1.  Theoretical Optimization is fun.

Many of us here in the CO optimize for the sake of optimizing.  We revel in the rules, in configuring powerful characters, in being creative and in manipulating the system.  Long before I played in a practical D&D game, I had built many characters for the sake of building them.  Theoretical builds, using rules loopholes or ambiguous interpretations, overly powerful combinations, and clever creative uses of resources, are cool.  Many people don't like them, and that's okay.  The builds aren't for them.  They are for those of us who appreciate and admire the theory.  
Pun-pun was not designed to play, but to enjoy the idea that the 3.5e rules allow the creation of a character with unlimited power.  At level 1.
Nanobots exist within the structure of the 3.5e rules.
4e has its own twisted possibilities.
These are cool concepts, cool characters.  They are fun to create, fun to read about for many of us here in the CO.

2.  Theoretical Optimization is a useful resource. 

You can divide science into two categories: applied and theoretical.  Theoretical science may have obvious applications, or it may not.  Even when it does not, one day it may...and the theory will be there when it is needed.
Theoretical optimization pushes at the boundaries of the ruleset, exploring territory that may one day be used, or may never be.
Long before I developed the Arcane Slasher, other TOers produced Infinite Minion Bamfing and The Apocalypse Twins.  Both were theoretical, because the concepts weren't really the sort of thing you wanted to pull on a DM (banhammers being what they are).  Eventually I was able to turn the concept into something more practical (and yes, I also pushed some theoretical bounds at the same time).  Without those previous posters, I might never have seen the potential that lay hidden in the Feytouched and Long Night Scion PPs.
Furthermore, many DMs use TO ideas for their NPCs or recurring villains with good success.
Theory is a resource with many dead-ends, and many hidden gems.  It is good for the CO to have that resource available.

3.  Theoretical Optimization highlights rules problems.

One of the objectives of game design, especially in 4e, is a tight ruleset that presents a balanced game.  Balance is generally good for fun.  If there's a problem with the rules, you can either pretend it's not there, or fix it.  DMs and game designers alike have an incentive to know what needs fixing and what could happen if it doesn't get fixed.  Furthermore, DMs have many different ways of dealing with balance issues, and it's good for those DMs to have more information so that their methods can be better.  Information is good.

4.  Theoretical Optimization encourages good rules understanding.

Theoretical exploration of the rules encourages people to learn the rules so they can know what is valid and what is not.  I'm sure we can all agree that it is a good thing for people to know the rules of the game they play.

5.  Theoretical Optimization is still Character Optimization.

Even if you ignore the positive qualities of TO, it is still optimization.  For a time in 3.5e, there was a separate board for TO.  That is no longer the case.  As such, theoretical builds have no proper place outside the CO.  TOers have a right to post here as much as any of us have a right to post on these forums.  TO has a place here even if the only reason you will accept is that this is the only place that it COULD belong.



Theoretical Optimization belongs in the CO.
When someone posts a theoretical build or idea, ensure that people know it's theoretical, then judge it on its own merits.  The fact that it is theoretical is useful only to tell you what kind of discussion to have, not whether the discussion should take place at all.