A reaction to "Systematically Improving Espresso: Insights from Mathematical Modeling and Experiment"

By Dylan McNamee 2020-02-20

What just happened?

In late January, 2020, one of my favorite hobbies (making espresso) made the front-page of most of the news feeds I read. More exciting, it was about a scientific article on the topic of optimizing the quality of espresso, which is something I've been doing non-scientifically for about 20 years. Most "science of coffee" research I've seen in the past didn't actually address how I, at home, could make better espresso, but it seems that this paper was exactly about that.

After reading the first summary (I think it was from Ars Technica), I scanned the others, and the theme of of the coverage was "making better espresso by using less coffee". Based on my own experiments with espresso roasting and brewing, this conclusion suggested to me that the paper may have investigated some parameters of brewing espresso, but not the "roast profile" (which is how the coffee was roasted, including how dark or light the beans end up). In fact, I suspected that their results were based on a medium-dark roast, and was curious to read more about their method, and any parameters around roast that they experimented with.

Reading more of the web articles I noticed none of them mentioned the word "roast" in their summary of the paper, so I went to the source and started reading. Much of the paper is approachable - I particularly liked their graph measuring the degree and uniformity of grind (between "fines" and "boulders" - great language there). I was a bit concerned by the use of a barista's subjective "tasty point" - I get it, but they're not talking about consistency, but rather navigating a complex surface of flavors and using one (?) person's notion of which places represent local maxima. Some of parameters in the espresso recipes they examined started to make their recommendations make sense to me. Further, the paper doesn't mention (or describe) the "degree of roast" or "roast profile" of the coffees they used for their measurements. So I did what we do in this century, I sent a Tweet:

I was excited (and a bit nervous) to get a response from the lead author of the paper, Professor Christopher Hendon:

To make a long story short (the longer version is in the Appendix below), in my experience you need to adjust your espresso recipe when you include degree of roast (how dark it is) to maintain drinkability (not just "tasty point", but just to be palatable). An espresso recipe consists of the coffee being used (which beans and and the "roast profile"), the exact weight of coffee (down the 0.1 gram, which is about the weight of one bean), the fineness of grind, the amount of pressure used to "tamp" the ground coffee into the portafilter, the amount (or whether) to "preinfuse" the coffee with water at a lower pressure, and finally, the amount of time to pull the shot, while measuring the weight of brewed coffee that emerges from the portafilter. Most people hold the other variables constant: water temperature to 94° C and pressure to 9 atmospheres (although the paper experimented with lower pressures to draw out the extraction time with coarser grinds and less coffee).

These many parameters interact with each other in complex ways, a lot of which are really nicely described in the article. For example, making the grind finer, while holding the other variables constant, will increase the amount of time required to get the same weight of brewed coffee. In my experience (and I don't see this discussed much anywhere), making the roast lighter, while holding the other variables constant, has the effect of reducing the amount of time required to get the same weight of brewed coffee. So for a light roast, to maintain shot durations, the grind has to be significantly finer than the same weight of a darker roast. Environment also comes into play. For example as humidity changes, baristas will often tweak the grind coarser or finer to hold the remaining variables constant.

My non-scientific findings are the following: Narrowing the roast parameter to two points - one lighter roast, which consists (in my case) of roasting until the end of "first crack" (about 12 minutes and a final infra-red measured temperature of 205° C), and one medium roast, which consists of roasting so that second crack just starts in the cooling tray (about 14 minutes of roast time with final temperature around 220° C). With those two roasts, my conclusions:

  1. for the lighter roast, use 20 grams of beans in a "triple basket" ground so that it takes 35-45 seconds to pull 20 grams of water through the puck after a 20-second pre-infusion,
  2. for the medium roast, use a double basket and 15 grams of coffee pulled without pre-infusion for 25 seconds, which totally agrees with the Hendon paper's recommendations.

Further, in agreement with the paper, I have found that applying the light-roast recipe to medium-roast beans results in a more burnt-tasting coffee. Also, my attempts to use light-roast beans with the medium roast recipe results in a pucker-inducing sour/green-tasting espresso.

I'm just a hobbyist, and I am open-minded. I expect that I have missed a parameter, and/or have failed to explore a portion of the search-space where some delicious coffees lay, but I think these are two interesting points in the parameter space that both agree with the Hendon paper and also point out some long-standing complaints that each camp (the medium- vs. light-roast fans) have about each other's preferences can be met in the middle by modulating recipe.

A modest proposal

Professor Hendon says in an interview with Popular Science: “There is a tremendous dependency on the preferences of the person producing the coffee. We are elucidating the variables that they need to consider if they want to better navigate the parameter space of brewing espresso.”

My observation is that the above quote (and the article itself) seems to implicitly claim that they're elucidating all of the important variables in navigating espresso. My observation, which is confirmed via Twitter, but not acknowledged in the paper or reporting around the research, is that the study did not include roast profile among the variables they elucidated, but is in fact a very important variable, and one (which when taken into account) directly contradicts the primary claim of the paper. My modest proposal is to offer to work with the authors to demonstrate this claim, and to help facilitate follow-on research that adds roast profile to the parameter space of quality espresso. If they're interested, I'd actually suggest they approach Rob Hoos, of Nossa Familia coffee here in Portland, who is my favorite "roasting science" practitioner, because he approaches roasting as one of the tools that affects what ends up in a cup of coffee, as exemplified by the title of his great book: "Modulating the Flavor Profile of Coffee (One Roaster's Manifesto)".

Appendix: Some asides

Having family in Eugene, I took an opportunity to visit the café mentioned in the paper, and ordered a macchiato. It was pretty good. As I was drinking it, I spoke with the manager, and asked about his involvement with the research. He said it wasn't a surprise to him, and that he'd been doing things this way "for a couple years".

I've been involved in talking about degree of roast for longer than I care to admit. When I first came to Seattle after going to UC Berkeley, and my wife started working for a local coffee company, we both got into reading about "specialty coffee". One editorial in the local coffee journal (this was in the late 1980's) ranted about how awful it was that dark roast coffees were getting so popular, saying things like "no ketchup on this steak, thank you." I wrote a letter to the editor saying something along the lines of "there are lots of ways to enjoy coffee at different roast degrees", and noted that while on honeymoon in Italy, my wife and I noticed that the degree of roast used in espresso varied quite a bit, roughly getting lighter as you traveled North. The original author, David Schomer, wrote back, dismissing my opinion. So yes, I've been participating in "degree of roast flame wars" (as a moderating force, I hope) for about 30 years now.

Here in Portland, where light roasts rule, I was a medium-roast holdout for the longest time. The closest coffee shop to where I live, Stumptown, used to have a specialty outlet just a few blocks from me, called "The Stumptown Annex", which was like a laboratory, lecture space, enthusiasts den - they had a wall of different brewing systems on display, hosted talks and tastings, and so on. I had just got into roasting and was learning first-hand about the milestones of "first crack", which marks (to me) the middle of a light roast, and "second crack", which marks the middle/end of a medium roast. Once second crack is over, you're in the realm of Dark Roasts. In the spirit of exploring coffee experience, I asked the gentleman behind the counter at the Annex if they had considered offering a medium roast (something when second crack starts in the cooling tray, I believe I said). His response was prefaced by the words "Not to sound completely dismissive", and proceeded to lecture me about the evils of darker roasts, and then said "if you like the darker roasts, we do sell French roast". While he didn't actually give me the middle finger, that's exactly how the conversation felt.

My "light roast awakening" happened around 2005 at Albina Press, also here in Portland. They used Stumptown beans at the time, and had recently placed highly in the international barista's competition, so I wanted to check them out. It was a crowded day, the barista was really busy mostly making lattes. I ordered a macchiato. As he handed me my drink, the barista said "we've been so busy, I haven't been able to taste the shots for a while - could you tell me how it is?" It was amazing. Up to that point my favorite macchiato in town was from Spella Café, who uses a classic Italian medium roast, that day I added the light-roast macchiato from Albina Press. They're different drinks, by far, but both just floor me with rich, complex flavors that I could drink all day long if I had the constitution. I'm pretty sure the barista at Albina press that day was Billy Wilson, who now owns Barista Coffee. Try as I might, however, I wasn't able to reproduce that Albina Press experience at home on my Rancilio Sylvia (with PID control) and Mazzer Mini grinder.

My ability to make delicious espresso at home from lighter roasts took another decade or so. I started getting more serious about roasting. I think I was the first in North America to get an Aillio Bullet roaster. Unlike my earlier roasters (a Gene Cafe and a Behmor 1600+) the Bullet connects to a computer and gives you a real-time graph of temperature (and derivative) measured by an infrared sensor and also gives you very precise control over the heat-energy input via induction heat applied directly to the roasting drum. Next, through good timing, luck, and the fact that my wife and I are both coffee fanatics, we purchased a used Slayer single group espresso machine. The thing the Slayer can do that no machine we'd owned previously (two Rancilio's, La Pavoni, and an Olympia Cremina) could do is precisely control pre-infusion, which is wetting the beans before pulling the shot. Pre-infusion lets you pull very slow shots from very finely ground coffee, which is how you get delicious shots from a light-roast coffee.

I've found, with only the one exception noted above, that the coffee community in Portland has been really open, friendly, and receptive to teaching an enthusiastic dilettante coffee hobbyist like myself. Here's an incomplete list of the coffee folks I've had great experiences with: David Griswold (of Sustainable Harvest) gave me a very welcoming introduction to coffee roasting almost 15 years ago, Andrea Spella, who welcomed me into his roastery when I was just getting started, Steph, Joey and the whole wonderful team at Buckman Coffee Factory, who host the amazing Cascadia Roaster's Competition (and sell small-volume roasters like me 25 lbs at a time), Trevin from Mr. Green Beans, who sells an amazingly diverse selection of green beans, who also introduced me to the idea of buying a Bullet (way before they were available), and whose staff is always fun to talk about coffee with, Rob Hoos from Nossa Familia coffee, who was so interested in the Bullet, he invited me to bring it to his roastery where we spent multiple hours having fun talking and roasting. Rob's book (mentioned above) is really good, and his talks (I saw one at OMSI) are amazing firehoses of coffee information. Oh yeah, and my wife, Heidi, who was a 15-year veteran of Starbucks, who lets me make most of the espresso these days, and gives me (I think) honest critiques of my technique and experiments.