One night, Kevin Jamieson sat at a bar and drank a beer that was particularly delicious. Maybe he even had a few or a few too many, because the next day, he couldn’t remember the beer’s name. But Jamieson was so smitten that he returned to the bar later that week just to track it down.
There was just one problem: The bar had 40 beers on tap.
Jamieson described the beer to the bartender as “pretty malty and hoppy”–which is vague at best–but the bartender was undeterred, and began sliding him samples, probing for similarities. Some tasted close. Others were way off. Yet within six samples, the bartender had narrowed down the elusive beer.
“Consider how remarkable it is that he found one beer among 40 after having me try just six,” Jamieson tells me today. “The only way this could have worked is if each question removed many possibilities for the beer I was looking for, and that is only possible if there was some simple, underlying structure to the beers.”
And with that realization, Jamieson embarked upon a new quest: To calculate and visualize the unseen relationships between brews, and to build that into an app: Beer Mapper, for iPad.
To understand the depth of the infographics in this post, you have to start by swallowing a somewhat odd concept: Jamieson has developed analytical algorithms that actually reason through space. Just like you might place a point on a 2-D or 3-D grid, Jamieson actually arranges traits of a beer–from very specific text descriptions he’s scraped from reviews, like “grapefruity” or “piney,” to very hard, scientific measurements of maltiness and bitterness–into six simultaneous dimensions. From the proximity of all these diverse data points, Jamieson can reason complex relationships of the beers based upon a multitude of interrelated criteria at once.
In other words, the algorithm is a spatial simulation. It’s like an infographic encapsulating a universe of 200,000 beers that only the computer gets to see. Cool, right?
There’s just one catch: Since the human mind isn’t so great at visualizing 6-D, Jamieson shows us the data in 2-D by limiting the user to two traits at a time to compare. (In the final app, you’ll be able to cycle through the six dimensions of relationships by toggling all of the various 2-D layers.)
Ultimately, the end user is left with a pretty handy tool. Not only can we understand, maybe for the first time, how porters relate to stouts, and how stouts relate to pilsners, but we can also grasp how individual beers within this cloud relate to one another by pure proximity. So you can see that, if you like Dogfish Head Midas Touch Golden Elixir, you may also like Killians Irish Red. Jamieson is capturing both the macro relationships of beer styles with the micro relationships of individual brewed, labeled beers.
Jamieson’s iPad app is due out in late June, and while it’s still unclear exactly how often it will recalculate its gigabytes of data into a meaningful visualization (or whether much of this will occur on the iPad or the cloud), it seems destined to become one of the most comprehensive, up-to-date analyses of beer, ever. And for those of us who’ve spent too many hours digging through message boards full of snobby beer nerds just to find a rare brew we’d enjoy, Jamieson seems to have kindly automated the process into a clever iPad app that’s not so different from the friendly, patient bartender who inspired it all.
Though, if I may be so forward–Jamieson, Jamieson, Jamieson–we need to find you a graphic designer. If only we knew a few of those types…