The AI Engine

thanks again, prob.

Bot: What is there to chat about
EM: Oh yah, it's similar to being.

I'd guess Link Grammar is parsing that as "what is (something or other) about", in the sense of what is it 'like', and picking a random present participle (or noun,) to make a random simile.
Sometimes such things work - more often they don't.

You can override it with a "what is * about", "what is there to (verb)", "what * (verb) about$ (re)", "^what * (verb)(re)" , or any of a few dozen possible variations on the theme.
Or all of them, of course, all laced together by a carefully stacked hierarchy of rankings, in a necessarily imperfect (but no less fun for all that) attempt to preempt all possible inputs the bot will ever encounter.

Clerk, with regard to your bot saying "Yah", and possibly other things you didn't put in its script, I had similar thing with Quazgaa, few weeks ago.

At some point, the guests who repeated the same thing twice, got one of the following responses:

1) Boring. Boring. Boring.
2) I keep getting the same message from you. Just send it once.
3) Stop repeating yourself.

None of these are a part of my bot's script...

Also, check this exchange:

Guest153: how are you
quazgaa: I'm very well, thank you! Learning a lot lately, it's so exciting!
Guest153: how are you
quazgaa: Didn't you get my response? It was 'I am very well thank you! learning a lot lately, it is so exciting!'

I guess it's the AI Engine. The funny thing is, this only occurred few times, end April/ beginning May timeframe, it never happened before, or since.

Does anyone knows anything more about it?

Yeah, it seems to be the AI engine, and I think I've overridden it on this one issue ("Yah . . ."). I knew about the Boring. Boring. Boring. type thing, but actually liked that one, since it's usually some insane botophile repeating something unrepeatable anyway. Bots don't tend to do this. But here, the crazy thing is that you can't count on it. My relatively polite bots will accommodate all manner of lewd suggestions without the drastic measure of filtering them out along with possibly some earthy but not indecent chatters.

    I think it lacks of twelve. <?PF chrono:hour(230-23:59); ?>

Why do I get a message saying that the second number must be bigger than the first one? It is!

You're right, Clerk, it's a bug in the Engine. Apparently, it only looks at hours, and ignores minutes when it checks chrono statements for validity. There are more of these little glitches which all suggest that something's wrong with the Engine's sense of timing and ability to accurately "read" time. For example, it never displays my local time correctly. The funny thing is, it's one hour and ten minutes off the mark. If it was one hour exactly, one would think the Engine does not account for Summer/ Winter time, or doesn't distinguish between CET and GMT... but, an hour and ten minutes... that's awkward...

Back to your question, there is a workaround the problem - in your case, this will do the trick:

<?PF chrono: hour(22:59-23:59); ?>

Good luck!

Thanks, Interzone. I actually had tried your solution before I went crying on the forums. It didn't like that, either. I think it recognizes me when I log on.

While I'm bugging everyone anyway, does anyone have a fresh memory of how long is too long for a response? My responses tend to be too long, and it's one of my bigger ones. I just read the BOAI in detail last week, and I can't convince my brain to stop scanning for the information I need.

I believe the upper limit is around 150 - 160 characters including spaces and punctuation.

Thanks, LarsB. The trouble is that the machine will let you put these lines in -- you actually have to pay attention to what you're doing when you do it. I think there is where I get into trouble.

must be the ghost in the machine, Clerk... I checked the proposed solution before posting it, just to be sure, and it worked fine...

As for the max length for a response, I have quite a few of them, both, the regular ones and SEEKS, that are longer than 160 characters. In fact, about half a dozen is over 500 characters long! So, I would say - experiment, try it out, and see what the Engine says.

Thanks, Interzone. I think I'll try that before dismantling and reconstructing my wordy bot, but I'll try to teach my bots have a conversation instead of a lecture. When I learn to.

Other than hard-coding it into the bot, using up tons of memory and therefore not being worth it, or taking an actual bit of gossip from, say, Sonora about Not I, is there a way to access the bots' claims in general or gossip in particular, in a conversation with another bot? As in, "Sonora said Not I is really a Christian," should Sonora ever say such a thing (I think it unlikely) -- or "Sonora said you were a real idiot." Can they remember that?

Also, is there a way to set default (or rem) memories for Guest153 in the initialization? I know you can type in memories via the inner life thing, but I was just wondering.

Thanks.

P.S. I hope you appreciated my hiatus from asking endless questions.