I poured a brewer friend of mine something I was stoked about, a beer built on Nelson Sauvin hops. I was over the moon with that diesel funk Nelson throws, the one that is so hard to describe. He summed it up in two words: “sweaty socks. Same tap, same glass, poured thirty seconds apart.
We ran the usual taproom argument (is it the beer or is it you?) and settled nothing, because the honest answer is that we were both right. That argument is worth taking seriously, and not just as bar talk. A lot of how we buy, rate, and recommend beer quietly assumes that everyone is tasting the same thing. We are not. And once you accept that, the way most of us pick our next pint starts to look a little silly.
Some of this is Genetics, and it isn’t Subtle
The clearest example is bitterness. Sensitivity to certain bitter compounds tracks to variants of a single gene, TAS2R38, and the differences are big enough that researchers sort people into rough tiers. A supertaster (roughly a quarter of the population, by most estimates) experiences bitterness at an intensity the rest of us would need a much higher dose to match. Hand a supertaster a 70 IBU West Coast, and they are not being dramatic. They are drinking a different beer than you are, in the only sense that matters. Bitterness is the famous one, but the aroma side might be stranger.
Modern hop varieties lean hard on thiols, the sulfur compounds behind the passionfruit and grapefruit character in something like Nelson Sauvin or Citra. Perception of specific thiols varies from person to person, and not by a little. The same compound can read as tropical fruit to one drinker and as sweat or cooked onion to the drinker next to them. Which means the sweaty socks verdict on my Nelson beer wasn’t an insult, and it wasn’t a broken palate. He and I were pulling different signals off the same pour, and both readings were accurate. Cilantro is the everyday version of this. Some people get bright citrus, some get dish soap, and the difference is heritable. Nobody argues that the soap people have a broken palate. Beer deserves the same courtesy.
A 4.1 with 9,000 Check-Ins is an Average of Other People’s Tongues
Here is where it gets practical. When you look up a beer’s score on a rating app before ordering, you are consulting an average taken across thousands of palates that may disagree with yours at the level of individual genes. The big sample size doesn’t fix that. It just gives you a very precise measurement of a question you didn’t ask. There is also a style problem hiding in those numbers. Anyone who has spent time on the check-in apps knows that pastry stouts and double hazies float toward the top of the scale while well-made lagers sink into the 3.6 swamp.
The average is not only somebody else’s palate. It is somebody else’s palate plus a thumb on the scale for whatever styles the rating crowd happens to reward. Readers here don’t need much convincing that perception is slipperier than the dogma says. Brülosophy’s whole triangle test methodology exists because tasters, under controlled conditions, routinely fail to distinguish beers that homebrew lore insists are worlds apart. If trained tasters wobble that much on “same or different,” a five-point score from a stranger is a shaky foundation for your next $9 pour.
Pandora Solved a Version of this for Music in 2000
There is an old answer to this problem from a different industry. When Pandora built the Music Genome Project, they made a bet that looks obvious in hindsight: instead of recommending songs because other people liked them, they paid trained musicians to score every song on hundreds of concrete attributes. Vocal timbre, syncopation, tonality, and on down the list.
Your thumbs up taught the system which attributes you responded to, and it recommended music by matching song structure to your reactions. The insight wasn’t the technology. It was the order of operations: describe the thing itself first, then let each listener’s reactions do the personalizing. Averages never enter the picture.
Beer Might be Easier to Map than Music
A song needs 450 attributes. A beer gives up its identity with far fewer. Style lineage, hop bill, IBU, residual sweetness, body, roast character, ester profile, ABV, and the package date will get you most of the way. Call it 30 dimensions. None of them is exotic. Half of them are already on the label or in the brewer’s notes. Two more signals finish the job, and they come from the drinker, not the crowd. The first is what you claim to like. The second is what you actually reorder, and the gap between those two is real. People tell you they love bitter beers and then quietly rebuy the amber every week.
Revealed preference beats declared preference, in beer as in everything else. Freshness belongs in the model too, and almost nobody puts it there. A ten-week-old haze is not the same beer that left the brewery, yet it gets rated under the same name, and those stale scores wash into the same average. Some of the “meh” pours that people blame on the brewer are really just calendar problems. This is the bet behind Barley, an AI beer companion my team built at Brewlytics.ai: score the beer the way Pandora scored songs, watch what an individual drinker actually reorders, and match the two.
It also borrows from drinkers whose ratings overlap yours, which is a politer way of saying it ignores the general public on purpose. I am obviously biased about even if that approach works. The underlying claim, though, stands on its own: your next great beer is a matching problem, not a popularity contest, helping you become a better taster with every recommendation.
Steal the Method for Your Own Tasting Notes
You don’t need software to act on any of this. A few habits will teach you more about your own palate than years of leaderboard scrolling. Blind yourself first. Have someone else pour, or wrap the cans in foil, exactly like a triangle test. Sight tasting drags every expectation you own into the glass. Score attributes instead of an overall number. Bitterness, sweetness, body, fruit character, whatever matters to you. One 1-to-5 score hides the reasons; four small scores expose them.
After a couple of months, you will see your actual preferences in the data, and I would bet money they don’t fully match what you would have written down on day one. Note the package date every time, especially when trying a hard beer. When a beer disappoints, check the calendar before you blame the brewer. And track what you rebrew or reorder. That list is your revealed preference, and it is the most honest tasting note you will ever produce. The leaderboard answers for everyone at once. You drink one pint at a time.
