Is choice always good in open source software?

In open source software communities, choice is often touted as a good thing.

While I agree with this in principle, I do not think that it
is always the case.

When choice is bad: Expense, confusion, low adoption

If choice results in confusing the adopters, increasing the
cost for developers, and fragmenting the market, resulting in less adoption, then it is not a good
thing.

An example is the KDE vs. Gnome, and how this has hurt
the adoption of Linux on the desktop. Corporations will write their
application for one or the other, but never both. The look and feel is
different and so is the user interface.

A similar story, mostly in the past, is the
large number of distros with different packaging systems, and the inability of one distro to run the packages of another. This is now less due to the consolidation we have seen recently into Debian/Ubuntu, Red Hat and SuSE.

When choice is good: Emacs vs. vi

Choice is good when it does not affect the API or adoption: for
example in the Emacs vs. vi debate. It does not matter what the programmer choses to write his software on the end users. It does not limit the end user's choice, or confuse the market.