Can Google Rig Elections?

August 23, 2015 1:40 PM

A fascinating article by Robert Epstein at Politico.com about Google's Search Engine Manipulation Effect (SEME) work, and whether and how they'd be able to influence an election.

"Given how powerful this effect is, it’s possible that Google decided the winner of the Indian election. Google’s own daily data on election-related search activity... showed that Narendra Modi, the ultimate winner, outscored his rivals in search activity by more than 25 percent for sixty-one consecutive days before the final votes were cast. That high volume of search activity could easily have been generated by higher search rankings for Modi."

posted by sneebler (34 comments total)

This article is completely goofy.
posted by Nevin at 2:09 PM on August 23, 2015 [7 favorites]

The answer google gives may surprise you!
posted by boo_radley at 2:13 PM on August 23, 2015 [2 favorites]

"That high volume of search activity could easily have been generated by higher search rankings for Modi.", from the article, pretty clearly demonstrates that they are disingenuous or don't really understand search engines.
posted by sagc at 2:15 PM on August 23, 2015 [9 favorites]

A bit reminiscent of Interface, though.
posted by sagc at 2:17 PM on August 23, 2015 [1 favorite]

From a Charles Pierce article last year;
"That would be Politico, aka Tiger Beat on the Potomac, which manages on a regular basis to cover the worst of our politics through the worst of our political journalism. It is a remarkable parlay to have achieved."
This piece - like so many in TBOTP - may contain something real, but it's wrapped in tripe and bacon and deep fried in stale fake oil.
posted by Alter Cocker at 2:18 PM on August 23, 2015 [7 favorites]

I'm sure Google could, in some small way, influence elections (rig is certainly not the right word). I'm also sure that this article is just a confused mess. Normally, we might blame a lazy journalist, but since this is written by one of the researchers, the article surely doesn't give one much confidence in the "gold-standard" research.

How about this? "we were able to boost the proportion of people who favored any candidate by between 37 and 63 percent after just one search session." That might have been true in whatever contrived scenario they set up for their research, but to suggest that it might be true in the real world is ludicrous.
posted by ssg at 2:22 PM on August 23, 2015 [1 favorite]

"Under this scenario, all of Google’s employees are innocent little lambs, but the software is evil."

Some very floridly-described research.
posted by sagc at 2:25 PM on August 23, 2015

From the author's Wikipedia page:

In 2012, Epstein publicly disputed with Google Search over a security warning placed on links to his website.[10] His website, which features mental health screening tests, was blocked for serving malware that could infect visitors to the site. Epstein emailed "Larry Page, Google's chief executive; David Drummond, Google’s legal counsel; Dr. Epstein's congressman; and journalists from The New York Times, The Washington Post, Wired, and Newsweek."[10] In it, Epstein threatened legal action if the warning concerning his website was not removed, and denied that any problems with his website existed.[10] Several weeks later, Epstein admitted his website had been hacked, but still blamed Google for tarnishing his name and not helping him find the infection.[11]

So clearly he's someone who understands how the internet works... I'm going to read his actual paper now, because this article is nonsense and apparently there's something wrong with me.
posted by zachlipton at 2:28 PM on August 23, 2015 [19 favorites]

I thought the Veep episode "Clovis" was fun, where the search engine owner effectively trades tax breaks for search result placement. Nothing explicit is said or done, of course, because algorithms are sacrosanct and Google Clovis would never, ever preferentially bump up other content over pure search results lol!
posted by a lungful of dragon at 2:38 PM on August 23, 2015

It’s hard to imagine Google ever degrading its product and undermining its credibility in such ways, however. To protect the free and fair election, that might leave only one option, as unpalatable as it might seem: government regulation.

Exactly what is the US Government going to regulate, the ranking algorithm itself? That would require intimate knowledge and direct control over the hardware and software that enables the algorithm. The US Government would effectively end up running Google search.

A far less silly option would be to statistically analyze the SERPs for signs of tampering, maybe compare the results with those from other search engines or collect SERPs data from end users? But then again the ranking algorithm is such a black box that millions of SEO experts have failed to understand to map its inner workings, so I'm not sure if it's even possible to truly understand it from the outside.
posted by Foci for Analysis at 2:50 PM on August 23, 2015

Proof that anyone can write anything on the Internet, here we have an article about elections and search engines written by a researcher who understands neither.
posted by Itaxpica at 3:04 PM on August 23, 2015 [4 favorites]

Politico, aka Tiger Beat on the Potomac

apply cold water directly to burn
posted by sidereal at 3:10 PM on August 23, 2015 [9 favorites]

For extra fun, read the comments.
posted by octothorpe at 3:34 PM on August 23, 2015

Robert Epstein is senior research psychologist at the American Institute for Behavioral Research and Technology and the former editor-in-chief of Psychology Today. Follow him on Twitter @DrREpstein.

Have you noticed that Psychology Today is a sensational rag?
posted by grobstein at 3:38 PM on August 23, 2015 [2 favorites]

Oh my god, the comments. I now more stupider than before. comments
posted by Johnny Wallflower at 3:53 PM on August 23, 2015 [3 favorites]

But - can Snopes rig an election?
posted by Going To Maine at 4:09 PM on August 23, 2015 [9 favorites]

For an article about misinformation, they sure are pushing you towards a conclusion that isn't supported at all by their research.

It's a shame because the underlying topic is very much worth being discussed.
posted by zixyer at 4:21 PM on August 23, 2015 [1 favorite]

I would not be shocked if this was somehow supported by Google's business competitors and / or political opponents.
posted by grobstein at 4:32 PM on August 23, 2015 [3 favorites]

I'm feeling lucky.
posted by JackFlash at 4:35 PM on August 23, 2015 [1 favorite]

Google, Inc., has amassed far more power to control elections... than any company in history has ever had.

For the 2004 presidential election, I believe that distinction belongs to Diebold. Though malfeasance was never proven, electronic voting machines were in charge of counting 22.1% of the votes, with an additional 29.6% of ballots being counted via optical scan.
posted by fragmede at 5:16 PM on August 23, 2015

It's a shame because the actual research is somewhat interesting, though it doesn't prove nearly as much as Epstein seems to want it to. Meanwhile, the Politico article goes off in all these angles about whether Google is manipulating elections by a top-down edict or through rogue employee and labeling the Search Engine Manipulation Effect (SEME) (a label that implies that there's actual manipulation already happening) "one of the largest behavioral effects ever discovered" (I'll give you a hint: repressive dictatorships have long ago figured out how to shift far larger proportions of the vote). Their studies also give widely divergent values for the size of the effect, which is interesting.

(As a sidenote, nothing in their paper indicates that they ever debriefed their participants about the manipulation after the study. This is particularly concerning with the study they did of 2,150 people in India during an actual election. The researchers handwave over the ethical concerns in the obligatory ethics paragraph by saying that 2,150 people isn't much compared to 430 million votes and that their study equally favored the candidates, so any impact should cancel out. To me, this is not enough to justify going to a foreign country and screwing with a few thousand people in the middle of their election, especially without revealing their manipulation to the participants after the study.)

The issues raised in the studies are worth further explanation, but they are largely just political adaptations of the much broader conversation about the role search engine rankings have on commerce and the spread of ideas. That's an important conversation, and it ties in with the open questions around echo chambers on social media, but it's irresponsible to call it "a serious threat to the democratic system of government."

It also seems to me that the issues raised are somewhat more applicable to small local races, where voters generally make their decisions on fairly limited information about candidates, rather than the 2016 US election, where billions of dollars will be spent on every side to deliver messages and different coverage is provided by every media outlet on the planet from Fox News to The Sportsman Channel to the (North) Korean Central News Agency.

Readers might also be interested in Jonathan Zittrain's article from last year on Facebook's get-out-the-vote campaign and how it hypothetically could be used to rig an election if targeted for evil. Zittrain, of course, understands the internet.
posted by zachlipton at 5:26 PM on August 23, 2015 [7 favorites]

For those interested in more serious takes on the regulation of search engines more generally, Mefi's own James Grimmelman wrote probably the most important piece on it, here.
posted by louie at 7:26 PM on August 23, 2015 [5 favorites]

So no one googled what "seme" means, huh? Because I could tell you some of the top results...
posted by maryr at 9:02 PM on August 23, 2015

sidereal: "Politico, aka Tiger Beat on the Potomac

apply cold water directly to burn"

Nope. Hot water, as that was an ice burn.
posted by Samizdata at 9:07 PM on August 23, 2015

sidereal: “Politico, aka Tiger Beat on the Potomac

apply cold water directly to burn”

Nope. Hot water, as that was an ice burn.

Given that Pierce has used that particular burn so often that there are ten pages of search results for it on Esquire, I think it’s kind of lost its sting (if not its truth).
posted by Going To Maine at 9:27 PM on August 23, 2015

Doesn't Google have more money than god? Or at least more money than anybody but Apple? Why couldn't they influence elections the same way every other wealthy corporation does?

Or perhaps the point is, they can achieve this level of influence without spending a nickel of the money they've accumulated by watching everybody like a hawk (an elephant-memoried and infallible hawk who is still somehow our special friend.)
posted by spacewrench at 10:16 PM on August 23, 2015

IIRC, a few years ago (last election cycle?) designers were pissed that googling for Paul Rand returned Rand Paul results something like 8 for every 10. Bit of it is an unfortunate coincidence that candidates with names like Handel shorten it to allude to their goddess, but a few people suggested people in Google let it slide to increase the exposure of a marginal candidate.
Right now, I'm getting mostly P. Rand results (then again; not in the US, google knows I'm way more into design than senators from Kentucky with roadkill on their head and R. Paul is apparently struggling for funds this time around, and it's not cheap to pay enough people to make "Rand Paul 2016" posts on every internet forum, blog comment box or facebook post)
posted by lmfsilva at 4:16 AM on August 24, 2015

I think the most important point here is really being obscured by the pulpit-pounding tone of this article: search results have such a powerful effect on voting preferences that search algorithms must already influencing results, even if those who design them are good actors and doing their best to remain "neutral".

In particular, it makes me wonder if Google's (and other engine's) attempts to individualize results, returning more pages that are similar to ones it thinks you've already been pleased with, has anything to do with what seems to be a widespread pattern of increasing polarization in political opinions. It's certainly not the only factor at work in this, but it seems like it might be more influential than I'd have thought.

And given the history of really, really bad actors in US elections, I don't think it's unreasonable to be concerned about all this influence being in the hands, for all practical purposes, of one company, particularly one that actually makes its money not on search, but on advertising (that is, in market for influencing over opinions). There don't seem to be deliberately malicious folks there now, at least in this space, but it's a disturbing amount of influence to see in any single set of hands.
posted by CHoldredge at 4:37 AM on August 24, 2015 [2 favorites]

To protect the free and fair election, that might leave only one option, as unpalatable as it might seem: government regulation.

Or, ya know, people could actually use more than one search engine ... No government intervention required.

I'm more concerned about Google's learning effect: if you search a lot for a particular candidate, then you're going to get a lot more results focused on that particular candidate (and less exposure to competing viewpoints, perhaps). Doesn't necessarily favor any particular candidate (but would tend to favor front-runners), but could lower the amount of public debate.

(I was about to write "lower the quality of public debate", but this is politics we're talking about ...)
posted by oheso at 5:51 AM on August 24, 2015

Or, ya know, people could actually use more than one search engine ... No government intervention required.

Google's monopoly issues are a little more complicated.
posted by a lungful of dragon at 12:07 PM on August 24, 2015

candidates with names like Handel

Handel; so hot right now.
posted by stebulus at 4:19 PM on August 24, 2015

The real question is, if Google really did decide to influence an election, would we be any worse off?
posted by Pipedreamergrey at 11:10 PM on August 24, 2015

Canada goes to the polls at the beginning of October.

If you want to learn more about the candidates (in our Westminster first-past-the-post system we vote for a party in local ridings, and the leader of the party who wins the most ridings or seats becomes PM) you don't use Google Search or whatever, you go onto Facebook or perhaps Twitter. Or read the online edition of the newspaper.

I tried using Google to search for candidates in a couple of neighbouring ridings (seats) and I couldn't find any information, mostly because the candidates are still in the process of being ratified following nominations and haven't set up websites yet. So I just went to the Facebook pages of the various political parties.

In the American presidential system, besides the utterly confused premise of the article here, the writer also doesn't investigate how much work the Obama campaign did to identify voters and get them out to vote. Obama did that in two elections. That's how he won.

Where I live in Canada the Green Party is doing the same thing: identifying potential voters, and then making sure they vote. A Green MP nearly got in in a by-election a couple of years ago, and the Green campaign machine helped the new Green-aligned mayor of Victoria squeak in a few months ago by identifying voters and getting them out to vote.

The same campaign team is doing the same thing locally for a candidate, and they have a very good chance of winning.

So I think the writer just doesn't understand how campaigning works either.
posted by Nevin at 3:36 PM on August 25, 2015

The real question is, if Google really did decide to influence an election, would we be any worse off?

We'd probably be worse off. The company is run by a someone who is comfortable with the idea of Google and the government as a whole looking into our private lives at their leisure.
posted by a lungful of dragon at 4:17 PM on August 25, 2015