They're not real "polls," people.
A brief etymological digression on the word "poll," per Oxford Dictionaries:
Middle English (in the sense 'head'): perhaps of Low German origin. The original sense was 'head,' and hence 'an individual person among a number,' from which developed the sense 'number of people ascertained by counting of heads' and then 'counting of heads or of votes' (17th cent).
The concept of "one person, one vote" is pretty explicit in that etymology and in our modern understanding of what a "poll" is about. And when it comes to what a poll is for, another important element is that the people being polled are plausibly the same people who will actually make the decision in question. That's why telephone polling for political campaigns looks for "likely [insert party here] voters," not just any old respondent.
And those are the two major problems with the Internet "polls" I'm referring to. They're trivial to stuff using e.g. web proxies and so forth, or for a campaign to flash-mob with an alert to people who like the candidate but will not be casting a real vote in the real relevant election.
They are, in other words, un-scientific and pretty much meaningless.
They're also sad, because between 2004 to 2008 the Libertarian Party actually began developing a real organic capacity for conducting and using real scientific polling.
In 2004, Libertarian presidential nominee Michael Badnarik's campaign commissioned several (IIRC, the number was five) polls from a reputable company (Rasmussen) for the general election cycle.
After 2004, Stephen Gordon (with whom I worked on Aaron Russo's pre-nomination campaign and Badnarik's general election campaign) ran a polling operation leading up to 2008 in which actual previous and likely national convention delegates were contacted -- one person, one vote, with an attempt to identify likely delegates -- and polled on their preferences.
And then that growing organic polling capacity just ... disappeared.
In the first century AD, Hero of Alexandria built a steam engine. It was apparently briefly a sensation in the Roman/Egyptian royal courts.
Then the steam engine just ... disappeared, for about 1600 years, before re-appearing as a world-changing thing.
I hope it's not that long before Libertarians return to a serious interest in polling.