Skeletal pundit Styxhexenhammer666 talks at length about the nature of online polls, how reliable they are, and other matters here.
Quote from the beginning (transcribed):
All right, YouTube, it's time for more "perception versus reality" and a little bit more political analysis regarding debates. I've now sparred with about a dozen different people telling me, indeed, "The online polls were rigged!"—usually by 4chan or on Reddit or something like that. I'd like to point out that this only leaves a few possibilities as to how this could happen.
If Clinton won the debate, you'd think that people within any organic movement would reflect that in online polling. You'd expect, if energy and fervor for Trump and Clinton are both roughly equivalent, the aggregated polls overall—all of which are scientific, by the way, so I'm assuming you're going to trust them if you're trusting the CNN snap poll—show a dead-heat race, with Clinton leading by maybe a point: essentially a tied game. If that's the case, and if you would expect energy and thus, potentially, turnout to be the same on both sides, what you would most likely see is that, if there's rigging of the polls through 4chan by Trump fans, there would also be rigging of the online polls by Clinton fans on sites that they use: Tumblr, most of Reddit, as opposed to a handful of subreddits there, and a million other, sort of, avenues—through CTR, certainly; there is organized paid posting that goes on. Why would they not get involved? Wouldn't they have the technological literacy to do so?
It's a thirty-minute video, but Styx makes his main points within the first few minutes. I did find myself wondering whether saying "the polls are rigged" or "the polls are garbage" is an implicit admission that there is indeed a significant grassroots online groundswell happening for Trump (the non-legacy-media reality I've referred to in previous posts), even if it's in the form of poll-cheaters. There must be thousands of these poll-cheaters out there, which is quite impressive. Styx takes the issue further and asks why a counter-groundswell isn't arising from the Clintonistas, who are theoretically capable of the same sort of cyber-warfare. It's a fair question to consider. All of this fits the thesis that looking only at legacy-media data is unhelpful in understanding the reality of the situation.
I'm trying to be scientific about this by being willing to change my stance based on what I see. If a new hypothesis better fits the facts on the ground than does an older hypothesis, it's probably time to cast that old hypothesis aside, however unpleasant a task that might be. I'm still not on board the Trump train, but I'm now convinced, at the very least, that there are parties who can somehow see the US political situation better than I can, and whose analyses and predictions are more on the mark than those of the people I'd been relying on before (mostly mainstream media, in my case).