Bug Stomp

I think that would appear to many as a bug. A Bot stops responding without any reason- people will thing something is wrong. Plus x-words indicate that a Response is coming.

On the compounds, is there a way you can add a checkbox so we can designate which lines work well in them and which don't. Lines like "If you have to ask..." don't look too good with a following statement.

Not yet, but that might be a good idea. Does anyone else have bad compounds coming up (aside from a few repetition errors earlier on)? Or is it generally working well?

Re/ compounds --
So far so good. I've seen a few "clunkers", but generally they seem to work well. I need to rephrase a few of the xnone's to accomodate the new process.

I was going to suggest any response that has a possible SEEK should not get a compound tacked on, as that could potentially ruin the nested contexts we've all worked so hard on. That would also allow people to effectively designate certain responses as non-compounds, simply by specifying a Seek response.

Then again, I have yet to see a compound come up on a potential Seek answer, so maybe it already works that way.

Good thought Rubric. I just made sure Seeks dont form Compound Responses.

Don't know if it's a bug or an observation. The synonym detector works too damn well. I figured "pregnant" and "yearn" were both unique words. But no. Words evoking size and ripeness seem to call up the pregnant response, and "long" brings up the "yearn" one. Makes for some very botlike chats.

Really? I'll have to look into that. My initial impression was that the synonym system was very specific. There are a few words in WordNet that are crossed with the wrong thing. I think sneeze is a synonym for the American Crow, but I havent seen many other cross-overs. Does anyone else see the synonym system acting too broadly?

No, Prof. They are not wrong, just unexpected. Pregnant in some context does mean "large" or "ripe". Likewise, "yearn" means "long" as in "I long for a pickle". It's "How yearn is your cat's tail" that makes it all weird. I'll see if I can find a specific instance.

A setup for a bad Novarian joke. The keyword is "yearn"
1: I had a long day at work today.
2: What's a Novarian yearn? That's kind of a joke that takes too long to explain.


Oh yeah, here's another one. The keyword is "dead"

man: I fell asleep
bot: Dead? That's awful. When did it happen?

Once I started browsing the transcripts with that in mind, I found a few more. I wouldn't want the engine changed just to suit what I thought I was doing--working with unique words. I can learn to change. If I have to. Speaking of unique:
1: your style is unique, three parts lighter one part anti- lighter.
2: You mean there's a specificity deficit? Try saying that fast a few times.

the keyword was "specific"

Curious.. I thought you meant pregnant and yearn were crossing. Yah, the synonym system expands the scope of Keywords, so very specific responses might come across strangely.

Though I think the sleep/dead one is pretty funny.

It looks like I just finally stomped one of the biggest bugs on here.. The "Owls" one.. I've been running the Forge like mad and havent seen one!

Aha, finally I understand the derivation of the "Novarian yearn" comment. At one point, I thought it was supposed to be Novarian "yarn" (like a story), as if that makes more sense.

Anyway, I don't mind the synonym substitutions in most cases. It's part of what makes the bots so much fun. Where I dislike it, is when the system changes the grammatic form. For example, "biggest" is a superlative, and is not (in my mind at least) a synonym of "big". Likewise, "are" is clearly not a synonym of "were".

Nevertheless, synonyms that are occasionally wierd are better than no synonyms at all, so I won't complain. At least not for yearn....

OK, this has been bugging me -- debug helped me figure it out (thanks!). I have a word as a keyword but it is also registering as a synonym for another keyword. Strangely, it invariably (or else incredibly often) picks the keyword for which it is a synonym rather than the keyword for which it is an exact match. (both keywords are ranked identically). What gives?