307 | Kevin Peterson on the Theory of Cocktails
What this covers
Kevin Peterson, a mechanical engineer turned mixologist, joins Sean Carroll to argue that cocktail craft rests on optimizing distinct, measurable variables rather than intuition or tradition. The conversation traces how temperature, dilution, and aeration — controlled by the method and duration of shaking or stirring — matter as much as what goes into the glass. Peterson's central insight is that cocktails decompose into families sharing fixed ratios, so mastering one template (say, the sour) unlocks many drinks across different spirits. The episode examines how taste vs aroma decomposition splits the drinking experience: the tongue perceives only five basic tastes, while aroma — which carries complexity, storytelling, and surprise — can be composed using perfumery notes framework, much like top, middle, and bass notes in music or fragrance.
The discussion moves through several distinct territories. Peterson explains why ice melting matters more than non-melting chilling, how shaking amplifies aroma while suppressing taste perception, and why a cocktail formulated to balance at 35°F tastes entirely different at room temperature. He addresses the social dimensions too — how glassware and color shape enjoyment despite having no chemical effect. On the practical side, he describes his Cocktail Genome Project, which uses structured questions about taste memories and scent preferences to design personalized drinks. He also demonstrates that complexity creates tolerability bands: intricate drinks like the Negroni forgive small errors, while simple ones like the Old Fashioned do not. Throughout, Peterson grounds his claims in controlled experiments — from empirically testing daiquiri ratios to studying how ingredient pairing works at the molecular level — while acknowledging where bartending knowledge and sensory science still diverge.
Peterson argues that a great cocktail is the product of optimizing separable, measurable variables — taste balance, temperature, dilution, aeration, and aroma — and that general principles (families, golden ratios, perfumery thinking) make cocktail craft a reproducible science rather than guesswork.
- Cocktails decompose into families with shared ratios, so mastering one template generalizes to many drinks (Mr. Potato Head theory)
- Engineering variables (temperature, dilution, aeration) set by shaking/stirring matter as much as ingredient choice
- Aroma can be composed like perfume/music (top, middle, bass notes), and added complexity widens a drink's tolerance bands
Aroma—not taste—creates a drink's complexity and intrigue; it is best designed using metaphor and understanding common aroma molecules.
- Studies suggest that training your sense of smell does not lower your detection threshold or make you more sensitive; instead you learn to describe smells better, build a richer descriptive vocabulary, and refine your expectations so you can categorize and pick apart aromas more quickly — like a sommelier or an experienced whiskey drinker primed to look for specific notes.
“as you train your sense of smell, you are not necessarily getting more sensitive... What you learn is how to describe them better.”
- The tongue perceives only bitter, sour, salty, sweet, and umami, so balance is a matter of getting those ratios right; everything else — the storytelling and intrigue of a drink — comes from aroma, where a single sweetener like honey carries floral or caramelized notes the tongue reads only as 'sweet'.
“your tongue only perceives bitter, sour, salty sweet, umami. And every cocktail can be defined as some ratio of those.”
- Because scent has few numbers and few literal words, aroma in a cocktail is best handled through metaphor — the perfumery/music model of top notes (light, citrusy), middle notes, and bass notes (deep, dark, wintry like tobacco or bonfire) — letting a bartender build harmonies, chords, and deliberate dissonances rather than relying on guess-and-check.
“you can smell something very light and bright, more on the citrusy side and say, okay, I can see how that would be a top note... that's more bass note.”
- Ingredients tend to pair well when they share common aroma molecules, which is why surprising combinations like blueberry and horseradish can taste good despite sounding awful — and the more terrible a pairing sounds, the more delightful the subversion of expectation when it works.
“ingredients pair well together when they share some common aroma molecules”
- The same smell can be identified differently by different people and even reacted to positively or negatively by the same person depending on context — e.g., the molecules in Baijiu appear in both foot fungus and Parmesan cheese — so you must be careful which referent you name when describing an aroma.
“There's a Chinese spirit called Baijiu, that has some molecules that appear both in foot fungus and in Parmesan cheese.”
- The difference between cheaper and more expensive spirits is not greater exquisiteness as price rises but fewer flaws, and many aroma compounds are good or bad depending on concentration — the dose makes the poison — so a molecule like ethyl acetate (green apple at low levels, nail-polish-remover off-note at high levels) can be a feature in a cocktail when dilution and temperature bring it into range.
“the difference between the less expensive spirits and the more expensive spirits is not more exquisiteness when you get more expensive, but fewer flaws.”
Taste perception depends on temperature, dilution, and aeration; shaking and stirring are engineering choices with measurable sensory consequences.
- Whiskey stones and non-melting chilling components don't work well because they have little heat capacity; ice's big advantage is the large energy transfer during melting, which gives a large temperature change from a small amount of melt, so chilling the glass is a better strategy than non-melting cubes.
“they're great. They just don't have a lot of heat capacity. So they don't actually buy you that much temperature change”
- Taste perception is temperature dependent: a drink (or food like ice cream or cola) formulated to taste balanced at 30-35F sends a different ratio of taste signals at 70F, so a correctly-made cocktail served warm is effectively no longer the right drink, since the same sugar reads as unbearably sweet warm.
“your taste buds are temperature dependent and when something's warmer, you're sending a different ratio of signals in.”
- Neither shaking nor stirring is right or wrong; they simply deliver different amounts of temperature drop, dilution, and textural change, so the choice depends on what you want in the final drink and whether your ingredient ratios will be balanced at that level of dilution and temperature.
“It is stirring and shaking give you different amounts of temperature drop, dilution, textural change. What is it that you want in the final drink?”
- Shaking traps air bubbles that boost aroma — they atomize aroma molecules that pop into the nose when you drink — while simultaneously suppressing taste, because bubbles cling to the tongue and block liquid from reaching taste buds; aeration is effectively all-or-nothing, so a shaken drink differs dramatically from a stirred one even at identical temperature and dilution.
“the shaking boosts aroma and drops the level of taste.”
- Shaking or stirring is not primarily about mixing ingredients (which happens in a couple seconds) but about setting the engineering variables — temperature, dilution, and aeration; the ideal shake is about 12 seconds and hard enough to slam the ice into the tin ends, which is far longer and harder than most home users and even most bartenders shake.
“It is stirring and shaking give you different amounts of temperature drop, dilution, textural change. What is it that you want in the final drink?”
Simple drinks demand precision while complex drinks tolerate wider error margins; balance point and dilution zone constrain what works.
- Counterintuitively, more complex cocktails like the Negroni have wider tolerance bands for ratio, temperature, and dilution, because the many overlapping components mask small errors — whereas simple drinks like the Old Fashioned are balanced on a knife edge with nowhere to hide a mistake.
“those more complex drinks tended to have wider tolerability bands”
- A normal 10-12 second shake produces about 30% dilution, and drinks tend to fall off a quality cliff around 35% dilution, so pre-chilling ingredients (e.g., keeping rum in the freezer) melts less ice during shaking and keeps the drink in the ideal dilution zone.
“If you just shake a normal drink for 10 to 12 seconds, you're gonna get about 30% dilution.”
- By holding rum constant at 2 ounces and varying lime and sugar in quarter-ounce increments until the drink failed, Peterson found the ideal classic daiquiri to be 2 oz rum, 3/4 oz lime, 3/4 oz simple syrup, with a wiggle-room region that empirically validated some anecdotal cocktail wisdom.
“kind of the ideal is 2 ounces of rum, 3/4 lime, 3/4 simple syrup. But there's a little bit of wiggle room on either side.”
- To make a shaken Old Fashioned that is still balanced, you must boost taste ingredients (more sugar and bitters, since shaking suppresses taste) and upgrade to a better whiskey (since shaking amplifies aroma, including off-notes), so an entry-level whiskey that is fine stirred reveals its flaws when shaken.
“because shaking inhibits some of that taste signal, you have to boost the amount of taste ingredients... But it also enhances the aromatic elements of the drink... really what I found there was I had to use better whiskey.”
Cocktails fall into families with shared structural roles; learning ratios and principles unlocks improvisation and beats memorizing individual recipes.
- Rather than memorizing thousands of individual cocktail recipes, you can recognize that drinks fall into families defined by component roles (spirit, sweetener, citrus, etc.), and within a family the same set of ratios applies — so a pisco sour and a whiskey sour share the same proportions, letting one set of ratios unlock many drinks.
“I know that a pisco sour and a whiskey sour have the same ratios. I don't have to memorize both drinks. I just have to memorize one set of ratios.”
- The best way for a home bartender to improve is not to work through every spirit and cocktail family but to pick one drink you love and build outward — running ratio studies on it, then varying one component (whiskey, sweetener, bitters) at a time — exploring one row or column of the possibility table rather than the whole table at once, which also avoids buying expensive single-use bottles.
“find a drink that you love... and work your way out from there.”
- Because cocktails are sets of structural roles, you can substitute any ingredient that fills a given role (any citrus for lime, any sweetener for sugar) and still produce a balanced drink, which both unlocks creativity and lets you improvise with whatever you have on hand.
“I know that a pisco sour and a whiskey sour have the same ratios. I don't have to memorize both drinks. I just have to memorize one set of ratios.”
Bartenders should discover a guest's preferences through structured questions, then design a drink that pushes them beyond their current comfort zone.
- Peterson's Cocktail Genome Project flips the usual bar interaction: instead of handing a guest a menu of unfamiliar ingredients and asking them to choose, the bartender asks structured questions about taste, scent, and culinary history (favorite childhood candy, cuisines, happy scent memories) to build a flavor profile and design a drink ideally suited to the person, while pushing them one step beyond their current comfort zone.
“Can we kind of flip the script and say, well, what if instead of me telling you about the drinks, what have you told me about your preferences?”
- The ideal cocktail is both a moment in time and something that evolves over years, so a good cocktail experience pushes a guest one step beyond their current preferences — meaning a drink you don't love in the moment can still be a pivotal, formative experience, as Peterson's own hated-but-pivotal first Negroni was.
“the ideal cocktail, it's both a moment in time, but it's also, it evolves over time... being able to sort of push people one step beyond where they're at today.”
- To build a 'Mindscape' cocktail matching a spirit-forward, bittersweet, complex, novelty-seeking palate with a happy scent memory of forest-after-rain (petrichor), Peterson designed a Petrichor Negroni: gin distilled with vetiver (a grass smelling of post-rain earth), St. George Bruto Americano for a fir-tree bitter note, and Antica Formula vermouth for a deep sweet-vanilla offset, stirred over an ice cube with expressed orange peel.
“we're gonna make a petrichor Negroni... distill our own gin using vetiver... let's use St. George Bruto, which is also a red bitter liqueur, but it's got a very prominent fur note... And let's use Antica vermouth.”
Bitterness tolerance is narrow and specific; taste preferences and social factors like gendered glassware shape cocktail enjoyment.
- Most people enter the cocktail industry from the people side — an entry-level job, working up to behind the bar, then adding technical details later — whereas Peterson entered backwards, with all the technical skills and none of the people skills, making his learning curve harsh.
“a lot of people that wind up in this industry come to it from the people side... I came to it the other way where I had all of the technical skills and none of the people skills.”
- Drinks and their glassware carry social gender connotations that affect enjoyment — Peterson witnessed a husband swap his pink drink with his wife — and although he dislikes these associations, they remain a reality bartenders must navigate.
“I've totally had guests, husband and wife, where the husband had to switch drinks with his wife because his drink was pink.”
- Liking one bitter thing does not predict liking another, because bitterness tolerance is acquired in a very narrow band — someone who loves black coffee and green vegetables may still find Campari, Fernet, or Malort terrible — so preference questions must probe specific bitter experiences rather than bitterness in general.
“it can be acquired in a very narrow band. Maybe you dislike bitterness, but you've drank enough coffee that you love coffee. But that doesn't mean you're gonna like Campari or Fernet or Malort.”
- Roughly 30% of a shaken cocktail is water from melted ice, which Peterson jokes makes a $15 drink partly priced water.
“hey, I paid 15 bucks for this, and 30% of this is water. Can you charge me 30 less, please?”