If you like this, then you should try that...

If you look at the beer profile page for Lagunitas Sucks you will see that it is listed as a "American Double / Imperial IPA" style. If you click that style link you will be taken to a list of the beers in that category, and the default sort order tends to be the list of beers in that category which are considered the best. It sounds like the feature you are asking for already exists...

If there was something specific that you like about Sucks then you should describe it and ask for other ideas with that same flavor profile, but with just a generic question I think you are only going to get a list of people's favorite DIPAs (which is already available).

 

It sounds like an unfathomable amount of work for the BA staff to create a tool like this. I'd suggest studying up on different hop/malt profiles and get your hands dirty with research.

 

Untappd has a feature like that and it is completely ridiculous. You're better off justing asking a question here for a specific beer or searching for past threads.

 

This may have been suggested, but something I've really wanted to see is a recommendation system, such as is employed by Amazon/Pandora/Netflix/etc. Given the number of ratings and users (but depending on how those data are stored), it should be easy to implement some collaborative filtering or something similar, which would also encourage people to do more rating (since more ratings=better recommendations). Presumably you could set this up to be optional, so this service wouldn't affect people who wouldn't want to use it, but would be there for those who do.

 

A recommendation system beyond what is already implemented would be nice, but those examples you listed would suggest things that are available nationwide (and internationally as well). The availability of beer is much more subjective considering where exactly it is and isn't available; which can make for some headaches if you are trying to provide a precise suggestion that someone could actually find.

 

Depending on what kind of databasing system they're using (that is, how the ratings/beers/users are stored), it might not be that much work. I've never built a recommender system before, but it's something I'd like to try (mods?)... As for whether it would work, that remains to be seen, especially because people can be so idiosyncratic in their preferences, but there is structure to a lot of beers, and for users without the wherewithal to delve deeper into yeast/malt/hop/schedule/etc variables (which is still probably an oversimplification, since I bet I could find two beers with completely dissimilar profiles that were more alike than two beers with almost identical profiles, just due to differences in brewer skill, etc), it could be a nice tool.

Not sure what Untappd does (I don't use the site, and almost never visit it), so I can't say whether what I'm proposing would be any better. I agree that asking around can be helpful, but taking a quantitative angle where users are organized according to the similarity of their ratings and then mined for relationships could actually be just as useful, if not moreso, just because you wouldn't have to try to sift through advice from fifteen people who love/hate/haven't tried Sucks/90 Minutes/Dirtwolf. Your recommendations would be tailored to you. And any decent recommender system would let you "wipe the slate" in some sense if, four months down the line, you burn out on Belgians and go nuts for ABAs.

 

That's a little trickier, but if you include users' location in the system, you could weight it such that recommendations from users geographically nearby are preferred, even if their overall similarity in tastes is lower. These are all knobs that could be tunable, as well.

 

Human Beernome Project is what you're looking for, make it happen. Hops used, malts used, yeasts used, water used (pH, region, whatever) and adjuncts used (type of barrel included) and weighted according to the beer style. So a DIPA would place most of it's emphasis on which hops were used, and in what proportion, then similarity of yeast used and malts used, etc when recommending a similar beer.
Humanity will be in the right place when we devote enough resources to this type of thing. We could eliminate, say, water polo and instead teach people boozy science. Just teach everyone science, the closer to Germany, the boozier (the good kind, more litre mugs).

 

That's pretty much the Pandora approach. Plenty of upsides, but the downsides are that it's labor-intensive, and it can actually miss things because it's basically impossible to human-define enough variables. In beer terms, even if you know the hop bill, you need to know when each variety was added, in what quantity, for how long, etc. Using collaborative filtering avoids both of those problems--it's automatic (once you've coded it), and it doesn't care about any of those variables per se. It just says, "People who like Sucks like Fortunate Islands. You like Sucks? Try Fortunate Islands!" Doesn't care that one's a 7.85% DIPA and the other's a 4.8% American Pale Wheat, with no overlapping hops. Actually, it does better than that--it finds people who like Sucks and also like and dislike as much of the same stuff as you as possible, and then figures out what else they like that you haven't tried.