I read an observation about artificial intelligence that went something like this:

"The more similar two options are, the harder it is to determine which
is better, and the less it matters."

(My paraphasing may not be totally accurate, but that's the best I can
remember it.)

Who made this observation?  Does it have a name?

---

Clarification of Question by rabbitrabbit-ga on 29 Sep 2006 11:22 PDT

I suppose this principle applies to any kind of decision-making,
whether or not it's done by computer.  If I recall correctly, I
originally read about this in a computer science context; it was
quoted as an epigraph in some article or book I read on the web.

---

Clarification of Question by rabbitrabbit-ga on 29 Sep 2006 11:37 PDT

This is similar to Buridan's ass (http://en.wikipedia.org/wiki/Buridan%27s_ass).

Does the observation have to apply just to artificial intelligence?
It seems that it has very general applications:  anything from
choosing a spouse (okay, only if they are twins) to making business
decisions.
I could envisage a business school professor using the statement after
letting his students spend the weekend wrestling with a case study
about two such similar options, pointing out that instead of wracking
their brains and punishing the numbers to find the better choice, they
should have recognized the overriding similarities and explained that
either option was acceptable  - or that both should be rejected.

Myoarin has a point that this could apply to many things in the world.
However, I can see how someone would explicitly say this about
artificial intelligence because people are deathly afraid of it.
People don't like to think that robots will be doing people's factory
jobs, or that if they call up a company they might get a robot. If the
robot sounds EXACTLY like a human does, however (which they eventually
will), they won't know and it won't matter.

This basically amounts to the Turing test-- a test developed by Alan
Turing to test whether artificial intelligence exists or not. If you
can't tell whether it's human or artificial, it's intelligent. Nothing
has passed the Turing test yet, but it will.

It's true that this principle applies to any kind of decision-making,
whether or not it's done by computer.  I call it AI because I remember
reading it in a computer science context -- I think it was quoted as
an epigraph in some article or book I read on the web.

It is not even a principle. It is a tutology or definition of 'similarity'

I mean tautology. Sorry.

But, Hedgie, as it refers to teaching ...

Should it not be 'taughtology'?

One application is in file fingerprinting but it is not a value issue
(ie better or worse), just that the more similar two files are the
more difficult it is to separate them or identify one uniquely. This
leads to a system of file fingerprinting or other indexing system....I
doubt this is what you mean, but I throw it in the mix in case it
might help