I think they’ve modeled espresso channeling? It’s well known by hobbyists. Although, they’ve quantified things nicely, and anyway having a result in the record that matches your gut is great!
No, but I think this was novel at the time for being able to model espresso puck dynamics well enough to produce a testable/tasteable prediction that led to a big shift in how coffee hobbyists think about espresso.
For an academic coffee paper, it is better than many.
A common sin remains: like most coffee papers, I was unable to find calibration procedures in the methods or supplementary sections for the espresso brewing instrument whose performance may vary between runs, days, or users. In this case, they claim/assume "The Opera allows for precise control of shot time, water pressure (PW), and temperature"). As a Decent owner, I'm less familiar with the Opera, but for either machine I would want to disprove any confounding variables by attaching independent sensors. Decent has openly discussed hurdles they've confronted for consistency and accurate measurement.
Their main takeaways, though, are interesting and track with how many now prefer to extract:
As we demonstrated in Figure 3, our model informs us that a reduction in dry coffee mass results in an increased EYmax (shown schematically in blue in Figure 6). Thus, a barista is able to achieve highly reproducible espresso with the same EY as the 20 g espresso by reducing the coffee mass to 15 g and counter-intuitively grinding much coarser (as shown in red, Figure 6B). This modification may result in very fast shots (<15 s), a reduction in espresso concentration, and a different flavor profile.
...
Beyond sensory science studies, a persistent difficulty is that there is no rapid route
to assessing the quality of two identical EYs made with different grind settings or
brew parameters. It is clear that espresso made at 22% EY in the partially clogged
regime tastes more ‘‘complex’’ than a fast 22% EY obtained using the optimization
routine presented in Figure 6. In an attempt to recover the same flavor profile as
the partially clogged flow regime, a shot must contain a mixture of higher and lower
extractions. Consider the tasty point in Figure 7: One can approximate its flavor pro-
file by blending two shots: (1) a low extraction/high dose (purple point) and (2) a high
extraction/low dose (green point). This procedure can more economically yield a
shot with a flavor profile that should approximate that which was previously only
obtainable in an economically inefficient partially clogged shot. Blending shots
does double the total volume of the beverage, and the procedure comes with the
added combinatorial complexity associated with calibrating two shots that, when
mixed together, yield superior flavor. We expect only the most enthusiastic practi-
tioners would consider this approach, but it may well be actionable in an industrial
setting where extraction is carried out in bulk.
Logic minimization is kind of boring? I had to solve a problem once and the answer was still to use the espresso software from the 1980s. It is a pretty specialized problem and honestly I don't see how you would improve on it, besides integrating the digital circuit design research. But in terms of software, there is not really any reason to use a Boolean logic formula instead of just passing around the truth table directly.
There is one important thing missing in the paper:
https://news.ycombinator.com/item?id=46526933
Amusingly I also added a coffee stain to my Tex-based e-ink newspaper:
https://imgur.com/a/diy-automatic-e-ink-newspaper-using-rust...
Not using that package, though.
I think they’ve modeled espresso channeling? It’s well known by hobbyists. Although, they’ve quantified things nicely, and anyway having a result in the record that matches your gut is great!
If I'm not mistaken this is the original "turbo shot" paper.
Oh wow, so this is what kicked off the whole channeling thing? Neat!
No, but I think this was novel at the time for being able to model espresso puck dynamics well enough to produce a testable/tasteable prediction that led to a big shift in how coffee hobbyists think about espresso.
For an academic coffee paper, it is better than many.
A common sin remains: like most coffee papers, I was unable to find calibration procedures in the methods or supplementary sections for the espresso brewing instrument whose performance may vary between runs, days, or users. In this case, they claim/assume "The Opera allows for precise control of shot time, water pressure (PW), and temperature"). As a Decent owner, I'm less familiar with the Opera, but for either machine I would want to disprove any confounding variables by attaching independent sensors. Decent has openly discussed hurdles they've confronted for consistency and accurate measurement.
Their main takeaways, though, are interesting and track with how many now prefer to extract:
As we demonstrated in Figure 3, our model informs us that a reduction in dry coffee mass results in an increased EYmax (shown schematically in blue in Figure 6). Thus, a barista is able to achieve highly reproducible espresso with the same EY as the 20 g espresso by reducing the coffee mass to 15 g and counter-intuitively grinding much coarser (as shown in red, Figure 6B). This modification may result in very fast shots (<15 s), a reduction in espresso concentration, and a different flavor profile.
...
Beyond sensory science studies, a persistent difficulty is that there is no rapid route to assessing the quality of two identical EYs made with different grind settings or brew parameters. It is clear that espresso made at 22% EY in the partially clogged regime tastes more ‘‘complex’’ than a fast 22% EY obtained using the optimization routine presented in Figure 6. In an attempt to recover the same flavor profile as the partially clogged flow regime, a shot must contain a mixture of higher and lower extractions. Consider the tasty point in Figure 7: One can approximate its flavor pro- file by blending two shots: (1) a low extraction/high dose (purple point) and (2) a high extraction/low dose (green point). This procedure can more economically yield a shot with a flavor profile that should approximate that which was previously only obtainable in an economically inefficient partially clogged shot. Blending shots does double the total volume of the beverage, and the procedure comes with the added combinatorial complexity associated with calibrating two shots that, when mixed together, yield superior flavor. We expect only the most enthusiastic practi- tioners would consider this approach, but it may well be actionable in an industrial setting where extraction is carried out in bulk.
I’ll wait for the Lance Hedrick / James Hoffmann video.
You'll have to wait negative six years. https://www.youtube.com/watch?v=BoYBLn9hRqs
MORE TO EXPLORE
Coffee. Vols. 1–6. R. J. Clarke and R. Macrae. Elsevier Applied Science, 1985.
Coffee: Botany, Biochemistry and Production of Beans and Beverage. M. N. Clifford and K. C. Willson. Croom Helm, London, 1985.
Caffeine, Coffee and Health. Edited by S. Garattini. Raven Press, 1993.
Coffee: Recent Developments. R. J. Clarke and O. Vitzthum. Blackwell Science, 2001.
Espresso Coffee: The Science of Quality. Second edition. A. Illy and R. Viani. Academic Press, 2005.
Association for Science and Information on Coffee: www.asic-cafe.org (sadly now a spam/gambling site)
International Coffee Organization: www.ico.org (seems to have a bad cert now?)
News from the industry of specialty coffee: www.scaa.org/chronicle/category/coffee-science (also dead)
from https://www.scientificamerican.com/article/the-science-of-a-...
a parameterization would be helpful for machine designers probably
I thought this was going to be about logic minimization. Was severely disappointed
Logic minimization is kind of boring? I had to solve a problem once and the answer was still to use the espresso software from the 1980s. It is a pretty specialized problem and honestly I don't see how you would improve on it, besides integrating the digital circuit design research. But in terms of software, there is not really any reason to use a Boolean logic formula instead of just passing around the truth table directly.
[dead]