Prompts for Presumed Normal

Starting the essay

Are the “Adaptive Reasoning Systems” slides a good seed for an essay for our blog, and is there a connection to “Answer Set Programming”?

Scope

Drop NARS and PLN from coverage. We can try drawing a connection to abduction, but that would be a novel result, I don’t see it right away. In the directory, you can also see my notes on ASP. For the essay, we should not bring up that these were my slides for a reading group, that’s not particularly interesting even biographically I think?

Decisions on the outline

Q1: “Presumed Normal” sounds best. Q2: (8) is out as written, not specific enough. Q3: not sure, so we can start with in-between: a box. Q4: I’d say induction and metaphor, as most relevant to cognition / AI. Question evocation sounds too explosive to me, and vagueness sounds like not a distinctive fit. WDYT esp. about vagueness? Q5: yes.

Line notes while reading the draft

“Reliability keeps the models whose abnormalities all lie in the unreliable set.” this is unclear. What’s the unreliable set and why?

A small parenthetical sentence explaining Cn_L is the consequence relation of logic L

Let’s add a box to explain Σ_3^0 -complete and Π_1^1 -complete, thanks!

Why isn’t “every extension that marks it has a further extension that unmarks it” Π⁰_2 sufficient?

That’s because “every extension that marks it” has internal Σ⁰₂ structure, right? Otherwise we don’t need to look for something to exist, we can start by quantifying over all marked lines.

Ah right, because we’re linearizing the proof tree. Can’t we do “cut” to only keep the necessary part of the tree?

Let’s add a separate box that works this out. Apologies for being dense.

“the Minimal Abnormality consequences are the sceptical conclusions, those supported in every extension. The Reliability consequences are the conclusions derivable from the intersection of the extensions.” sounds like this states the same condition: intersection of extensions equals members of every extension

“Under the naive, preferred and stable semantics” is a forward reference (right?), which is ok but needs to be marked (with a pointer)

Is the logic of here-and-there a proper three-valued logic?

Worth a box?

Go ahead, thanks! Not a worry for me (boxes are intended as optional, the essay is still relatively short).

“We have not found this identification stated as a theorem” – which identification precisely? ASP with Adaptive Logic of classical lower limit?

“the solver reports a sequence of improving answer sets, each usable, while it searches for an optimal one” – are the improving answers already stable models? What’s the condition of optimality?

This reads strange: “LA∀ even admits relational formulas, and the answer x≐y in that essay’s example is a relation.” Does it intend to separate propositional and monadic cases from general predicate logics? Is “that essay” the “What Must be True” one?

Does ASP ever count models? Is it used in applications where model counting, rather than cardinality of a model, is important?

Yes, this is connected to probabilistic logics and, for example, Pedro Domingos’ research program. It’s, I guess, not a good fit for this essay, though. Which puts me in a bind because I’m not sure it’s worth dedicating a whole essay to this topic.

Editorial pass

Very good. Let’s do a final editorial pass. Thanks.