There are three common arguments made about voting by libertarians. The first is made by those who are too ready to sacrifice principle; that a person must vote for the lesser of two evils. This is a wasted vote. Voting for a Republican or a Democrat on the premise that the candidate from the other major party is worse really makes no difference. First of all most districts are safe districts and the winner of the general election is really determined in the primary. For state-wide races, such as senators or the president, most states are safe states as well.
Voting for the lesser of two evils doesn’t send the message the voter generally assumes will be sent. If a person votes against candidate A by voting for candidate B, all the vote totals show is another vote for candidate B. Votes do not come with labels saying "this is actually a vote against the other candidate", it is counted as a positive endorsement of all the flaws of candidate B.
Maybe in some particular races the counter argument could be made. In Minnesota, for example, the margin of victory in the 2008 senate race was less than the third party total, and Republicans bitterly complained about Libertarians allowing a Democrat to win the office - but Republicans forgot that the votes do not belong to the Republican Party, that they had not earned them, and in fact had acted in such a way as to guarantee that those voters will vote Libertarian. In attempting to make the lesser of two evils argument in that particular case, defenders of lesser of two evils not only undermine their own case, they support showing that the case is the exception and that most races are not nearly that close.
Except for the very extreme case, the vote for a third party will not decide the race between the major party candidates. In Texas, a voter trying to choose between Cynthia McKinney of the Green Party or Barack Obama of the Democratic Party will not swing that state. In California, a voter trying to choose between Chuck Baldwin of the Constitution Party or John McCain of the Republican Party will not swing that state. Also, a voter choosing between McKinney or Obama in California or Baldwin or McCain in Texas will not change the outcome.
But put it another way, take a theoretical voter trying to decide between Cynthia McKinney as his first choice or Barack Obama as the lesser of two evils candidate. If he votes for McKinney he increases her vote total from 161,603 to 161,604 (a percentage increase of 0.0006), but if he votes for Obama he increases his vote total from 64,639,738 to 64,639,739 (a percentage increase of 0.000000015). Clearly voting for McKinney would have a greater impact.
The two arguments on effective voting center on whether or not someone should vote at all. Well reasoned arguments are made on both sides of the issue. Those against voting are attempting to withhold consent from the state, consent that the state claims to have from participation in the system. Theoretically if someone votes, the person agrees to abide by the outcome of the election. Those in favor of voting, and not for the lesser of two evils, say that only by voting can the voice of the voter be heard, however faintly, alerting those in power to the wishes of the voter.
On the consent issue, the state has constructed an inherently contradictory case. If a person does vote the person is said to have given consent through participation; however, if a person does not vote the person is said to have given consent through not bothering to participate by expressing that the voter is content with any outcome. The two arguments contradict each other, but that is no concern to those who support gaining the illusion of consent. Whether or not someone votes, it is counted as consent, so therefore there is no reason to not vote; better to vote in a way that sends a clear message on the voter’s preference.
It is true that if there is no good candidate then there is no point in voting, but if there is actually a good candidate then by voting for that person it does increase, in however small a number, the chances that said candidate would win and does relay the message of who the voter actually supports and what the voter actually wishes of the government. Since the leaders of the major parties seem to believe that the votes belong to the major parties, by voting outside the two party framework a voter sends a disproportionately loud message by not "giving" to the parties that which "belongs" to them. The more voters fail to "deliver the goods" the more the parties start to work on how they can adjust to cause those voters to return. It is even possible, though unlikely, that the major parties could move towards greater liberty without ever attracting a pro-liberty vote. They will never move in that direction if they can get the pro-liberty vote without effort, or if they do not know that the vote is out there.
But there is one area in which the message sent by voting is unmistakable: ballot propositions. There are only two sides to a ballot proposition: yes and no. One of those options is clearly better, and there is no splitting of the vote with third options on ballot propositions. Sometimes the right side of the ballot proposition wins, such as Proposition 13, and sometimes it loses, such as Proposition 14, but every vote on a ballot proposition counts. There is no argument against voting for a ballot proposition.