Apr 06 2015

The Google University Effect

I remain endlessly fascinated with the incredible social experiment we have all been living through over the last decade (and I can say, if you are reading this, you are part of the experiment). The internet and social media have changed the way we access information and communicate. The traditional top-down systems of information and opinion dispersion are eroding, being replaced by a largely bottom-up free-for-all.

I think we’re still figuring out all the consequences of these changes, both intended and unintended. One effect that has been casually observed is that many people believe they have expertise they do not have because they have been able to do “research” online. The democratization of information has led to a false sense of democratization of expertise.

While free access to information is great, there is no systematic way in which the public is taught how to use this information to maximal benefit, and avoid the most common pitfalls. Schools are generally behind the curve in terms of teaching students how to manage their online information access. Most adults were done with their formal education before the wave of social media.

The result is the “Jenny McCarthy Effect.” She is a celebrity who feels that she can substitute her own non-expert opinion for the strong consensus of expert opinion on the safety and effectiveness of vaccines because she “did her own research.” She is an obvious example of how searching for information online can give someone a false confidence in an unscientific opinion, illustrating the fact that relying on “Google University” can be extremely misleading. There are some specific pitfalls at work here.

The first pitfall is the subject of a recently published series of experiments by Matthew Fisher, a doctoral student in cognitive psychology at Yale University. He looked specifically at the effect of searching online for information and confidence in one’s knowledge on that topic. Of course it makes sense that if we search for and read information on a topic this will increase our confidence in our knowledge about that topic. Fisher, however, tried to control for as many variables as he could to see if there was an independent effect of just searching, regardless of how it affected our actual knowledge.

He found that people had higher confidence in their knowledge even when they searched for a subject vs being taken there directly, when the topic had no relevant information online, when their searches were filtered for relevant information, and when they read the information online vs in print. So even when the actual information was controlled for, the act of searching online itself seemed to raise confidence in one’s knowledge.

These types of experiments are obviously complex and we’ll need to see this replicated from different angles, but so far it does seem that having access to a vast store of knowledge that one can sift through raises our assessment of our own knowledge (beyond access to relevant information itself).

It seems to me that there are likely other effects at work as well – chief among them is confirmation bias on steroids. Searching online gives us the opportunity to mine a vast amount of information and select (even if unconsciously) that information which confirms what we already believe or want to believe. Search online for information about vaccines and you can find plenty of information that supports the safety and effectiveness of vaccines, and plenty of information that vilifies vaccines. Choose any controversial topic and the results will be the same.

Confirmation bias is powerful and dangerous specifically because it creates the illusion that the data supports our beliefs, because we are unaware of the degree to which we have filtered and biased that information. The internet is a setup for confirmation bias.

The extreme version of this phenomenon is what we call “echochambers.” Filtering information can be formalized into online communities where only one perspective is expressed, and information that supports that perspective is shared, while opposing information is filtered out or directly contradicted. This is a pervasive effect, and is true of scientific and skeptical sites as well as pseudoscientific ones.

Another potential problem is the confusion of knowledge with expertise. This is often what leads to cranks – people who may be very smart and have a great deal of factual knowledge, but come to absurd conclusions in which they have high confidence. One problem with cranks is that they do not properly engage with the relevant intellectual community.

It is critically important to engage with the community, especially in highly complex and technical areas of knowledge. It can be very difficult for any individual to see a complex issue from every angle, and to consider all perspectives. Left alone we will tend to create a neat narrative, and become increasingly convinced in the truth of that narrative. Engaging with the community will tend to challenge that narrative, leading to a deeper and more nuanced understanding of the topic. This is the heart of true expertise.

Studying a subject alone by searching online can be a crank factory – giving factual knowledge without really engaging with the ideas. Then the echochamber effect can give the illusion of engaging, but only with a biased community rather than the broader community. The result are people who falsely believe they have sufficient knowledge in areas they do not truly understand. The Dunning Kruger effect kicks in as well, and they likely do not appreciate the gulf between their Google University understanding of a topic and the depth of understanding of true experts.

Conclusion

The internet may be creating an army of overconfident pseudoexperts. There are a number of fixes to this problem on the individual level:

– Be humble. Do not think a little knowledge makes you an expert. Respect the opinions of actual experts. (You don’t have to agree, but at least take them very seriously.)

– Understand the inherent advantage of a consensus of expert opinion over the opinions of any individual.

– When searching online, go out of your way to search for information which goes against your current belief or conclusion. Try to find what both or all sides are saying, and reserve your personal judgement until you think you have heard all sides.

– Understand how online searching is a setup for confirmation bias. Google itself can bias the results of your search. You can turn this feature off.

– Understand that, in addition to confirmation bias, there is organized bias on the internet – echochambers, astroturf campaigns, and deliberately biased ideological information. Be on the lookout for false information, and carefully vet a source before you rely upon it.

– As always, there is no substitute for skepticism and critical thinking.

20 responses so far