<$BlogRSDUrl$>
Be sure to follow me on Twitter @leftiblog

Saturday, January 20, 2018


 

Twitter and Facebook and Russia: Consider the numbers


Remember how we were initially told that "Russia" (in quotes, because we have no idea who actually placed these ads, whether they had any relationship to the Russian government, or whether they were Russian at all) placed 3000 ads on Facebook? Only to learn later that 56% of them were placed after the election, 25% of them were never shown to anyone, 50% had ad budgets less than $3, and 99% of them had ad budgets less than $1000? Leaving approximately 14 ads seen during the election which had a budget exceeding $1000.And by the way virtually none of these ads mentioned Trump or Clinton at all; most dealt with issues like Black Lives Matter or gun rights. How that had an influence on the election is never quite specified. We're told this is because Russia is trying to encourage "divisiveness" in America. As if we aren't doing quite well on that score all by ourselves, with a little assist from FOX and MSNBC, and why that would effect that election anyway isn't clear. Oh and most of those ads were geographically targeted, and some of the ones that were were targeted to non-"swing states".



And then we were told by FB that those "Russian ads" reached "up to" 126 million Americans. There were 198 million Facebook users in 2016. According to Politico, 128 million people across the US generated 8.8 billion likes, posts, comments and shares related to the election. Which means that the ads reached 98.4% of all people who were posting or seeing posts about the election! Really? Bear in mind that are only 144 million Americans in the 18-65 age range, which is the default age range when you place an ad on FB (it can be changed of course). Which means that, according to FB, "Russian ads" reached a whopping 87.5% of all FB users aged 18-65. Again, really? Did you see any? I know I didn't.

Now along comes Twitter with its own math challenges. They tell us they found 50,000 (!) accounts "linked to Russia", which were followed, liked, or retweeted by at least 677,775 Americans, all of whom have received dire warnings from Twitter that their thinking may have been swayed by these tweets. 3,814 accounts were operated by alleged "Russian state operatives" connected to the Internet Research Agency (IRA), a company allegedly "linked to" the Russian government (how has never exactly been specified). Twitter said these IRA accounts posted 175,993 tweets during the election period, and 8.4% of these were election-related.

Wait, what? This was the big attempt to influence our election, but 91.6% of the output of these alleged "trolls" couldn't even be classified as "election-related" according to what were undoubtedly extremely loose criteria. So that's not 176K tweets, but 15K tweets. Is that a lot? Well, to begin with that's over an 18 month period, so we're talking about fewer than 1K tweets per month, or fewer than 30 per day*. What's Twitter's volume? According to them, more than 1 billion tweets about the election were sent out. 15K election-related tweets amounts to 0.0015% of that amount. Not 1%. Not a tenth of a percent. Not a hundredth of a percent One and a half thousandths of one percent.

*ADDED: Twitter says this was actually a 10-week period, which brings the number of tweets up to 200/day. But remember that's spread over 3,814 accounts, which works out to 0.06 tweets/day/account, or one tweet per account every 16 days! I hope those trolls aren't being paid much, they're not very productive! On average you'd have to be following 16 of those accounts just to see one tweet from them every day.

And what about the daily rate of all tweets? There are 500 million tweets sent out every day. 21% of Twitter users are American, and I'm going to guess (because I can't find the statistic) that they send out more than their share of tweets, so lets round up to 25% and say that Americans send out 125 million tweets/day (obviously people from other countries were also tweeting about the election, but we're going to disregard that). So if these Russian trolls were sending out 325 tweets a day on all subjects (not just the election), that amounts to 0.00026% of the tweets on any given day. 2 and a half ten-thousandths of a percent of all tweets being sent out. Have you stopped laughing yet?

None of this, by the way, goes to the effectiveness of these FB ads or Tweets. If you're following a rabidly pro-Trump account, chances are you're either a rabidly pro-Trump person yourself, or perhaps the opposite just monitoring the opposition. How many neutral people whose minds were susceptible to be changed follow such accounts? There has yet to be any measurement, or for that matter even anecdotal accounts, of such things happening. "Well, I was going to vote for Hillary, but then I saw this Russian troll account tweet that Hillary lost the debate to Trump, so I changed my mind." Really?

And now FB is going to let users rank news sources for trustworthiness. I can't think of a worse idea. Trump supporters will rank FOX as credible and everything else as "Fake News". The 99.5% of the population who have never even watched RT or listened to Sputnik will rank them as "untrustworthy", because they've been told repeatedly that's what they are, leaving the 0.5% of us who do to recognize they're both as, or more credible, than any U.S. corporate media (and carry viewpoints we can't get on the latter). Actually I can think of a worse idea, and that would be to let FB itself do the ranking. We've already seen what happens when Google does that, as progressive and left-wing sites, once easy to find when searching Google or Google News, have been harder or even impossible to find.

This page is powered by Blogger. Isn't yours? Weblog Commenting by HaloScan.com High Class Blogs: News and Media