When I was in Rotterdam in August, I had coffee with my friend Han. She asked me two interesting questions when I told her I had been in coffee for 30 years: 

What changed over the past 30 years? 

What beliefs did you once hold and decide were wrong? 

Those questions seemed like the basis for an interesting blog post, so thank you Han for the inspiration. 

What changed over the past 30 years? 

Well, everything and nothing, depending on your point of view. The big trend changes are obvious: light roasts are more common, high-quality green is more common, data collection and technical proficiency in coffee has improved, processing styles have gone wild, and baristas have more tattoos than they once did.

This may surprise readers, but I don’t think top-quality green has changed that much over those years. In the 90s, Kenya, Ethiopia, and Colombia were pumping out green offerings that would shine on any of today’s cupping tables. I remember having a life-changing Kenya Kirinyaga in 1994 at Coffee Connection, George Howell’s former company, just before he sold it to Starbucks (truly, one of the most unfortunate takeovers in coffee history.) While I didn’t know how to score then, my internal heuristic for an 88+ coffee is “I will remember this coffee vividly in a year.” I remember that cup pretty well 30 years after drinking it, so I suppose that puts the coffee around 90 points. 

In the 90s, it was easy to get what I would call 87-point green from Bill McAlpin’s projects (Costa Rica La Minita being the most famous), Kai Janson’s farm in Panama (Kai: I’m still waiting for those geisha samples!), importers such as Tim Castle, author of The Perfect Cup, and Kenyans from Jeremy Woods. Supply and demand were such that those coffees were affordable and available spot. Not many roasters sampled pre-ship samples months ahead of time in the hope that the coffee would be a similar quality level upon arrival. These lovely coffees cost an average of $3.50/lb (roughly $7/lb, or $14/kg in today’s dollars.) 

It wasn’t until Kevin Knox of Allegro paid, if I’m not mistaken, a record $10 per pound for a Kenyan at auction in ~1998, that green prices began to wake up and high-quality lots slowly began commanding the differentials they deserved relative to average specialty lots. Ironically, many roasters at the time were critical of Mr Knox for his bold move, because they didn’t want to pay more for green. Many of those same people would now call such a buy a sign of fairness or a well-deserved reward to a producer. I don’t know Kevin, but when I first read about his auction “win,” I thought “it’s about time someone valued those Kenyans much more than other coffees.”

Interest in both precise and experimental green-processing techniques have of exploded over the past three decades. Although I’m not a fan of the average “experimental” processing, I greatly value the experimentation, as some experiments will lead to breakthroughs we will need to maintain and improve quality, especially due to the challenges of global warming.

For example, I posted a story on Instagram about a robusta I tasted in July, processed by Diego Bermudez and his team. I couldn’t tell it was a robusta, which surprised me. Diego and his team are experimenting with processing robusta as part of a ten-year plan to be ready for the future. That cup wouldn’t be my choice for a daily driver yet, but it made me more optimistic about the future of green. Note that coffee was worlds better than the mediocre “specialty” robusta people claim is the future of specialty coffee. 

Thirty years ago, I had not tasted coffee outside of the US, so I can’t speak to the trends in other countries at that time. In the early 90s, about 50% of sales were batch brew, 20% retail bean sales, and milk drinks were steadily growing in popularity, partly due to the spread of Starbucks. In 1995, the most common customer requests were French Roast, Colombian, and hazelnut-flavored drip coffee. Everything outside of those three was a little too exotic for the average customer. We won over customers by demonstrating that one didn’t have to roast dark to prevent sourness or weak cups.

“Light” back then was a roast dropped closer to the beginning of second crack than to the end of first crack. “Light roast” weight losses have dropped steadily over thirty years from around 16% to 12%. For reference, Starbucks has proudly stayed near 20% the entire time.

For those newer to coffee who love light roasts, please don’t be too hard your 90s brethren: without good data collection or control systems, no one was capable of roasting light with enough consistency or precision to prevent frequent, and often disgusting, underdeveloped flavors. Even in the 2000s, the transition to very light roasts was ugly, and there were years where I was not served a single competent light roast in a cafe. Roasters were becoming famous simply for roasting lighter than anyone else, regardless of competency or consistency. Unfortunately, the “lighter is better at all costs” mentality hasn’t completely died. Lighter is better only if you know how to do it well. In 2000, almost no one knew how. Today, I would put fewer than ten US roasters on the list. Most “Nordic style” roasters, regardless of location, proudly sell the celery/corn/twiggy flavors of underdevelopment, and too few consumers are aware one can have lovely, transparent, nuanced roasts without those flavors. 

I owned my first cafe from 1994-2001. I roasted on a 12-kg “Sasa Samiac” roasting machine (the French build roasters!) with a gas dial similar to that on a kitchen stove, and a slide-gate damper to influence airflow. The first day I roasted a batch, the fire department showed up. The learning curve was shallow. A few years later I bought a UG22 and it felt like driving a Corvette after owning a Ford Pinto. Cropster was still merely a gleam in Andreas’ and Norbert’s eyes, and I filled piles of notebooks with time and temperature data, sometimes staring at it for hours searching for answers. 

In the 90s in the US, typical drink sizes were 12oz and 16oz (360ml and 480ml, respectively). The “venti” mercifully hadn’t been invented yet). I put the first-ever timer on an espresso grinder to control doses, I stacked two hopper on top of each other to prevent the dreaded grind coarsening and popcorning of beans when the hopper got low, I installed a commercial humidifier in my cafe/roastery to stabilize green moisture content and shot quality (the difference was shocking), and I threw out batch brew that didn’t sell within 30 minutes of brewing. In year one, we threw out more than we served. Everyone thought I was nuts. By year five, we were serving over 1000 drinks per day, mostly batch brew, seven days per week, in a town of 30,000 people. Maybe not so nuts after all. I learned latte art from the photos in David Schomer’s book (not David’s fault, but my latte art was never that great), I briefly put Monsooned Malabar in my espresso (ultimately not for me), we weaned people off of flavored syrups, and mastered the crafts of ice-blended coffee beverages and toddy-style coffee brewed cold overnight in a lineup of 20-liter buckets. 

When I sold my cafe in 2001, I spent four years traveling the world, and fortunately stumbled upon the coffee scenes in Australia and New Zealand. While there probably wasn’t a decent filter brew to be found in either country, there were civilized, small, Italian-sized beverages, and unrivaled milk quality and texture. I remember getting schooled in milk texture by Kate at Coffee Supreme (sorry Kate, I forget your last name, but you were amazing!), Dave Lamason at People’s Coffee (he won’t remember me, but I remember that cappuccino!), and Jason Moore at Mojo (now roasting his own at Vanguard Specialty Coffee in Dunedin). I swear it took ten minutes for Jason’s milk to separate. I still don’t know how he did it.

After that adventure, I opened a cafe/roastery/bistro in the US with more focus on milk texture, smaller beverages served in ACF cups, and tried to do well by antipodean standards, while navigating an inherited, overpowered 23kg Gothot with a drum made of sheet iron. The Giesen also served as my sample roaster, roasting 100g batches with a constant air temperature of 420°F (it works). Meanwhile, places like Stumptown and Ritual began roasting light enough to make George Howell blush, beverages in the US were mercifully, if slowly, shrinking in size outside of the walls of Starbucks, and blueberry-tinged natural Ethiopians were providing more and more people their “aha moment” about specialty coffee.

It was 2007 when Hacienda La Esmeralda’s Geisha won the Best of Panama competition and fetched $130/lb at auction that the industry really woke up to the sort of prices the best lots could command. Kevin Knox probably laughed out loud at the haters. The rest is a history you are probably familiar with. 

With the advent of the coffee refractometer in 2008 and Cropster in 2011 (?), data slowly began to influence how coffee was roasted and brewed. Data has helped improve consistency and quality, and data is here to stay, whether one likes it or not. There is little doubt that the next 30 years of coffee roasting and brewing methodology will be shaped by data-driven learning. It’s only a matter of time until AI-driven machines outperform the best humans. I just hope I retire before that happens :0. 

What beliefs did you once hold and decide were wrong? 

My answer is again version of “everything and nothing.” The biggest change in my beliefs over the years has been what is an appropriate roast level. Although I saw “cinnamon roasts” (not too far from today’s “Nordic style” roasts) in open barrels at Zabar’s in NYC during my childhood in the 80s, such light roasts were otherwise unheard of. Like most lighter roasters of the 90s, I dropped coffee shortly before second crack began. I never saw anyone other than Zabar’s do that, and I suspect they did that to maximize yield, not flavor. 

My old friend and mentor James Marcotte visited my cafe in Amherst circa 1998 while I was batch brewing a blueberry-tinged Ethiopian Harrar in an old American Metal Ware urn-style brewer (a la Peet’s, complete with a gentle stir of the grounds during prewetting!), and told me I was grinding too fine. It was one of those “aha” moments that made me realize grinding too fine often led to astringency, and I became a lot more conscious of finding that “highest non-astringent extraction level” for all brewing methods. At that time, I became obsessed with astringency and eliminating it anywhere I could. Of course, we weren’t yet measuring extractions (Vince Fedele patented the coffee refractometer in 2008), but we knew grinding finer or using a lower brewing ratio resulted in stronger coffee. It seemed that “too fine” was the point where a brewing method produced astringency (now we believe that happens due to an increase in bypass with finer grinds.)

Another area where I was wrong was in the source of astringency in coffee. I, along with almost everyone, believed that astringency came from “overextraction” along channels. However, I began to have doubts in 2019, when I extracted up to 30% making “blooming” shots on my Decent Espresso Machine and didn’t detect more astringency than I did from more typical extraction levels. Jonathan Gagné seems to have solved this mystery; his plausible hypothesis is astringency comes from larger, undissolved molecules that are able to “escape” the coffee bed through larger flow paths, such as channels and areas with bypass. 

At one point I believed that the tighter the grind-particle-size distribution, the better, at least for filter coffee. Regardless of the practical impossibility of a fines-free or truly “unimodal” PSD, this seems to be another one of those “the happy medium is best” situations. Some fines seem desirable, and we seem to be in the early stages of learning what an “ideal” PSD looks like, if there is one. 

In roasting, we knew nothing but folklore and old wives’ tales before Cropster and Artisan appeared on the scene. I knew certain combinations of batch size, roast times, and final roast colors seemed most likely to produce results I liked, but we didn’t have the data or the control systems to be confident of much else. I targeted total time as well as time-before and time-after first crack, as well as final bean temperature and weight-loss numbers, and tried to be as consistent about those as possible. In the 90s, I was happy if I hit my targets within a 10-second time range. These days I consider a three-second miss to be cause for concern. 

In the 90s, air roasters and that Sivetz guy were quirky and interesting, but it was hard to take a machine seriously that had virtually no controls, and no cooling bin. (The modern Sivetz machine is much more sophisticated.) It took many years before seeing a quality air roaster in action to form a valid opinion about it. 

In 2008, I did a consulting job at 49th Parallel in Vancouver, and it was a turning point in my career and understanding of roasting. The owner, Vince, wanted me there, and was serious about quality at any cost. His staff resented my presence (this often happens on jobs; the boss wants progress, and the staff want to defend the status quo and to not look bad), and I had to hit a home run to win them over and get them to buy into a better system. 

The first night of the job, I spent four of five hours poring over reams of roasting data in spreadsheet form, tried to discern what the better batches had in common, and what I discovered there gave me the confidence to write The Coffee Roaster’s Companion. 

The next day at 49th, we roasted two coffees with 88-point potential, a washed Kenyan and a washed Ethiopian. On the third day, we cupped and brewed day two’s roasts and I remember Vince sitting down, staring at a cup of coffee, and saying “I didn’t know the coffee could taste like that.” The staff were won over, we implemented a system based on a new methodology, and 49th reigned for a while as the best and most consistent roaster in North America. It was one of the most rewarding consulting jobs in my career. 

I always knew writing the first professional-level book about roasting would cause endless controversy and bring out the haters. Certainly, I made mistakes, most of which I righted in Coffee Roasting: Best Practices. But CRC accomplished two undeniable things the critics don’t seem to have noticed: it introduced systematic approaches to roasters (no more “smell-a-vision” as Mark likes to say), and it opened a public conversation about roasting that had not yet existed. Roasting was still in the “we can’t share our secrets” phase (translation: we are too insecure to share our beliefs, because they may be wrong), and I stuck my neck out, ready for the trolls to chop it off. (Editor’s note: I still have my head, and now “walk the talk” with Prodigal.)

The most famous (or infamous) advice in The Coffee Roaster’s Companion is a recommendation to drop coffee at a DTR of 20%—25%. Was that a mistake? Perhaps in delivery, but not in substance. Do I roast to 20%—25% DTR? Not often. Huh? At Prodigal, for example, we use an air roaster and average lower than 12% weight loss per batch. We’ve occasionally neared 20%, but most roasts end up with much lower DTR. How do I square the advice in CRC with my current approach?  It’s simple: CRC was written for all of the world’s roasters, not just the 1% who roast the lightest. We light-roast lovers tend to live in a bubble and sometimes forget that perhaps 99% of the world’s roasts are dropped after first crack has ended, and most of those are dropped during second crack. In CRC, I invented the concept of DTR, and I made a recommendation that is fitting for a very large proportion of the world’s roasters. Of course, if you choose to be in the extreme 1% of a spectrum, typical approaches may not work well for you. 

The mistakes I made in CRC are things that don’t get much notice. In the years prior to writing the book, I had the experience many times of the first batch of the day being underdeveloped if the machine was not adequately warmed up. That led me to think higher thermal energy at charge is important for adequate development. With experience, I’ve learned that is not necessarily true. These days I see finding the “happy medium” in phase balance and initial thermal energy as best practices to help prevent underdevelopment. Likewise, I underestimated the importance to development of minimizing airflow and roasting slower, especially in drum roasters. 

Another famous piece of book advice was to attempt to achieve a constantly declining rate of rise (ROR), preferably with a somewhat constant slope. I have no way of proving that is the optimal approach to roasting, but I follow it today with excellent results. Perhaps a better question than “is that the right target?” is “what would be a better target?” I have yet to read or hear a plausible suggestion of a superior approach. Optimal or not, it’s a target that keeps many roasters on the right track.

I’m sure I’ve updated or replaced countless beliefs I cannot even remember. That’s okay, as my goal is to change my mind and grow, not to be dogmatic. The internet isn’t always so understanding, and writing a book has a way of causing others to believe I haven’t changed my mind since publication, which is absurd. I change my mind routinely. Prodigal’s roast quality and consistency improve every week. This learning allows to help clients more precisely every year. However, most roasters don’t seem to improve or change much over time. When I stop learning, I will retire. 

I assume 99% of what we believe at any time is incorrect. Even scientific belief has been incorrect for 99% of science’s history. But science has a mechanism for updating beliefs and facts as they evolve, and that’s probably the best we can do. I hope the next thirty years teach us all the ways we are wrong now, and bring us new, better beliefs and processes that help us to make progressively better coffee. 

********

Prodigal drops a new, fabulous coffee each week. Join our mailing list to get exclusive, first access to new offerings