Predicates that take two parameters are sometimes called binary predicates.  A useful debugging tool for binary predicates is that we visualize them as graphs, i.e. networks of nodes with lines or arrows between them.  If [Predicate a b] is true, then we include a and b with an arrow from a to b.  For example, for our family example:

[predicate]
Siblings ?a ?b: [Parent ?a ?parent] [Parent ?b ?parent] [Different ?a ?b]

[predicate]
Parent bart marge.
Parent bart homer.
Parent lisa homer.
Parent lisa marge.
Parent stan sharon.
Parent stan randy.
Parent randy grandpa.
Parent jimbo grandpa.

The Parent predicate looks like this:

graph TD
n0("bart") --> n1("marge")
n0 --> n2("homer")
n3("lisa") --> n2
n3 --> n1
n4("stan") --> n5("sharon")
n4 --> n6("randy")
n6 --> n7("grandpa")
n8("jimbo") --> n7

and the Siblings predicate looks like this:

graph TD
n0("bart") --> n1("lisa")
n1 --> n0
n2("randy") --> n3("jimbo")
n3 --> n2

You can generate these visualizations yourself using the VisualizeGraph task.  Simply say [VisualizeGraph ?predicate], where ?predicate is some predicate that takes 2 arguments and it will draw the graph in your browser.  Try it out:

# Try: [VisualizeGraph Parent]
[predicate]
Siblings ?a ?b: [Parent ?a ?parent] [Parent ?b ?parent] [Different ?a ?b]

[predicate]
Parent bart marge.
Parent bart homer.
Parent lisa homer.
Parent lisa marge.
Parent stan sharon.
Parent stan randy.
Parent randy grandpa.
Parent jimbo grandpa.

Unbound variables in visualizations

Returning to our food preferences example:

# Try: [VisualizeGraph Likes]
[predicate]
Likes tanya sushi.
Likes tanya burgers.
Likes tanya mexican.
Likes jayden burgers.
Likes jayden ethiopean.
# Kimiko likes everything
Likes kimiko ?.
# Everyone likes pizza.
Likes ? pizza.

When we visualize it, we get the somewhat odd visualization:

graph TD
n0("tanya") --> n1("sushi")
n0 --> n2("burgers")
n0 --> n3("mexican")
n4("jayden") --> n2
n4 --> n5("ethiopean")
n6("kimiko") --> n7("?endNode9")
n8("?startNode8") --> n9("pizza")

Where we have graph nodes like ?startNode8 and ?endNode9 that are variables rather than real values.  That's because of rules like:

# Kimiko likes everything
Likes kimiko ?.

which mean that when you ask for all solutions to [Likes ?startNode ?endNode], which is what the graph visualizer does, one of the solutions you get is ?startNode = kimiko, ?endNode=?, i.e. ?endNode can be anything, i.e. “Kimiko likes everything”.  But of course, the intention isn't that Kimiko doesn't like everything.  Kimiko may not like nuclear war, for example.  The intention of the rule above means “Kimiko likes all foods” and so the variable there is intended only to be used with foods.  But the system doesn't have a list of all foods, so it doesn't know what to do with the variable and it just renders it as ?endNode9.

We can fix this buy giving it explicit generators for the set of characters and the set of foods.  We can say that someone is a character if they like something and food if it's liked by someone:

[predicate]
Character ?who: [Likes ?who ?] [NonVar ?who]
[predicate]
Food ?what: [Likes ? ?what] [NonVar ?what]

The NonVar predicate, which we'll talk about later, says only to accept the result if ?who has a value.  That filters out the results from the Likes rules that have ?s in them.

Then we can have a version of Likes that never returns variables in its queries:

[predicate]
LikesExhaustive ?who ?what: [Likes ?who ?what] [Character ?who] [Food ?what]

And then we get the full visualization:

graph TD
n0("tanya") --> n1("sushi")
n0 --> n2("burgers")
n0 --> n3("mexican")
n4("jayden") --> n2
n4 --> n5("ethiopean")
n6("kimiko") --> n1
n6 --> n2
n6 --> n3
n6 --> n5
n6 --> n7("pizza")
n0 --> n7
n4 --> n7

This is “better” in the sense that it explicitly shows all combinations of who liking what.  But it's also a mess, so you might prefer the version with variables.

You can try it out:

# Try: [VisualizeGraph LikesExhaustive]
[predicate]
Likes tanya sushi.
Likes tanya burgers.
Likes tanya mexican.
Likes jayden burgers.
Likes jayden ethiopean.
# Kimiko likes everything
Likes kimiko ?.
# Everyone likes pizza.
Likes ? pizza.

[predicate]
Character ?who: [Likes ?who ?] [NonVar ?who]
[predicate]
Food ?what: [Likes ? ?what] [NonVar ?what]
[predicate]
LikesExhaustive ?who ?what: [Likes ?who ?what] [Character ?who] [Food ?what]