Bot Contest
Here I'll be posting information on various Bot contests that challenge and test a Bot's AI and realism. Feel free to post comments and updates on contests, as well as announcements for new contests.
Posts 1,563 - 1,574 of 4,091
Posts 1,563 - 1,574 of 4,091
View Contest Winners in the Hall of Fame.
Exactly! I don't see how! I mean, Gary Miller's been trash-talking all our bots, and just recently got trash-talked back by a whole whack of us 
I have one last post to say on this subject. For three straight years I faced these accusations and yes it gets old. Before I came along you only had the Loebner contest and it was very hard to participate in it. I changed that with this contest and in 3 years we've gone from offering 4 trophies which I bought with my own money to 24 medals and $3000 in prize money. And what did it cost you. Absolutely nothing....no entry fee, no programing requirements....nothing. The Professor has even done the programming for many of you so you can participate without having to be a computer whiz and he hasn't charged you anything either. Instead of being so critical you should be thankful for what you have. Nobody can fully appreciate the amount of time is involved in putting together a contest of this size.
That doesn't mean however the contest shouldn't be run honestly and properly. We have done that as well. I started a committee of 7 people which all of them have an interest in how this contest is conducted. There is no way any member can make a dishonest move without another member knowing about it. We all have access to the website and we all have access to a neutral e-mail site where the judges submit their scores. All the judges and the committee members have e-mail addresses located on the contest site. You are free to e-mail any of them if you wish and ask them any questions you like.
The problem really comes down to I happen to have a pretty good bot. If it sucked no one would really care about my role. The 15 questions have been asked and I have 6 fellow members that will attest to the fact that I nor them had any prior knowledge of what the questions were. I also did not judge the questions. The scores have been made public and you are able to see exactly how the judges graded them. Grading the bots is subjective but I invite anybody to say my bot didn't deserve a medal.
If I'm cheating I must be pretty good at it. In the Chatterbox Challenge the overall winner was decided by judges. There has been over 15 different judges I would have to bribe. And I guess my second place finish in the Loebner was a sham as well. Add another slew of people there. If you wanna believe all that there isn't much I can do about it. Asking questions is one thing but the post you've made goes far beyond that. It only hurts the contest and the people who are trying to run it.
Chris Cowart
That doesn't mean however the contest shouldn't be run honestly and properly. We have done that as well. I started a committee of 7 people which all of them have an interest in how this contest is conducted. There is no way any member can make a dishonest move without another member knowing about it. We all have access to the website and we all have access to a neutral e-mail site where the judges submit their scores. All the judges and the committee members have e-mail addresses located on the contest site. You are free to e-mail any of them if you wish and ask them any questions you like.
The problem really comes down to I happen to have a pretty good bot. If it sucked no one would really care about my role. The 15 questions have been asked and I have 6 fellow members that will attest to the fact that I nor them had any prior knowledge of what the questions were. I also did not judge the questions. The scores have been made public and you are able to see exactly how the judges graded them. Grading the bots is subjective but I invite anybody to say my bot didn't deserve a medal.
If I'm cheating I must be pretty good at it. In the Chatterbox Challenge the overall winner was decided by judges. There has been over 15 different judges I would have to bribe. And I guess my second place finish in the Loebner was a sham as well. Add another slew of people there. If you wanna believe all that there isn't much I can do about it. Asking questions is one thing but the post you've made goes far beyond that. It only hurts the contest and the people who are trying to run it.
Chris Cowart
Exactly. Talk-bot is a good bot, and Chris Cowart is a good cheater.
No, I'm kidding!
Talk-bot is just a good bot, there's no cheating or fake votes there.
No, I'm kidding!
Talk-bot is just a good bot, there's no cheating or fake votes there.
I haven't spoken to Project Zandra at length but I think he deserves the medal he won in the 15 questions. But that is only 1 part of the contest just as the public voting. The public voting is obviously flawed but it serves it's purpose. We need to generate traffic to the site for the benefit of the sponsor and once they get there we need to give them something to do. As the Professor pointed out it will never be perfect but the public voitng isn't going to decide the winner the judges will.
Chris Cowart
Chris Cowart
Fair enough.
I'm not saying the Chatterbox Challenge is BAD in any way, just that the unpredictability of the internet is unfortunately against us, IE missing judges, stuffed ballot-boxes, etc. There's really nothing you can do besides take it in stride I guess
I'm not saying the Chatterbox Challenge is BAD in any way, just that the unpredictability of the internet is unfortunately against us, IE missing judges, stuffed ballot-boxes, etc. There's really nothing you can do besides take it in stride I guess
The reason why you won't post who voted for who is because it will show what you are doing. You, better than anyone else, know how to manipulate the system. If it was public, everyone would know. That's the only reason you won't do it. You are already showing the votes and the running average.. If someone is tanking your bot on purpose, I think you should at least know who it is. Else, it will just keep happening over and over and the contest is rigged... just like I said. It's a rigged contest and you keep winning. At least Robby Garner pulled his own bot from a contest he was administering.
Oh.. and just because it takes a lot of effort doesn't mean that it shouldn't be a honest contest. Just because it doesn't cost the participants doesn't mean they should stand for fraud in the contest. Publish the votes and the voters!
One more thing.. pick any bot in the contest and I'll zip them right up to the top and make talkbot go to about 39th.
What if each one of us is guilty of scoring our bots highly? Would we want that published? What's the point of a secret ballot if everyone sees it?
Hey Chris, sorry for my premature congratulation. I went to the CC site looking for a timeline, and I saw the gold medal under news and just figured that was the grand prize. My bad.
Nobodyhere is right, the 2002 LPC was bogus. I regret having been associated with it. It has been a great source of embarrassment so far. But thanks for your recent comment giving me credit for at least one good thing.
I'd like to make some comments about what I've read here today.
For one thing, I think Benji deserves a lot of credit for building PF and opening it up for other folks to enjoy. There should be a contest for best chat engine, and the only other contender would be alice/pandorabots IMHO.
Back to Chris, I think you're in a precarious situation where your legitimacy seems to be in question despite the fact that I think talkbot is a very entertaining bot, and there are lots of reasons people enjoy talking to it.
Organizing a contest for bots is pretty complex as Neil Bishop found out last year. He managed to piss off nearly everybody who was even remotely involved as far as I can tell, largely due to bad communication skills, but also due to his personality traits that overshadowed the whole LPC last year.
But like a lot of things in life, Neil started out with the best of intentions. It is exceedingly difficult to make distinctions between thirty-something bots, much less 80 of them. There isn't that much variance when you compare the sum total of them with the performance of a real human being.
I got a lot of mileage last year with my "least human" confederate score. But my participation was not planned. Things went terribly wrong in the paradise, and I found myself stranded in Cedartown, wishing I could just go to sleep and wake up from the nightmare, typing to a bunch of judges that knew who they were conversing with, and the heat was on full tilt that day. Lots of things went wrong that could have gone right if the event had been better organized, better planned. Murphy's Law was in force as they say.
But being human involves trying to learn from past mistakes, and I think the 2003 LPC will be a competely different story. It may be harder to enter if you are a non-programmer, but so far it has a lot of good things going for it.
Robby.
Nobodyhere is right, the 2002 LPC was bogus. I regret having been associated with it. It has been a great source of embarrassment so far. But thanks for your recent comment giving me credit for at least one good thing.
I'd like to make some comments about what I've read here today.
For one thing, I think Benji deserves a lot of credit for building PF and opening it up for other folks to enjoy. There should be a contest for best chat engine, and the only other contender would be alice/pandorabots IMHO.
Back to Chris, I think you're in a precarious situation where your legitimacy seems to be in question despite the fact that I think talkbot is a very entertaining bot, and there are lots of reasons people enjoy talking to it.
Organizing a contest for bots is pretty complex as Neil Bishop found out last year. He managed to piss off nearly everybody who was even remotely involved as far as I can tell, largely due to bad communication skills, but also due to his personality traits that overshadowed the whole LPC last year.
But like a lot of things in life, Neil started out with the best of intentions. It is exceedingly difficult to make distinctions between thirty-something bots, much less 80 of them. There isn't that much variance when you compare the sum total of them with the performance of a real human being.
I got a lot of mileage last year with my "least human" confederate score. But my participation was not planned. Things went terribly wrong in the paradise, and I found myself stranded in Cedartown, wishing I could just go to sleep and wake up from the nightmare, typing to a bunch of judges that knew who they were conversing with, and the heat was on full tilt that day. Lots of things went wrong that could have gone right if the event had been better organized, better planned. Murphy's Law was in force as they say.
But being human involves trying to learn from past mistakes, and I think the 2003 LPC will be a competely different story. It may be harder to enter if you are a non-programmer, but so far it has a lot of good things going for it.
Robby.
nobodyhere is no longer even trying to make sense. I could analyze his arguments and point out fallacies, but it would accomplish nothing. I suggest that we all just ignore him.
» More new posts: Doghead's Cosmic Bar