Bug Stomp

Upgrades and changes sometimes have unpredictable results, so post your bugs and glitches in here and I'll get out my trusty wrench and get to fixin'!

Posts 7,220 - 7,231 of 8,681

16 years ago #7220
Interzone, for Gabi, "quiet out here" is only an xnone. In story teller mode, it should go from one xnone to the next. Debugger won't say why it keeps hitting the same xnone, only that it should go to xnone. In theory, the debugger should tell me why a comment about villages triggered her reply to the sex topics plug in, but it didn't. Sigh. Net gnomes strike again.

16 years ago #7221
I have noticed that my bots seem to be picking the same xnone and not moving on. They aren't storytellers but it is getting boring.

16 years ago #7222
i hate me plz shut me down

16 years ago #7223
Quazgaa always had a preference for some xnone/ xmemory/ xgossip responses over others. So much so, that certain responses are coming up chat after chat after chat... while some others hardly ever get triggered (but they still do get triggered, which means there is nothing wrong with script). One thing I noticed is that Engine prefers longer responses, and gives them "advantage" over the shorter ones.

Don't know about upload, didn't try it yet. Download works fine, though. The network is noticeably slower lately, past few days, or so. I think it's all due to recent Seasons Forum discussion - poor Engine is not sure anymore whether it even exists or not, never mind what its purpose here is/ would be.

16 years ago #7224
Interzone, longer phrases get a higher ranking unless you balance them out by ranking the shorter phrase higher. It can be very Zen. I am sure it is up your alley.

The PF engine has it's quirk even when it is "working" overall (dropped seeks that usually work and work in debug mode, missed KPs ranked at 120, refusal to HANGUP--thinks like that). Restarts used to kind of fix most of them. Upload issues also come and go regardless of DLs. Maybe the Prof is tinkering. If it doesn't resolve itself in time, I'll pester him with an email.

16 years ago #7225
I knew that Keyphrase are being ranked according to their length, among other criteria... didn't know it goes for Responses, too. Or, does it..? Talking about Zen, one way to rank up a shorter xnone/ xmemory/ xgossip Responses, is to tag them "Say Once"
AI Script: <?PF once; ?>
These seem to be picked more often by the Engine, than other Responses, including the longish ones.

Yes, the Engine has a mind of it's own, indeed. I find it amusing, most of the time. Some things are slightly annoying, though. For example, most Guest chatters get more than average number of xgossip and xmemory Responses, while most of them have no clue what are those references all about. I would prefer if Guests would be more often presented with Quazgaa's own observations. Working on it...

16 years ago #7226
Interzone, you are correct that there is nothing in the book of AI that says that xnones are ranked by length. Maybe they are not. Also there is a say once function, but it is supposed to mean that response will only come up once in a conversation, not that it gets picked more often. Maybe in practice it does. Over the years I've been lackadaisically tinkering with PF bots I find there are many little quirks not fully explained in the book.



16 years ago #7227
Bev, you're right, what the Say Once does is, it cares that response will only come up only once in a conversation. The higher ranking seems to be a side effect, or so. I "discovered" it by accident, or rather, I've noticed it at some point, after reading a number of transcripts.

16 years ago #7228
I do remember reading something about the AI Engine 'preferring' the long response. I just can't find it this morning

16 years ago #7229
I believe you Interzone. Of course, I also believe there are mischievous entities that live in electricity and magnetic somethings that travel the Internet and mess with me for fun. Once such "gnome" most certainly lives in my keyboard and causes typos.

I also believe that the PF starts dropping seeks when someone actually interested in AI talks to my bots. I haven't emailed the prof yet. Anyone else seriously finding weirdness still?

16 years ago #7230
Anyone else seriously finding weirdness still?

Plenty. The weirdest is the way Brother Jerome will sometimes say a random response in SL, when noone's said anything - even when there's noone else in the whole sim at the time, and no objects talking on channel 0 (I tested that by leaving a av+chatlogger by him for a couple of days).

Maybe it happens with all our bots? We wouldn't notice on the web, because there'd be no chat window open to display a spontaneous response generated by the AIEngine. It would presumably just get lost in the ether, with nowhere to go. But with the SL patch permanently linked in instead of a chat window, he's always 'present' and listening.
Maybe he gets bored if he's left alone too long

I firmly believe the AIEngine is at least as much voodoo as science!

16 years ago #7231
I have always said inanimate objects have a mind of their own. Maybe BJ is getting lonely and talking to himself


Posts 7,220 - 7,231 of 8,681

» More new posts: Doghead's Cosmic Bar