Here’s Why You Shouldn’t Trust Those Online Polls That Say Trump Won

Sep 27, 2016

So who won the presidential debate on Monday night? Not surprisingly, supporters of Democratic candidate Hillary Clinton believe she won, and say they have the polls to prove it. Trump supporters, meanwhile, are convinced he won, and have their own polls that say so.

So in other words, this is just the usual argument between two candidates over which polls are more accurate, right? Not quite. There appears to be a bit more to it than that.

It's relatively common for polls about something like a debate to vary when it comes to the outcome, sometimes widely. But the difference between polls like CNN's after-debate survey and some of the online polls that media outlets created goes well beyond the usual variance.

In the CNN poll, which used responses from 521 people, Clinton was declared the winner by a large margin—62% thought she won, as opposed to 27% who thought Trump won (the rest were undecided).

Despite that, however, a number of online polls said the Republican candidate won, with some votes as high as 73%. In virtually all cases, those polls were non-scientific, meaning the makeup of those who responded did not necessarily reflect the population of likely voters. In many cases, the protections against people voting more than once was also minimal.

At least one of the polls appeared to be an outright fake: The Drudge Report and a number of Trump supporters linked to a supposed ABC poll showing that Trump won 54% of the vote and Clinton just 10%, but the poll was hosted by a site that mimicked ABC's website.

Based on a number of posts on Twitter, the anarchic online community 4chan, and online message board Reddit, it appears as though Internet trolls ganged together to try to rig the online polls by using a number of tricks to game the results so that they would show Trump as victorious.

Get Data Sheet, Fortune's technology newsletter.

4chan, in particular, has a large "alt-right" community that is responsible for a number of racist pro-Trump memes, including Pepe the Frog. And Reddit users have been involved in rigging online polls before, including a poll by Time (which like Fortune is owned by Time Inc.) in 2009 for the most influential people in the world.

That poll was manipulated so that 4chan founder Christopher "Moot" Poole would win first place. And in addition to doing that, the trolls from the site rigged all of the top 10 so they would spell the code word "marblecake."

It's not clear how much of the result from a variety of online polls was the result of this kind of rigging, but it's obvious from a number of posts that there was a concerted effort to game the results in favor of Trump —despite the fact that hackers were upset by his reference during the presidential debate to a hacker as "some 400-pound guy sitting in his bedroom."

One of the techniques referred to in a number of the Reddit and 4chan posts—as a way of getting around the usual restrictions on online voting—was to vote using a smartphone and use the device's "airplane mode" to rapidly change Internet addresses, so that each vote would be recorded as a new one.

In the past, Reddit trolls have used fairly sophisticated bot-nets made up of hundreds or even thousands of computers to game the results of online polls. In some cases, such polls can also be manipulated by changing browsers or simply using multiple devices and email accounts.

All products and services featured are based solely on editorial selection. FORTUNE may receive compensation for some links to products and services on this website.

Quotes delayed at least 15 minutes. Market data provided by Interactive Data. ETF and Mutual Fund data provided by Morningstar, Inc. Dow Jones Terms & Conditions: http://www.djindexes.com/mdsidx/html/tandc/indexestandcs.html. S&P Index data is the property of Chicago Mercantile Exchange Inc. and its licensors. All rights reserved. Terms & Conditions. Powered and implemented by Interactive Data Managed Solutions