Latest Conversation Spaghetti Chart

Well, here it is, then: the current version of the spaghetti chart for conversations. This is rather large, so I shrank it so that you can see the whole thing in one screen:

The new glyphs are all in the lower section. The top row of this consists of two glyphs: “that’s true” and “that’s not true”, which are modified by uncertainty adverbs. The middle row has “maybe I’m wrong”, “that’s true”, and “X told me that” (notice the clock with the past indicating arrow). The bottom row has “meh…”, “that’s not true”, and “goodbye”. 

Following up on reports
Now I must address that glyph on the upper right, the one with just a big question mark. That means “ask”, and it is meant to provide a check against reports. Suppose that Fred decides to tell a lie about Joan to Tom. He tells Tom that Joan badmouthed Tom to him. Tom might find this difficult to believe, and would therefore wish to follow up on this report by asking Tom himself. 

Of course, this might be too fine a consideration; after all, is Tom really going to admit that he badmouthed Fred? Of course not! Therefore, asking him would be pointless. The only way that Fred could get a reliable assessment of the report is to go to a third party who might have heard. But who would have heard? 

Oops, I forgot about spies in hiding. Suppose that Dave was in hiding and observed Mary badmouthing Joe to Mike. Then Dave could tell Joe that Mary badmouthed him to Mike. This would permit Joe to ask Mike about the veracity of the report, and Mike would have no compelling reason to deny it, other than his relative loyalty to Mary versus Joe. 

The conclusion is that verification should take place only when the reporter is neither Subject nor DirObject of the report. 

Planning versus Opportunism
This raises another problem: how should Joe handle the matter of verification? He could either make a plan to verify with Mike as soon as he receives the report; or he could, upon next meeting Mike, search through all previous reports for things to verify. I think that the latter would involve complex and time-consuming searches through the HistoryBook. Better, I think, to rely on the planning mechanism. There’s a hitch with this, though: how does the planning get inserted into the conversational flow? Refer to the spaghetti chart above. Ah, I see: as a Subject reaction to the “that is not true” verb. Yep, this can be made to work. But now I must move the “ask” verb to its new location.

Figuring the credibility of a report
After some deliberation, I have decided to use a rough Gaussian curve to calculate the credibility of a report. The starting point is the translation of a report into its personality implications. For now, here is my translation table:

Badmouthing: Good = BInverse(Quantifier)
Betrayal: Trustworthy = −0.6
Lying: Trustworthy = −0.3
Told aura count: value of reported aura count
Told lost an aura: no implication

I just realized that this table must also reflect the Desirable scripts used to choose these options. Another thing: uncertainty applies only to estimates of aura counts. There simply is no point in declaring that somebody may or may not have lied; either you know they lied or you don’t. That in turn means that all of the reports are either truthful or stone-cold lies. Of course, the listener must still make a decision as to the plausibility of the report. This will be tricky with a report of telling aura counts.

Joe witnesses Mary telling Tom that Fred has 2 shial pons, with a level +1 certainty. Joe then reports to Jane that Mary said that Fred has 2 shial pons, again with a level +1 certainty. Jane must evaluate the credibility of this report. Here’s where we get into the trickiness. She has four numbers to process: her own pon count for Fred, her certainty of this number, and the two reported values. I’ll have to handle this with a custom method in the Interpreter.

But other problems loom. Consider what the report of telling aura counts will look like:

demosentence

“Camiggdo tells Koopi that Caronycoorck told Skordokott that Zubi has 2 good aura pons, 
but Caronycoorck is moderately uncertain about the count.”

This is too damned long. It occupies too much screen space: 700h by 340v. It’s too difficult to understand. It’s too long to input. The dilemma is that this sentence is also useful. But I think I’ll have to dispense with it. Ouch.