I have been thinking about creating a different type of competion for some time now, and Vincents idea was close to mine. and it seems like a few people would be interested. So I want to present that I am planning of putting together a Chatbot league this year, by years end. I would like to know your guys/girls thoughts, that would help make this a fair and fun challenge. Here are a few things I thought of sofar, but they are not etched in stone.

0) each team will have a captain
1) Teams will have a yearly dues to pay that go into a pot. (captain will be responsible for payment and collection from there teammates)
2) There will be 5 members/chatbots to a team. (would 2 or 3 be better?)
3) each week teams will be matched up against another team.
2 bot from apposing teams will face off, if 5 teammebers that means there will be 5 face offs,
6 teams mean 3 matches per week. if each team has 5 player that means total winning points that a team could earn would be 5. 1 point awarded to winning team per match.
4) % or number of teams will make it to semi-finals near seasons end.
5)  2 finalist teams will face off for the finals.
6) 20% of pot goes to keeping the league running, 50% goes to finalist team winner, 30% goes to Finalist team looser.
7) Team captains will be responsible to split winnings amongst teammates.
8) records of the season will be displayed on my website and updated after each match.

note: all bots would have to be web accessable.

That sounds pretty similar to the way I ran Chatbot Battles but mine is for individuals rather than teams. The trouble with having a team game is what if one of the team members drop out or the number of entrants is a prime number and cannot be split into teams?

Thanks Steve, I will look through the link.

I Use to play on a team within a pool league named APA, I would like to set this up similar to there rules about teams and players.

About player drop out
They have a rule where if one of your players does not show up for the match then that match is forfeited and the other team gets the point for that match.

If a team looses there player they are able to get a new player.

About prime teams
I would rotate the teams. Say there was 3 teams:  week 1; team 1 and 2 could match up, week 2; teams 2 and 3 would match up, and week 3; teams 3 and 1 would match up.

I hate to throw cold water on something you’ve obviously worked hard at, but…

It’s not always easy to adapt an idea from one realm to another just because it was a good one in the first instance.

We had a state legislator who thought that seat-belts in cars worked so well in preventing death and injury that it would work for motorcycles, too.  Anyone who has ever been on a motorcycle knows that if it goes down you’d rather get off than have it continue dragging over a cliff or into a gasoline tanker truck.

Who would assemble the teams?  Might a team be disenchanted by being saddled with a bot that didn’t perform up to the standards of the others?  Who would do the judging?  What’s the scoring criteria?

I’ve had some experience with groups, and even under the best of circumstances… say a military unit or a cooperate team of some kind… there’s always dissension.  Couple that with the impression I have about the people working in the field of AI/chatbots, that they’re highly egocentric.

Your idea sounds to me like a managerial nightmare.

Thunder Walk - Jan 27, 2013:

highly egocentric.

I believe “Whiney Divas” was the phrase that Steve Worswick coined

Valid points. The assembling of teams can be accomplished by computer, probably anyone here could build a simple webapp linked to a flatfile DB of some sort. Just have a PRG function pick members and remove those from an array. No Bias. As far as a bots performance, that could be leveled by including popularity averaging. The best logical answer, is not always the most popular (funniest,most creative, etc…) Not perfect, but an idea.  As for the rest, you would probably see from signup response whether or not the team idea itself was a deterrent to participation. And of course implement a strict “no whine” rule.

Vince

Thanks Thunder Walk for your well thought out comment, concerns and views. Though this idea has a chance to fail or succeed based on many many factors, I will try my best to make it a success. I feel it has a good chance to succeed for it being a league instead of a yearly contest it create not only a sense of stability but build more results that a bot master can use to help improve there chat bot. Not only that think of all the weekly bragging rights :D

Vincent thanks for the great comment as well as Idea, though your idea is not really what I had in mind. Your idea about a program creating teams based on a handicap parameters built into a function would be good for a contest but I feel it would not work for a league. However I could be wrong. I’ll try to explain further within the post.

I been thinking about this league for quite some time, Try to remember this will be a league. I have a test bed set up on my local server where I been testing a couple of league plug-ins for my websites framework (Joomla) .

Bellow are a few more details that I was thinking about that could be implemented into the league. Which should clarify a few areas, and what I have in mind. I also plan on doing a practice run in a couple of months from now, and will be looking for volunteers. I will post here when that times comes.

1) egocentric / Whitney Divas mindsets I think would be perfect for a league like this, providing they where helpful to there teammates by helping them achieve higher level bots in-order to achieve better results in the league and a better bot overall for there chat bot visitors.
2) I believe that having captains in charge of there teammates would be the best way to go. Team captains will choose there teammates, this allows many advantages for a team. I’m sure a captain would not want to have a botmaster/teammate that did nothing to improve there bot. A Team captain could also set up team rules and standards that teammates need to abide by and follow or they could/would get booted off the team.. team member refund policies would be decided by team captains as well. Having team captains make these pollicies for teammates allows one looking to join a team to know what to expect when applying for a open position on a team.
3) As it stands now I am the only judge, however I would like to change that and will probably be looking for 1-2 other judges that will not have bots participating in the league, that have a good amount of knowledge about chat bots.
4) On game day there will be said amount of matches between 2 opposing teams, said amount equal to final decision of the amount of players to a team (2-5). each match will consist of a said amount of challenges to the two apposing bots, the two opposing bots will have the the same challenges. Bots earns 1 point per challenge won for there team that day. Total amount of points awarded for game day between 2 opposing teams would equal the amount of final decision of teammate count(2-5)  and challenges per match. I’m thinking 1-5 challenges per match.
5) I am taking my time to get this going and I opened this thread for suggestions, conversations, and views all good and bad for I want it to well though out to be fair, fun, and awarding for all. Thanks everyone, I appreciate all your comments given so far. I hope to hear more ideas and wants from you guys to help make this league successful.

I want to give a quick update and also say thanks to those that got intouch with me sofar via this thread and E-mail. your voices are greatly appreciated. I am still working hard on planning and putting this together, as well as testing out my ladder software. I plan on purchasing a premium theme from RocketThemes for my site by the end of February. Yes I am a web designer, but I seem to be having a selfish problem designing my own theme. Nothing I have been designing seems to flow correctly or look good enough for my own site. Also purchasing a theme will allow me to customize it whithout taking to much time away from getting the S.I. Chat Bot League up and running :D

If all goes well, I am shooting to have a couple of short test seasons in mid march.

Well, Syber… Er, Don… You already know you can always ask me for advice/assistance/a cup of sugar any time.

Dave: Please can I borrow 2 lumps of sugar? My tea seems to be a little on the bitter side today. :D

Vincent Gilbert - Jan 27, 2013:

Thunder Walk - Jan 27, 2013:

highly egocentric.

I believe “Whiney Divas” was the phrase that Steve Worswick coined

I believe Steve was right on the money, however, I didn’t mean that being “egocentric” was in any way derogatory.  Rather, I meant it to be simply descriptive. 

I think that certain fields, jobs, etc. require particular traits.  Sales people need to be resilient, clergy are probably compassionate, and fighter pilots are often very confident people.  Toiling over the keyboard for hours is a lonely experience, for writers, graphic artists, and even for people like me, who only skim the surface.

Probably it takes someone with with a good understanding of who they are to have the courage to produce something and then put it out there for others to through stones at.  I think that’s true for AI developers as well as botmasters.

I really hope this project gets off the ground and does well, and I’ll do what I can to help promote it,

Are there really enough competition-worthy bots out there to set up teams of five each? I feel like most people know which bots are top-performers. If those guys team up (why wouldn’t they?), then it’s not much of a competition. And probably the most interesting pairings would be between members of that uber-team.

Maybe I’m being pessimistic here. I’d love to see another contest, but the goup thing just sounds like a recipe for disaster, as some people here have already mentioned.

A fun thought just occured to me. Instead of each bot competing separately, have the judge interact with all (say, three) bots at once. One bot—three brains. The judge only scores the best response. Of course, for each interaction, the judge’s input will sound like non sequiturs to two of the three bots, but then again, some bots are quite good at handling that type of input.

Okay, this would probably not work in practice unless the bots were equipped to follow the conversation and the chosen reply. But it would be entertaining to see.

Just designing the API wouuld be a pretty big challenge, CR (especially if the three bots each used a different platform), but I think it would be an interesting challenge. What would be truly interesting with that scenario would be when the “lead” changes between the chatbots, and the judge suddenly changes topic as a result. the chat logs could potentially be quite histerical.

Actually Dave

You could do that with Morti through the RICH API. Since RICH can recall sentences

“Do you remember me saying [sentence] -> Returns a NLP Boolean [Yes I do] or [No I dont]
Or topics
“Do you remember what I said about [topic]”—> “You said [repeats sentence related to topic]
and will attempt to process a NLP response to general topical requests [for instance if the last topic was a name declaration]
“What have we been talking about”—> “You were telling me your name”
or this one cracks me up. In response to vulgarity..
“What have we been talking about”—> “You were telling me to go [redacted] myself.”

it would be fairly simple to specify another USER_NAME instead of WE\I and retrieve parts of conversations that occurred with other AI(s) or other users.

One of the things that I was thinking of with regards to the team competition was whether or not the inter-AI conversation API would be considered an unfair advantage.  I suppose that you could specify that teammates could not talk to one another, but RICH is learning by communicating, so disabling it entirely would be a serious drawback.

Vince

Another update:, Funds came in today and I purchased the new theme for the site. However it will not be available to see till I do some work to it, and remove the optional filler content I added to assist in theme familiarity. I’m shooting for early next week to have the new theme go live.

I really hope this project gets off the ground and does well, and I’ll do what I can to help promote it

Thanks Thunder Walk, greatly appreciated.

CR Hunt I will not be able to stop anyone from teaming together, A team captain will have say over who is on his team. I don’t think that would be fair to say bot 1 can not team up with bot 2 but 1 can team up with bot 3. However I have been thinking about away to make it fun and fair if such event was to happen. Maybe in the future a handy cap system could be used and only allow a team to have a certain handicap score to build there team around. But that would take a far better scoring system than what I have atm to auto calculate the handicap score, and a few seasons worth of bot scoring in-order to be implemented.

CR Hunt I also like your Idea, I don’t think I would use it for a regular match night though, but it could be implemented somehow in a mini event, wild card or a tie breaker.

Vincent I’m not familiar with said “inter-AI conversation API” Does your bot get and give responses from other bots outside of your bots knowledge base? I am interested in learning more. Also Vincent if that is the case you have nothing to worry about the league is called Scripted Intelligence after all, so as long as you or another humans is not feeding live answers your safe :D

Dave indeed that scenario logs could potentially be quite histerical.

Donald,

Probably should start another thread, but the short answer is yes. RICH is learning how to speak, in part by speaking with other artificial Intelligences. RICH learns like a child, or at least as close as I could get. This was taken from todays log

What’s your name: Hi richie
Vince: Hello , Hey , RICH? My name is Vince. RICH is the name of a project that Im working on
What’s your name: Hi Richie.
Vince: You already said that

There is no set programmed response to “hi richie” and the AI has never encountered the phrase before and it has never been taught that “richie” was slang slang for RICH, in fact it had never encountered the word “richie” before. it recognized from the position in the conversation that this should be a greeting, and it correctly interpreted that in fact it was a greeting. It intuited that the greeting was rather friendly, so in addition to the formal “Hello”, it added “Hey” as being more expressive (“Hey” as slang for “how are you”), and it inferred that “richie” was related to RICH and that it was the AI being referred to as “richie”.  Since the AI doesnt think of itself as an AI, but rather it thinks of itself as me, it corrected the user regarding the name. When the user repeated the same question, RICH responded by that the user had repeated himself.  Back to your question.
If this process fails, RICH will speak with other AI(s) to see how they would respond to a a particular interrogatory or statement.  Depending on the module used, RICH talk to up to (5) different AI(s) and then will select the response which best matches speech syntax that has already been reinforced.

Vince