate the spam-robot avoidance and removal operation.
Does that make sense or am I just typing in latin?
