February 25, 2010

Why online comments are negative and online product reviews are positive

OK, there's been enough time for people to have thought about the question I posed a few days ago. To reiterate the paradox: anonymous online comments left at blogs, discussion forums, and newspaper sites tend to be much more negative than real-life comments; while anonymous online product reviews left at Amazon, YouTube, Ebay, etc., are much more positive. We only notice the former -- everyone's an asshole on the internet -- but the pervasiveness of the latter means we're missing something.

And when I say "positive" or "negative," I don't just mean the judgment the person is passing but also their tone of voice: "You absolutely MUST drop everything you're doing now, and GO SEE THIS MOVIE!!!" vs. "omg gayest idea evar."

First, some replies on why the answers given so far aren't it.

In self-interested competition people accent differences in comments: negative to the extreme of playing the devil's advocate (taking a position they might not agree with for the sake of argument). Sympathy is positive and people will cooperate to enhance a liked product.

It's not just devil's advocates, though -- we're talking about people calling each other morons, using moralistic language that we normally reserve for child molesters, etc. It's not unemotionally giving the other side a fair hearing. As for cooperating to enhance a liked product, people also do the opposite -- they gang up and make fun of a common disliked thing to feel positive sympathy among themselves. And yet we don't see product haters piling on in product reviews, in the way we'd see a group of jocks surround and taunt an emo kid. We also don't see product reviewers playing devil's advocate by downgrading a product they've bought but whose reviews so far are overwhelmingly 5-star ratings.

People who leave comments on products have to admit that they bought it. People who bought a poor product would rather not admit that they made a bad decision, while those who made a "good decision" (i.e., they love the product) are more likely to review it.
On blogs, since there is no such product, it may be a simple case of not liking the idea via Not Invented Here syndrome.

But on blogs, discussion forums, and newspaper sites, you are definitely consuming a product and spending your time on that -- just not necessarily paying money out-of-pocket for it. Hence the thousands of resentful comments left on such websites that say, "I want my five minutes back, bitch!" Negative commenters are admitting that they've been wasting a bunch of time, the scarcest resource, on reading idiotic ideas and arguing with clueless morons.

You can try to fix this by saying that they're happy to waste their time and admit they consumed a bad product, as long as they get a later boost in happiness from signaling how great they are by really tearing into the stupid ideas, calling others morons, etc. But again, why don't these people signal their superiority or just enjoy some plain old sadistic fun by tearing into products in online reviews?

Now for the answer. I'm going to call the commenters or reviewers "the speaker," whatever it is they're commenting on or reviewing "the target," their comment or review "the remark," and those who the speaker expects to read the remark "the audience." What separates the negative from the positive cases is that in the negative case, the speaker expects the audience to have also experienced the target -- they too have read the blog entry, the post in a forum, the newspaper article, etc. -- while in the positive case the speaker expects the audience not to have also experienced the target -- they have not (yet) read the book, watched the YouTube video, bought from the Ebay seller, eaten at the restaurant, etc.

Indeed, that's the whole point of a comments section -- to bring together an audience who have all experienced the target. And the whole point of a product review is to bring together speakers who have experienced the target and a much larger audience who have not and who are coming to the review to read the remarks from those who have.

Why does this difference matter? In the Theory of Moral Sentiments, Adam Smith elaborates on how profoundly people are motivated by a desire to sympathize with others. Here, "sympathy" does not mean pity, compassion, benevolence, etc., but merely feeling what another person feels, being on the same wavelength, and so on. When the whole world smiles with you, or when your neighbors commiserate with you when a tornado tears through your house, these are examples of sympathy.

Crucially, our moral approval of someone's actions that affect another person depends on our sympathy with the actor's motives and sympathy with what an impartial spectator would feel about the effects on the person affected. So if an actor benefits someone else purely by accident, we don't give him moral approval. And if the actor has good motives but an impartial spectator would say the effects are actually harming the other person, we certainly don't give the actor moral approval. But even if we approve of the actor's motives and their effects on the other person, we may still disapprove if the actor did not observe propriety -- for example, if someone with just cause punishes someone but does so to an improperly extreme level, we disapprove. And conversely, if the actor with just cause punishes someone to an improperly low level -- a slap on the wrist when more is required -- we disapprove.

In the real world, it is often difficult to know what the actor's motives are, especially if we have "just walked in on" their action. We see someone chewing out another when we walk into Starbucks -- are they crazy or justly motivated? We can't tell. And without knowing the back-story, we also cannot tell how an impartial spectator would feel about the action's effect on the other person. Nor can we tell whether propriety is being observed -- even if the other person deserves to be scolded, is the actor being overly harsh -- or not harsh enough? Again we don't know since we're ignorant of the back-story.

In cases of punishment, though, despite our ignorance we give the benefit of the doubt to the person being punished. So when we walk into a bar and see someone lunging out to punch another person, we actually feel more for the person about to get hit! If you're a male, you may even rush in to restrain the hitter. Even though the hitter may have worthy motives, the hittee may deserve what he's getting, and the degree of punishment may be just right, we don't know the back-story and so side with the hittee. Perhaps if we'd seen the entire interaction unfold, we'd approve of the hitter's action and even cheer him on -- or join in ourselves. Perhaps he just threatened everyone with a gun, someone distracted him, and the hitter made a dash to punch him out while he wasn't paying attention.

Now you see how this applies to blog comments vs. product reviews. In comments, the speaker expects the audience to have experienced the target, so they know the full back-story. If the blog entry's idea really was stupid and deserved to be panned, the speaker can lash out knowing that the audience will approve of the remarks. This just says that it is possible, not inevitable, that speakers in comments sections will be negative. I think the status-jockeying arguments apply at this level -- making the possible negativity a realized negativity.

Why aren't there positive speakers to counterbalance the negative ones? After all, any part of the audience who liked the blog entry's ideas would sympathize with such speakers. I think people who generally enjoy the blog entries, newspaper articles, etc., just consume them and stay out of the comments because they know they'll get dragged into a nasty war with the haters. They may disapprove of the negative comments, but unless it's really bad, they aren't going to join in to defend the blog post, newspaper article, etc. That leaves mostly negative speakers to write remarks.

On the other hand, the audience of a product review is entirely ignorant of the back-story for a speaker's remarks -- the audience didn't sit over the speaker's shoulder while they read the book, didn't sit by their side as they watched the YouTube video, didn't chat over the phone with them as they tried out their new espresso machine, etc. In fact, the audience hasn't experienced the target in any way whatsoever, so they're ignorant of how any fair-minded person would review the product. Therefore, if a reviewer tears into the product, particularly in the curt way that newspaper article commenters do, the audience will go into "side with the punished" mode and disapprove of the speaker. No sympathy will form between speaker and audience. So, any speaker inclined to demolish the target will instead bite his lip and move on. That only leaves those inclined to praise the target to leave their remarks. Unless there are no such people (in which case there would be no reviews at all), the target will get overwhelmingly positive remarks.

This explanation does not capture everything, but it does explain the hardest part, which is why it's possible for the blog comments to be so negative (not surprising) but impossible for product reviews to be so negative, causing them to be rather positive (very surprising). It's remarkable when you think about it: entire armies of haters shut themselves up in product review places because they expect to get little sympathy from the audience, and to have the mob stare at them in disapproval. They are never going to interact with anyone from the audience face-to-face, and probably even not even impersonally through market exchange (except at 20 degrees removed). They don't even have a reputation to maintain since they aren't planning to hang around in the product review place -- just to leave an anonymous poisoned review and then leave.

Of course some people still do leave negative product reviews, but that will only happen when the speaker values meting out punishment more than he values not being disapproved of by a group of onlookers. That could either be because the product is so terrible that it needs to be said -- if a laptop overheated and burned a hole through your desk, say -- or because the speaker just doesn't care that much whether or not the audience approves of his actions. He thinks he's above them.

And this explanation also captures why some book reviews are withering critiques -- the speaker of these remarks expects a fair amount of the well-read audience to have already read at least pieces of the book. It's some hot new book whose hype has led a lot of literary types to browse through it. Now the reviewer is safer to launch an attack. But most book reviews on Amazon are written for an audience that is just about entirely ignorant of the book's contents. They just heard about it and want to see what's up. Those reading the book review sections of high-status newspapers and magazines, on the other hand...

I can't recommend Theory of Moral Sentiments highly enough, and I'm not just saying that to stay on my audience's good side. It's one of the few social science books over 100 or so years old that's still worth reading. It's a mix of social psychology and moral philosophy. EconTalk ran a six-part podcast series on the book, and I'd encourage everyone to give it a listen. The book is not long but very dense with ideas, so it'll take awhile to absorb it all. I did the podcasts first in my down time and then read the book. I found that the guide helped me to grok the ideas.

It really is one of those books that changes your entire perception of how the human world works, and in all of the modern social science I've ever read (starting with William James or so), I haven't read anything that has mentioned or independently rediscovered Smith's insights. (Certainly some have mis-attributed various ideas to him.) It's not like Wealth of Nations, where most modern econ textbooks will mention the benefits of specialization, division of labor, ending mercantilism, and so on. All the social sciences have a depressingly crude picture of what motivates human beings, and why they interact socially the way they do. If you invest the time in Theory of Moral Sentiments, you'll come away with a far richer view.


  1. Thanks for the interesting theory and the book reccomendation. One question, though: Why do you lump YouTube together with Amazon and Ebay rather than with blogs and discussion forums? It seems to me that with a YT video, as with a blog post, most people experience the product first and then read the reviews, many of which tend to be extremely negative.

  2. I should have clarified: the YouTube thing I'm referring to is the star rating, not the verbal comments. (The WSJ article linked to in my first post said YouTube star ratings are very high.)

    You're right that the comments tend to be negative for the reasons you say, while the star ratings are there to influence people who haven't seen it yet.


You MUST enter a nickname with the "Name/URL" option if you're not signed in. We can't follow who is saying what if everyone is "Anonymous."