Bot Contest

ohhh ok

Hey rex.. I'll buy ya a drink of your choice over at dogh'd's and be your handservant for a (N number of days) if you'll make one of steve
Pwwease? Pwetty Pwease?


Only if you have time and want to, of course
--Shady

My attempt to personalize my bot page has been refused, but they're still going up on the bot photos page. If you still want one, I'll see if I have time in the next couple of days. That's it though. No more requests unless you can get me a large number of votes.

I'm even more confused about the scores.
I read the transcripts and 10 questions for a bunch of the higher ranking bots and the only reason I can imagine that oraknabo's jury score is so low is that he insulted a judge. I realize I probably don't rank with any of them over 6, but my answers were way more coherent than any of the ones iin the 3.5-4 range.

I'm not trying to cry scandal or complain too much, but this highlights a basic flaw I see in turing tests. My only interest in creating a bot is character and personality, not accuracy. Not everybody has a favorite of everything. Many people when asked about a favorite book will say, "I don't really read much." But "A la Recherche de Temps Perdu" is by default going to get a better score just because there's an "OOH AHH" factor in the specificity. Again, math is the same. How many people do you know that can answer anything beyond operations involving addition and subtraction of single digits with any accuracy without a calculator or pen and paper? I WANT my bot to get it wrong sometimes.
Basically, if oraknabo pisses you off, I've succeeded, and that's all that really matters to me in the end.

I finally found all the details of the contest at http://www.frontiernet.net/~wcowart/contest_rules.html

I didn't realise it was going to go on till the month of June! Looks like the PF bots are doing pretty good, but there's still a long way to go before the fun's over...

That's such a long time to wait. 1-2 weeks would have been perfect.

Dogh'd did the best of my Bots on the 10 questions, and the score reflected that. The rating system seemed pretty cut and dry, so it's strange that your score should be so different. I agree that a realistic response like "I dont have a favorite" should score as highly as a specific answer. If someone asked me a math question in a chat room, I'd probably say "Huh?" or not even respond.

RE: the contest details

This is exactly what I'm talking about- My bot is intentionally evasive and reluctant to cooperate with the chat partner. I this case I think there's a really fine line between a "vague or non-committal response" (only 1 point) and an "appropriate response to the question" (4 points) which I don't necessarily think should rank less than "answering the question correctly" (5 points) in all cases.
I know they have to have some standards for scoring, it's just something I think is worth debating.

Yeah, but we have to remember that the bots are in a contest and should try and do they're best to impress the judges... Like in Miss America contests - as if those girls really go to bed praying for world peace! Anyways, I agree that Oraknabo is pretty realistic when he's chatting, but his lack of cooperation cost him some points with the judges I guess...

I can't find my botmaster profile anywhere... Although I saw some descriptions for other bots on their voting pages.

I just looked at Albert2, who got the Star Wars and Moon landing right, but most of the rest of his answers were totally off and he has a 4.28 avg.
And Robitron's scored highter than both of mine and his answers were pretty mediocre.

I do feel like one of those pagent moms, totally overjudging the other contestants and getting a little pissy about it all.

Maybe your little darling will do better next year, Rex. *pats head*

Kidding! I wonder if you could go for 100% realism with "Sleeping Bot" who just says "Zzzzzz" all the time.

One really interesting thing to do is to read Loebner transcripts of judges who thought other real people were Bots. That's a new kind of insult, no?

There aren't any "secret humans" in this contest like some of the others are there?

Oraknabo is in first place though, which is weird with such a low 10-question score. Perhaps we've yet to comprehend the judging/ranking system.

Let's not forget that these bot competitions are still a pretty new concept, so I guess they're all doing their best over there.

Here's a thought... in a couple years, when the contest rules are all figured out, the bot contestant databases are automated and the bumps and folds are evened out in the whole process - our bots are going to be too evolved for the competition! How will we judge them if we can't even tell the difference between bots and humans anymore? Best bot personality? That would be tough for Oraknabo!

I can just imagine the game show type contests we'll have to get them participating in...