Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
Breaking the Information Monopoly?
This story is not yet trending on Google news. (I’ll be curious to see if it does, although I won’t draw any conclusions from it.)
Robert Epstein is a research psychologist. (I’ve drawn this conclusion by using Google to look him up and peruse his publications.) Here’s his Wikipedia page. (It was the top-ranked entry when I searched for “research psychologist Robert Epstein.”)
It says he was born on June 19, 1953. He’s “an American psychologist, professor, author, and journalist.” It also says has a doctorate from Harvard. I believe all of that, although I’ve confirmed none of it independently. I also believe this, though I haven’t confirmed it independently:
In 2012, Epstein publicly disputed with Google Search over a security warning placed on links to his website. 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.” In it, Epstein threatened legal action if the warning concerning his website was not removed, and denied that any problems with his website existed. 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.
Epstein has just published a piece in Politico warning that Google might throw the 2016 election:
America’s next president could be eased into office not just by TV ads or speeches, but by Google’s secret decisions, and no one—except for me and perhaps a few other obscure researchers—would know how this was accomplished.
Research I have been directing in recent years suggests that Google, Inc., has amassed far more power to control elections—indeed, to control a wide variety of opinions and beliefs—than any company in history has ever had. Google’s search algorithm can easily shift the voting preferences of undecided voters by 20 percent or more—up to 80 percent in some demographic groups—with virtually no one knowing they are being manipulated, according to experiments I conducted recently with Ronald E. Robertson.
You knew about this research already, of course. I’ve brought it up before.
Funny thing is that the first four or five times I saw a reference to him, it didn’t occur to me to look him up. After seeing this reported again today, I finally decided to look into it more carefully.
Here’s the full paper on PNAS: The search engine manipulation effect (SEME) and its possible impact on the outcomes of elections. His co-author is Ronald E. Robertson; if you throw his name in Google, you’ll find his LinkedIn page first, and you probably won’t do any further searching. (I basically believe everything Robertson says about himself there, too, although I’ve confirmed none of it.) My willingness to believe what I find in the top-ranked search result confirms what they suggest in their paper:
Studies using eye-tracking technology have shown that people generally scan search engine results in the order in which the results appear and then fixate on the results that rank highest, even when lower-ranked results are more relevant to their search. Higher-ranked links also draw more clicks, and consequently people spend more time on Web pages associated with higher-ranked search results. A recent analysis of ∼300 million clicks on one search engine found that 91.5% of those clicks were on the first page of search results, with 32.5% on the first result and 17.6% on the second. The study also reported that the bottom item on the first page of results drew 140% more clicks than the first item on the second page. These phenomena occur apparently because people trust search engine companies to assign higher ranks to the results best suited to their needs, even though users generally have no idea how results get ranked.
That more or less describes what I did when I decided that I’d like to know more about their research, so no conflict with intuition there.
I’ve now read the whole paper. I do think they’re raising a serious concern. That said, there’s a limit to the amount of research I can do quickly. You see, when I search for more information about Robert Epstein, for example, I hit a wall very quickly:
Some results may have been removed under data protection law in Europe. Learn more
Just curious: What happens when those of you who aren’t in Europe search under these researchers’ names? Do you learn the same things I do? Do you draw similar conclusions?
Published in General
Parentheses can certainly be useful. (If they could not, why would we have them?)
But they can create a problem when they are “echoed”, just as words can. (I didn’t realize this until I read the beginning of this post.)
I’m going to read the rest of the post now. (I wouldn’t want to be one of those people who are obsessed with form, to the exclusion of substance.)
That’s all. (I don’t want to belabor the point.)
Your loyal reader (and faithful reader),
M.
Just for fun, I went to Google News and Bing News, and searched both for ‘Ted Cruz’.
The top article in Bing is a positive article about Ted Cruz from NewsMax.
Next, an analysis of how big the evangelical vote might be for Cruz, published in the Washington Post.
After that, “Ted Cruz Remains a Force in Crowded GOP Field” – USA Today
“Ted Cruz to lead 100,000 pastors in a 50-state attack on Planned Parenthood” – The Raw Story.
“Ted Cruz and Ellen Page spar over gay rights” – The Inquisitr
“Ted Cruz seeks to lock down conservatives in South Carolina” – Politico
Overall, I’d say the results are fairly non-partisan, and the large sites you’d expect to be at the top of the rankings are represented.
Over at Google, the top story for “Ted Cruz”:
“In Bizarre Moment, Ted Cruz Says Gays Should Be Able To Discriminate Against Christians Too” – ThinkProgress
After that are the same politico and Washington Post articles Bing hit.
Next, “Ted Cruz Talks to Pro-Slavery Radio Host About “Anchor Babies” – Dallas Observer
“Tuscaloosa preparing to host Ted Cruz in sold out event” – WRBC
“Ted Cruz enlists pastors for all-out assault on Planned Parenthood” – MSNBC
Can we learn anything from this? Probably not. But in this case, it’s pretty clear that the two search engines give results with a very different slant. A ‘low information voter’s opinion of Cruz would possibly be quite different if he chose Bing over Google. In this case Google’s results are slanted more to the left than Bing’s. Whether this would hold for other candidates or over time, I have no idea. This could just be a fluke.
Okay, I just tried ‘Scott Walker’ for both engines.
Bing’s top story is “How Scott Walker’s Bargain Hunting at Kohl’s Informs His Economic Policy” – from ABC News
Google’s top story: “Scott Walker’s nonsensical China statement reveals one of his biggest weaknesses” – from Vox.
Again, doesn’t necessarily mean a thing, but does show how a difference in search algorithms can affect real-world perceptions of politicians.
Right. But if they could do that consistently, SEO would be impossible, for exactly the same reason that if poker were gambling, you wouldn’t have the same people playing in the World Series of Poker year after year.
I know it is.
That’s what “Bayesian machine learning” means.
Because Bayes requires you to provide prior probabilities, and the better those are, the better your results will be, the more ad impressions you’ll sell, etc. The entire 2nd half of Probability Theory: The Logic of Science is devoted to calculating good priors. Likewise chapters 5 on of Data Analysis: A Bayesian Tutorial. Chapter 7, “Experimental Design,” is the sort of thing that’s especially important to Google, but they could write their own book, and basically have.
So back to my question: are Bing, DuckDuckGo, etc. better, worse, different, about the same…?
Not necessarily. It could be that SEO manipulations are not affected by this kind of bias – especially SEO in areas where the bias doesn’t exist. Or it could be that the effect exists but is subtle enough that it doesn’t totally swamp SEO manipulation, so SEO still works.
Didn’t you contradict yourself here? You first said that the human input is completely irrelevant over time, but when I asked why Google does it, you said that good priors lead to better searches. I don’t think you can have both – either the human input has no effect on the search, or it does.
I don’t think anything is as good as Google – right now. That doesn’t mean I want the competition to go away. I’d just rather see them get better.
But then the bias doesn’t matter, statistically speaking—it could be (and I think part of our hope is “is”) countered by oppo SEO. But then, we are talking about the Republicans…
Sure, in the short term. No contradiction. Bayesian probability measures information you have. If you start with a lot, Bayes’ theorem won’t budge much. If you don’t, it can. You can summarize Jaynes as “use all the information you have, all the time.”
You seem to be ignoring “over time.”
But what does “get better” mean? The way search engines get better is by biasing their starting data!
I understand its a tough target to hit. A good search result is like pornography – you know it when you see it. (-:
Basically, my point is that with competition, the pressure is on the search provider to provide what people are looking for, and not game the results for some purpose other than customer satisfaction. If Google becomes known as the ‘liberal search engine’, it will lose a lot of traffic to Bing. On the other hand, if Bing or other competitors don’t exist, no one would ever know that Google was gaming the results.
haha, and I always “don’t click” on anything with the ad link, but go underneath to search results.
even if what I am doing is shopping.
not sure if it matters, but if I’m searching for information, I don’t want an advertisement.
Dan & GGofG,
Although I tend to see things the way GGofG sees things, I think there is a legitimate question of perception. However, anecdotal evidence followed by paranoid presumptions about Google won’t help.
To make clear what is going on we need to bring in a ringer. Amazingly, we don’t even need to bring him in. Dr. Tim Groseclose is our in house ringer for this job. He has a peer reviewed proven method for monitoring political bias. Tim needs to turn his modeling mind to internet searches. Only he can give us the empirical data that would verify a legitimate case against Google. Demanding that Google turn over its search algorithms is a fools game. This can only lead to bureaucratic meddling in the internet. We already have the immense danger of Obamanet. We don’t need yet another reason to destroy the greatest free innovation perhaps of all time.
Let’s help Tim get some serious grant money and go at this thing full tilt. That’s the only way we can get Dan to sleep easy about the bias and the rest of us to sleep easy about the idiots at the FCC.
Regards,
Jim
the best way I’ve found to keep my sanity is to
thinkdrink.just kidding, but sometimes my eyes see things when I read that aren’t there.
Jules,
If on the other hand you were searching for a low price from a quality company with a big selection sometimes the paid advertising represents people who know what they are doing and will save you time and aggravation.
Regards,
Jim
To be clear, I’m not claiming that Google is biased, and certainly my little game above is not proof of anything. I’m just pointing out that we are putting faith in a company that has the power to swing elections and otherwise subtly and non-transparently manipulate the public. Google may be the most powerful company that has ever existed because of that. When you’re the filter through which the world finds its information, you’re controlling a pretty commanding height.
That seems like a risk worth considering, regardless if anything is actually going on right now or not.
At a more philosophical level, what does this mean for freedom of the will?
Dan,
All the more reason not to do this through anecdote. I’m no lover of statistical social science but this is a situation that cries out for it. It’s all about the numbers. I don’t like Al Gore anywhere near anything that might be affected by bias but what good does it do to scare people unnecessarily. We need some hard evidence. Tim could go after Google searches for environmental information. He would work through how to discount all of the factors and work them into a usable model of bias. If he generated a report that showed major bias in search results that would be major news and then Google must explain why. We get a chance to critique Google’s answer and so it would go. If Google were found to be biased, even over a single issue, it wouldn’t bode well for their bottom line. For the moment Google has total market domination. If they were exposed as biased it would be the perfect reason for people to give Bing a try. I don’t think Google wants this so the thought that somebody out there was watching them in a way that could be authenticated would seriously make them consider anything that would wind up as political bias.
Where can we score some serious grant money for Tim quick?
Regards,
Jim
It is still interesting to note that Google obviously could influence people, and that they do also skew left (ever look at the ever changing Google logo?). There is clearly a potential for harm, whether any of us think that is likely or not.
I just know when I go to Bing and type in “wattsup” looking for Anthony Watts’ great website, it is the top of the selections offered. When I do the same on Google it is nowhere to be found. Is this an accident? Not likely.
That IS interesting, because according to Google’s own documentation pages on how their search works, if a search term is a significant part of the name of a web site it’s a ‘vital’ search and that web site should be at the top of the rankings. For example, if you type ‘yout’, the first link is YouTube. If you type “Microso”, you get microsoft.com. The only exception should be if the string better matches a different site.
However… after typing that I tried it myself, and on Google the first site that comes up for me is ‘WattsUpWithThat.com’, which is as it should be. Could it be that you mistyped something? If not, is there perhaps a regional site close to you that has a closer name? What was the first site that came up for you?
If it’s always coming up different for you, it would be interesting to figure out why.
I’ve observed more than once the joke’s on us: alcohol and caffeine are both diuretics!
I got the same result as Dan H, but the Google does (or so I’ve read) tailor its results based on your previous searches.
Google tailors search based on your previous searches, your geographic location, and a number of other factors. I imagine it probably looks at your browsing history to see what kinds of things you are interested in, perhaps even things like the time of day.
That makes sense from a search efficiency standpoint. I’m glad Google knows that if I type in ‘Microso’ I’m probably more interested in Microsoft.com than ‘Microsolo’, the web site for piccolo players. And if I were a piccolo player, I’d be annoyed by having to look at Microsoft.com every time I have a burning need to discuss my tiny flute.
However, I’m not sure Google (or anyone) is capable of knowing how such personalization will affect society at large. For example, if my personalized search profile means that I will be directed towards right-wing web sites when I search political issues, and my leftie counterpart will be directed to left-wing web sites, how does that affect political polarization in the country?
When one company acts as arbiter and filter of all our information, it has inordinate power. It becomes a meganode, a choke point in our complex society. I’m not sure that’s a good thing.
Then there’s the risk aspect. Nassim Taleb talks a lot about managing a complex world by adopting systems and institutions that are ‘anti-fragile’. For example, the internet itself, which is designed to route around damage automatically. A transportation system consisting of a large mesh of roads is anti-fragile, while a transportation system consisting of a hub of rail lines is not.
Google brings fragility to the internet in the sense that it’s a point of failure which can be systemically disruptive if it goes away or turns out to have some sort of pernicious influence.
Dan,
I am not pre-judging whether Google is biasing searches or not. However, I think your evaluation of Google is prejudicial. Google is the least fragile of any of the search engines and their approach was the most neutral. They pioneered search and are a huge multiplier of the internet. The value that they have added to the net by their efforts is immense. However, whining about their capacity to track data doesn’t improve anybody’s privacy or nail down bias. The only way to check bias is to “stress test” their system regularly. For every way to track you there is a way to defeat the trackers.
We are engaging in a mentality that will not make things better.
Regards,
Jim
I’m not talking about fragility in the sense that Google itself is fragile. It could be the bestest, most powerful, most secure search engine mankind can devise. ‘Anti-fragile’ in this case means organizing things in a way that a single point of failure can not cause massive disruption to society. For example, if an asteroid hit Google’s main server farm, or a hostile power decided to wipe out Google with missiles.
I use those examples not because I think they are likely, but to say that this isn’t a fault of Google’s at all. I’m not criticizing Google in any way. I’m saying that there is risk involved when you have critical choke points or hubs in a complex system.
Dan,
Given what we’ve seen of the gross incompetence of the government in information technology and Google’s mastery, I’d say that in all likelihood it’s Uncle Sam that is more exposed than Google. Google plan’s ahead to protect it’s business I’m sure. Your point still is well taken and one more reason we need desperately to be rid of Barack the Magic Dragon and his JV team.
Google like all powers needs a check and balance (where have I heard that before). I think independent evaluation (Tim Groseclose) could produce the check and just feeling free to try another search engine would be quite a balance. Even if Tim discovered for instance that Google does lean left and Bing leans right that’s a problem for Google. They have 90+% of the market. If it really is true that Google tilts left and Bing tiltes right then I suspect that Bing would come up to at least 25% of the market very shortly thereafter. Google doesn’t want that big dragon from Redmond let loose.
Checks and Balances.
Regards,
Jim