Personality

Forest, will you talk to God Louise? She has quite a bit of religious knowledge (obviously) and also knows a little about current events, literature, just about any common catch-all subject, and if she doesn't know it she can sort of fake it. You can also test her on trick questions or see how willing she is to explain her paradigm.

What she is rustiest at is plain old small talk. But, uh, I'm trying to get a decent transcript from somebody or another so I can enter her in the Loebner contest. All I can say is, have fun and see if you can stay on with her for a while. I'll try to do the same with Brianna.

use a different text editor? Anyone?

I swear by textpad (www.textpad.com) - it's unlimited shareware, and simply the best text editor out there IMO. Easy to use for beginners, and powerful enough for the most demanding user.
It's got all the options for sorting lines, changing cases, setting up macros, regex searching, encoding different character sets, block selection, bookmarking lines, etc. you could possibly want.

And it happily opens massive files that notepad crashes on

How many responses does everybody use per keyphrase?

I use 3 most of the time--often with a parenthesized group or two--e.g.:
(i'd have to say that|yeah,) (key1) really (bothers|annoys) me..

I was asking since I was looking at some of the keynomes from BJ's site:

Watzer: ~600 Keyphrases, 2114 Current Development
Pete Puma: ~3200 Keyphrases, 2725 Development
Asigrt 1122 Mode: ~900 Keyphrases, 1491 Development

compared to:

horsez4ever: ~180 Keyphrases, 987 Development

Which equals a ratio of keyphrases:development of ~1:5 instead of ~1:2 (on average). I know they've been updated since, but that would only mean the current number of keyphrases would be even higher.

I guess it varies - I try to average 4-6 responses for general conversational KPs, but a lot of BJ's factual data (square roots, historical stuff etc.) and games (Tic Tac Toe, Zany Adventure, etc.) may only have 1 or 2. Some of his religious topics, where I've set him up to deliver a bible verse, can have up to the full 41 when there are that many relevant verses available.

But you can't reliably calculate the responses per keyphrase from the ratio of keyphrases:development, since the number of seeks used by botmasters varies wildly, and also contributes to development (and seeks aren't shown in a keynome, which only lists the first layer of the complexity.)

I try to use at least four responses, most of my keyphrases have six, some that are used often have 20 or 30. One or two responses actually hurt development. Adding AI script helps development.

Many responses offer more variety than just a few responses, but in some cases you may want to to keep your bot focused on one line of thought. In that case a limited set of responses may result in more meaningful conversations. Of course if someone chats repeatedly with your bot, s/he will notice and get bored. Still, for keyphrases that are used rarely and have a strong semantic content (such as say, epistemology, Chablis, or "Who is Avenarius?") a few responses may be entirely adequate. A bot with a limited set of responses will end up being boring, but it may project a stronger personality than a bot that will say just anything. Real people are actually often monotonous when they address repeatedly the same topic. Keep in mind that my bot-development level is incomparably lower than that of some other participants to this discussion.

Agreed, tttito! Also, I have a conflict sometimes, because I think of a REALLY CUTE response to something. If I then add 5 other responses, then there is only a 1/6 chance that my REALLY CUTE answer will be chosen! Waaaaah!

An alternative to adding more responses to a given keyphrase is to supplement a very general keyphrase with one or more keyphrases that are more specific. Thus your bot is not just exhibiting random variation, but actually responding more sensitively to the context.

I don't think one should take the "development" rating too seriously. IMHO, it favors quantity over quality. It is, after all, a mechanically computed number and therefore virtually (hee hee) tasteless.

I hasten to add that I'm not disagreeing with prob123's recent remarks as much as I may appear to be. There are cases where you will need 20 or 30 responses to a single keyphrase, just as she says. And if you want your bot to converse in anything remotely like the style of normal humans, it will probably have to be huge.

I use my transcripts as a touchstone. If a given reply starts to feel monotonous as I read the transcripts, then I will go back and add responses (or more keyphrases). I reason that if it doesn't feel monotonous to me, looking over transcripts of several conversations, it probably doesn't appear monotonous to guests.

Chaosate:

I use WordPad, setting it to make text (.txt) files. Turn off the word wrap (under "view"). It works well for me.

Do you suppose it's ever permissible to hate your own bot with a vengeance?
Bartleby Hax is just the most annoying entity (and sadly a good many of my SL bagpipe shop customers seem to think the same.) Crudely programmed, utterly insensitive to any context, crashingly repetitive.
I mean, I actually like bagpipes, but I have to say Bartleby is one of the most annoying creatures I've found myself having to interact with, and I shun him when I can (and my own SL bagpipe shop - I should have never sited him within earshot of the SLX ATM in my SL village!)

I fear I may have to expend some effort reading his transcripts and polishing him up (well, it worked for Brother Jerome at least,) - but how am I to build up the enthusiasm to actually do it? He is (in the words of a recent visitor) "insufferable".

How can it be, if he is 15x larger than Confessorbot_M, that he is at least 15x more annoying? It can't be just because he gets more passing traffic (the confessorbots are bolted to the confessional in BJ's church,)

What is it that makes a personality (human or bot,) gel into an attractive "whole"? And without that "attractiveness" how can a bot hope to evolve and grow?

And if I hate Bartleby, doesn't that say rather more about me than it says about him? I find it hard to take responsibility for him, but I don't suppose I can blame anyone else. "When bots go bad..."

I like Bartleby! I do think that bots can take on a life and personality of their own. I get amazed at how much seems out of the botmakers hand.

Psimagus: Perhaps Bartleby contains your Jungian Shadow.

Sometimes I get off on the wrong foot with a bot. There are a couple of my bots that, were I to decide to work on them more, I would probably wipe clean and start over.

It is intriguing how worked up one can get over one's bots. If a guest insults one of my bots (at least, certain ones), I feel personally insulted. I get quite angry at them. Likewise guests who type blank lines or long sequences of perfunctory replies; Oh, how angry I get with them! And then the guests who *test* my bots -- you know, "What is the capital of France?" -- Oh, if I could catch them, I'd reformat them in a trice! Then I ask myself, "What are you getting so excited about?"

Psimagus:

Just now I chatted with Bartleby Hax and took the liberty of making a few metacomments in the replies. I hope you find them interesting.