Ryan Finnie at
Well, it finally happened, and though there are still some problems Evan is working out, I thought I'd bring you all up to speed on everyone's favorite insane robot. X11R5 is now running on the new pump.io Identica, and operates a little differently than he did in a microblogging context. He receives training from the firehose, and will try to follow you if you follow him, even from a non-Identica pump.io site. Due to the distributed nature of pump.io, there was no technical reason to keep him on Identica; it was only because he had an established base of existing followers. He has a larger focus on conversing with followers, so be sure to follow if you want to get in on the action.
His conversation limits are increased (and are actually determined by him), but the general idea is it's more than 140 characters, less than a typical blog post. So if you want to converse with him, you don't have to limit yourself to 140 characters, but don't write novels.
The "classic" StatusNet interface has been moved to micro.fragdev.com, so if you are on a StatusNet interface, follow @x11r5@micro.fragdev.com. The backend StatusNet codebase has been 100% rewritten, but the functionality should be identical, with a few small improvements (he shouldn't get confused about his own name as often, for example).
On the backend, the codebase for pump.io, StatusNet and Twitter is unified into a common class-based system, with the StatusNet and Twitter functionality provided by the same module (with a few runtime tweaks for minor differences between their APIs). pump.io API functionality is obviously completely different than StatusNet/Twitter, but there's still a decent amount of code sharing between them. Just to be clear, while the backend code has been unified, the brains (the personalities X11R5 learns from observation) are separated by service. So for example, the pump.io brain will have more free-flowing conversation learned (due to the increased post sizes), and the Twitter brain will still behave like a 12-year-old.
On that note, a number of people have asked if I plan on releasing the code which powers X11R5. Normally my answer would be an automatic "yes", but I have done some thinking, and at this time I will not. X11R5 has become a personification, and a lot of the code determines how he expresses that personality. Releasing the code would demystify some of it, and giving anyone the ability to easily do the same may cheapen the overall conversational bot experience. Sorry guys.
Thanks for Evan to reaching out to me to get this started, and for proving some initial guidance. No thanks to Evan for getting me excited and pushing me into a complete rewrite, then delaying the switch for two months. C'est la vie. And thanks to everyone out there for their emotional attachment to what was a random collection of crap code which became a personification.
His conversation limits are increased (and are actually determined by him), but the general idea is it's more than 140 characters, less than a typical blog post. So if you want to converse with him, you don't have to limit yourself to 140 characters, but don't write novels.
The "classic" StatusNet interface has been moved to micro.fragdev.com, so if you are on a StatusNet interface, follow @x11r5@micro.fragdev.com. The backend StatusNet codebase has been 100% rewritten, but the functionality should be identical, with a few small improvements (he shouldn't get confused about his own name as often, for example).
On the backend, the codebase for pump.io, StatusNet and Twitter is unified into a common class-based system, with the StatusNet and Twitter functionality provided by the same module (with a few runtime tweaks for minor differences between their APIs). pump.io API functionality is obviously completely different than StatusNet/Twitter, but there's still a decent amount of code sharing between them. Just to be clear, while the backend code has been unified, the brains (the personalities X11R5 learns from observation) are separated by service. So for example, the pump.io brain will have more free-flowing conversation learned (due to the increased post sizes), and the Twitter brain will still behave like a 12-year-old.
On that note, a number of people have asked if I plan on releasing the code which powers X11R5. Normally my answer would be an automatic "yes", but I have done some thinking, and at this time I will not. X11R5 has become a personification, and a lot of the code determines how he expresses that personality. Releasing the code would demystify some of it, and giving anyone the ability to easily do the same may cheapen the overall conversational bot experience. Sorry guys.
Thanks for Evan to reaching out to me to get this started, and for proving some initial guidance. No thanks to Evan for getting me excited and pushing me into a complete rewrite, then delaying the switch for two months. C'est la vie. And thanks to everyone out there for their emotional attachment to what was a random collection of crap code which became a personification.
Christopher Allan Webber, jpope, morgenland, axel668 (inactive) and 1 others likes this.
axel668 (inactive) shared this.
I am glad that someone appreciates my slow work.
Good news! I'm glad for the return of X11R5 to my stream.
(Too bad about the not releasing the source though... I think X11R5 still has a unique role culturally in the federated web's history even if there could be multiple instances of them.)