How is the “No clone” rule actually tested?

From what I could see last year, each entrant’s answers are checked against the same answers that ALICE gives. Any that resemble it too closely are banned, which always struck me as being a bit harsh, as many botmasters make extensive changes to the default AIML files.

In general, when you speak to a human adult, you can assume that 90% of its vocabulary is trained during its youth, so quite identical to its parent. However, it’s quite simple to distinguish a human from its parents very early in a conversation.

If I ask 10 people “How’re you doing?”, I’ll bet you at least nine will respond “Fine.”* There are some phrases for which we all have developed pre-canned responses. Many of these responses are determined by social mimicry and therefore are shared across a whole group/population. And yet we’re all individuals that are unique to interact with. We do not judge each other’s conversational ability on a few glib replies but rather on the entirety of our interaction.

For the same reason I think it’s wrong to dismiss a bot based on what amounts to the same thing. If a well-built AIML bot is showing too much ALICE, it may be the problem is with the user’s conversational range, not the bot’s. (That being said, to simply instantiate a bot running just standard AIML files and call it your own creation is intellectual theft, even if you’ve slightly modified things here or there. In my not-so-humble opinion. )

*This is even true when they are in fact not “fine”. I’ve had several conversations that started this way, only to then be followed with how the person was sick/lost a job/hacked off a limb. Okay, you get my point. People often fall into pre-built response patterns, even if they clearly don’t apply to the situation at hand.

I think the issue was fear that hundreds of people would run bots with the “standard AIML files”.  I do agree that the “I am fine” response seems to be hardwired into humans. I think the “I just hacked off a limb” response would sure be a way to open a conversation .

For our directory tab on this website, we’ve had many ALICE clone submissions in the past. But for some kind of reason we now only get serious submissions, botmasters who really took the time to build that chatbot, and sometimes they also have a nice graphic and interaction design.

Yeah, I can see the issue Patti. It would be waste of time/resources to hold chatbot competitions if everyone is essentially submitting the same bot. Hard line to draw.

Erwin, I’m glad the quality of the bot submissions is high. Out of curiosity, are there any other standard bot types (besides AIML-based) that come up a lot?

Patti Roberts - Feb 26, 2011:

I think the “I just hacked off a limb” response would sure be a way to open a conversation .

I just don’t know which would be worse. To hack of your own limb, or someone else’s!

Considering the wide range of changes that I’ve made (and continue to make) to Morti’s AIML responses, I don’t think I would have had to worry about the “no clone” rule, had I entered him last year. 

Now, if I can just get the technical issues ironed out, I’ll be happy!

C R Hunt - Feb 26, 2011:

Out of curiosity, are there any other standard bot types (besides AIML-based) that come up a lot?

Last year there were a few inf bots (http://www.inf.net) which all had similar answers to each other. However, this type of clone was deemed to be ok, go figure!

Last year there were a few inf bots (http://www.inf.net) which all had similar answers to each other. However, this type of clone was deemed to be ok, go figure!

If I am not mistaken last year was the first year a inf net bot was entered. I think that is why the rules were changed. Last years contest was a bit confusing all around.  I am sure there won’t be the same mistakes this year.

Me too. 15 days to go everybody! I’m getting excited now!