The definition of Cause is circular

[alanruttenberg]

Cause:

A Process that is the cause of or one of the causes of some other Process.

[swartik]

So what about:

A Process whose existence explains, in whole or in part, the existence of some other Process.

[alanruttenberg]

What would be examples of processes whose existence doesn't explain, in whole or in part, the existence of some other process?

[swartik]

Assuming you believe the Universe will end, then yes, there is some final process. Frivolous examples aside, is there something wrong with saying every process is a cause of some other process? A knowledge graph doesn't have to assert every cause-and-effect relationship. If you are concerned about explicitness, how about the following:

A Process whose existence is recognized to explain, in whole or in part, the existence of some other process.

The "is recognized to" phrase intends to justify including the "is-cause-of some process" equivalence axiom. I thought of using "is necessary and sufficient to" instead, but it seemed too formal.

[alanruttenberg]

See also #728 (comment)

[neilotte]

@mark-jensen @johnbeve @mark-jensen To @alanruttenberg 's point, the definition is circular. If we maintain this class, we should address this. As Alan's link to the discussion indicates, I think we should strike this class completely. Thoughts?

[APCox]

Cause and Effect are (and should be) defined classes. They are defined in terms of the object properties is cause of and caused by, which in turn are defined in terms of one process being the consequence of another process.

The following makes the current definition of Cause less verbose:

A Process that is a cause of some other Process.

If people really don't like the use of "cause" in the definition of Cause and @swartik's proposal above is too verbose, the following is an option:

A Process that brings about some other Process.

[johnbeve]

As @APCox points out, "Cause" is a defined class and so no more circular than any other. "Cause" should then be understood in terms of processes, one of which is a consequence of the other.

This is not to say the definition of "is caused by" is satisfying. Point to a definition of causation that is. I take "is caused by" to be primitive here and so earns its keep based on what it can be used for, such as disambiguation. I actually think "is caused by" does a well enough job in this respect, as it provides grounds on which to distinguish specifications of causation, as represented in expressions such as:

1. The drought caused the famine. (Token-Token Causation)
2. Drowsy driving causes crashes. (Type-Type Causation)
3. How much I water my plant influence how tall it grows. (Token-Token Influence)
4. How much novocaine a patient receives affects how much pain they will feel during dental surgery. (Type-Type Influence)

The famine is plausibly a consequence of the drought, crashes a consequence of drowsy driving, a growing plant a consequence of watering, and felt pain a consequence of novocaine.

Cause and causation are notoriously ambiguous. I say we lean into the ambiguity by formalizing it explicitly. "is caused by" seems a reasonable foundation from which to start.