From what I understand, the best reason that the current system favors two parties is Duverger's Law ( http://en.wikipedia.org/wiki/Duverger's_law ), which states that a first-past-the-post election system will naturally lead to a two-party system.  For this reason, I would think campaigning for alternate voting methods would be a major concern of the Libertarian party, and indeed, any third party.

A search of the lp.org site for "approval voting" only turns up this LP News article ( http://www.lp.org/lpnews/article_62.shtml ) about instant runoff voting, in which a few members mention approval voting as another possible alternative. If you haven't heard of approval voting, please see the first link at the bottom of this post.

I believe strongly in the potential of approval voting as a simple adjustment to the current system, and have my reservations about instant runoff voting due to its failure of several mathematical criteria (see http://en.wikipedia.org/wiki/IRV#Criteria_failed ).  Eventually, a move to an even better ranked choice voting system could be pursued, such as Schulze method, but this is a big change and one that I think should wait until the change has been made to approval voting.  I think that the Libertarian party would do well to have a consistent stance on this issue, and mention that position in its platform.  I voted for Michael Badnarik in the 2004 election, but in 2008 I'd like to be able to convince others to vote for both their favored major party candidate AND the Libertarian party candidate.

Let's discuss this a bit - I could be wrong in my support of approval voting, and certainly reasonable minds could disagree.  Mathematically though, while there is no such thing as a perfect voting method, approval voting seems to be the strongest method that doesn't require too much change from our current system to be enacted.

Some background material:
http://electionmethods.org/approved.htm
http://electionmethods.org/IRVproblems.htm
http://en.wikipedia.org/wiki/Approval_voting
http://en.wikipedia.org/wiki/Schulze_method
http://www.condorcet.org/rp/IRV.shtml