**A program for Getting to known the user
**How do you teach Hal to guess what the user is going to want?
I think that we can use the tools of the fortune-teller or a physic. There is a show on the Sci-Fi channel. I’m not sure how the show works. It could be that there actually is the ability to channel into unknown powers, but thinking about the conversations they psychic host has with the audience and eventually the guests on the show is a brilliant exercise in deductive logic. By learning about the relationship between the questions and answers we learn about the common denominator of conversation.
To begin with everyone in the audience knows what kind of show they’re on. Presumably it is possible to screen the audience to remove or isolate (identify) the gawkers and those who would debunk or mock the conversation.
Once the conversation begins the host can make one assumption correctly. Everyone in the audience has known someone who has died.
Armed with this knowledge, the host can begin asking open ended questions and listening to conversations. The more a person will openly speak the more information the host can extract from the conversation. With a more knowledgeable history of the individual and the conversation he can form a hypothesis, or a model of the conversation.
Once an accurate model is found the conversation can be continued in a natural progression.
Information that can help determine the model includes:
Infered
Language
Tone
Body Language
Tempo
Stated
Home Life
Attitude
Reactions
A common denominator is a series of number that Numeral can be divided by. To make a relation;
A numeral is a subject
A devisor is a question that will relate the numeral to a field.
A field is a set of number with a true connection
By correctly identifying several positive or negative results to field comparison through division a hypothetical model can be created. The model is simple, however it can aid in identifying a still larger common denominator.
I will for the sake of argument say that you need have four answers to determine a model. These answers must be derived from questions whose angles are radically different. That is to say, the answers must be able to represent a position in three dimensional spaces.
If the correct questions are asked a reasonably complex model of probability can be extrapolated. That model could represent up to three correct answers.
These new incorrect or correct responses can make a further clarification of the model.
Granted, Hal cannot use the same non-verbal cues a human host would use, he can use that type of question looking for paters in database response.
Is everybody still in the boat? That seemed like rocky writing to me.
WholyChao
