Re: What is the definition of "definition"?

Eliezer Yudkowsky (
Mon, 03 Feb 1997 03:35:33 -0600

[John K Clark:]
> I have not the slightest idea because your sentence makes no sense to me,
> you are using words I do not understand. Remember what the subject title of
> this post is.
> Since I adequately nerdsnaped "symbol formation", what's left to nerdsnap
> about "diphelzeation"?

Yeah, so you asked a circular question. So what? How is that my
fault? I still gave you a non-circular answer. Besides, now that I
understand that you're objecting to your own question rather than my
answer, it's still easy enough to get out of. First, I use the
subjective definition I scorned earlier so you understand what nerdsnap
and diphelzeation are, and then I give you the cognitive definition.
Still easy as pie. So what new alteration will you make to the rules
now, to prevent me from winning?

The subjective referent of "explanation" is easy enough to establish: I
ask you a few "why" questions, and then say: "See? You just explained
something." If you think this is circular because of "why", I can
exercise my subjunctive omnipotence and videotape your own brain as it
goes about its course, and flash a red light whenever I detect you
explaining something to yourself. Then a red light symbolizes

This does presume that you start out by explaining things to yourself,
but any human *will* do that. And if you DON'T start out by explaining
things to yourself, I'll take a <bleep>ing AI that I programmed with my
own <bleep>ing bare hands and print out the <bleep>ing assembly and
define "definition" right down to the basic <bleep>ing constituents of
computational causality!

And finally, if you ask for a non-circular definition of "definition",
and I give you back an explanation with the note "this is a definition
of 'definition'", and then you refuse to know that I am answering your
question, so that my non-"explanation"-mentioning cognitive explanation
doesn't even get processed, you can take your <bleep>ing <bleep> and
<bleep> it up your nose. Of course defining "definition" requires you
to enact the cognitive events associated with definition! That doesn't
make it circular. Circular reference requires that the explanation
refer to itself, not that the process of comprehending the explanation
refer to the explanation!

Our "visual cortex" is areas 17, 18 and 19 of the brain. Even though
you took that in through your visual cortex, "visual cortex" is NOT a
circular definition! "Computer screen" is NOT a circular definition!
"Quark" is NOT a circular definition! And "sticky note" is NOT a
circular definition, EVEN IF we have to refer to some actual sticky
notes in order to define it!

You are torturing the rules of the game! Defining definition requires
that we *use* our cognitive facilities for definition, not that we
*refer* to a previously constructed symbol *representing* our cognitive
facilities for definition. It isn't circular! Give it up! Definition
is part of objective reality just like everything else, and can be
reduced to its causal constituents like all good computational

--       Eliezer S. Yudkowsky

Disclaimer:  Unless otherwise specified, I'm not telling you
everything I think I know.