Media Center

The CBW Archives

How easily, part two: the expert opinion

This is a followup of sorts to “How easily our ship is steered,” which is funny because I’ve been meaning to write about the article in question since September. While the initial post was more science-y, however, this one’s going to have a bit of th’ ol’ soapbox in it, so I thought it best to split it into its own post.

Let’s talk about McAuley J & Leskovec J  (2013), “From Amateurs to Connoisseurs: Modeling the Evolution of User Expertise through Online Reviews.” They take a different angle at the issue: instead of only considering what we like, it discusses what we like and how that might be subject to change, how a person “grows” over time as they experience more things.

This is a great example of how science reporting is often pretty terrible. I came across this study when Ethan emailed me a Business Insider article on it:

They developed a model to help them separate the “expert” users from the “beginners” on a review website. Briefly, it takes into account the number of reviews a user has written, and how the user’s ratings have evolved compared to the rest of the “expert” community (those who have written more than 50 reviews).

I was fairly annoyed! Ethan and I have solidly entered the realm of post-geek, wherein we think people should drink what they damn well please (uh, I mean, drink Frank!), and here this article is talking about how someone is an expert just because they reviewed a lot of beer.

Friend of the show Matt Kahn, Resident Mad Scientist and impending brewer, read the full article and told me it wasn’t quite so bad. Indeed, what I initially thought was “these people are experts they are so much better than you” is actually only using “expert” as a shorthand:

Starting with a simple definition, experience is some quality that users gain over time, as they consume, rate, and review additional products. The underlying hypothesis that we aim to model is that users with similar levels of experience will rate products in similar ways, even if their ratings are temporally far apart.

So it’s not that an “experienced” drinker is more adept at rating, but that as people gain “experience” their tastes follow a similar path. Okay, that hypothesis I can dig. Fury abated; let’s continue.

Their findings are fairly conclusive: there’s an image in the article and the Business Insider post that I won’t reproduce here so as to not violate copyright, but essentially they classified beer into three types: lagers, mild ales, strong ales (“mild” here not being actual milds but less hoppy styles). Not only is there a clear hierarchy — strong ales are better than mild ales are better than lagers — but as a user gained “experience” they liked lagers less and liked strong ales more. Their review scores became spaghettified.

Not only that, but they claim to be able to predict scores, using a model that “significantly outperform[s] traditional recommender systems.” There’s math involved, of the sort that I admit is beyond me (I have two degrees in liberal arts), so you’ll have to see for yourself if that pans out. I do like that it accounts for a user changing over time. Call me “experienced” or just “different,” but I’m not the same person as I was five years ago. For one thing, I’m significantly less into Coheed & Cambria.

In section 5 they go back to saying things with which I significantly disagree. After a series of equations that, again in fairness, were completely unintelligible to me, they go back to the initial data of lagers < mild ales < strong ales, saying:

Of course, it is not surprising (to a beer drinker) that experts dislike lagers while preferring India Pale Ales.

Excuse me? I think any definition of “expert” as a person who dismisses that many styles out of hand is incredibly suspect. Indeed, I think the ability to appreciate a good Dortmunder Export or Classic American Pilsner is one of the marks of a true expert, inasmuch as I’m willing to admit such a thing exists. After a few years of homebrewing and aggressively immersing myself in beer culture I stewarded the “Light lager” category at the Amber Waves of Grain homebrewing competition. The judges were Terry Felton and Becky Dyster, two people who have been beer experts since I was in short pants. They picked out flavors and complexities in the beer that I simply could not. To me (a novice) they all tasted more or less the same, but to them there were clear differences. Turning up your nose at lagers does not make you cool and it does not make you an authority on beer.

So, then, I have no problem with their analysis of the results. Instead, let’s go back up the chain and burn down the whole damn system. I disagree that being in a community for a long time and writing many reviews makes you an expert. In fact, that the more you rate the more you agree with other raters to me says less about experience and more about a tendency to fall into step with the crowd. Maybe it’s their use of “expert” instead of something like “long time member.”

Saying that “strong ales” (which includes IPAs) are categorically better than “mild ales” is false. It’s just wrong. Strong ales are not better than mild ones; they are merely higher rated. Don’t confuse the shadows on the wall for the animal that made them.

Soon after reading this article in September I came across an article about music but which applies to beer and, really, everything.

There’s only one thing worse in this whole world than a misguided nerd, and that’s an apologetic one. You like what you like. You should always be looking to expand your palate but, at the end of the day, you can’t force it.

I think that’s my real underlying issue with this concept: that some opinions are better than others. Call me a socialist, I guess, but when people tell me they don’t like IPAs? That’s fine! You don’t have to like everything. Recently I’ve even seen a trend of people liking hoppy beers but not malty ones, throwing the entire idea of a “progression” right out the window. McAuley and Leskovec’s data is solid, but I question their premise. Experience does not equal expertise. We are not moving inexorably toward some teleological conclusion. And, besides: you’re a grown-ass person. Your opinion is valid.

I’m confident everything you like sucks. I know it. But there is not a reason in the world you should care about my opinion. There’s no reason you should value ANY person’s opinion over what your ears tell you. Never change for the guy at the record store, the geek in a popular band, or some faceless blog. Don’t bother defending your position. Just like what you like.