busybee Food Coffee Talk

Explanatory videos

Here are the videos that Hoffmann produced about the project as a whole as well as the three decaffeination processes he went with:

Background

I’ve had plenty of Swiss Water and Ethyl Acetate coffees in the past, but had never had (or even heard of) Subcritical CO₂. I’ve had supercritical CO₂ but that was not a part of this test. The “common knowledge” I’d heard about supercritical CO₂ is that it has the least impact on the flavor but ironically is only used on very cheap coffee since it only makes sense in extreme bulk; the supercrits I’ve tried have not impressed me, as such.

In general I’ve held the impression (untested) that Swiss Water has the closest flavor to the original, and that ethyl acetate changes the flavor a little to make it a bit sweeter and more caramelly. But I’ve never done a side-by-side of them.

Test setup

For my test setup, I have the following equipment:

• Grinder: Timemore Sculptor 064s
• Scale: Timemore Black Mirror Nano
• Kettle: 1L Bodum Bistro
• Water: Seattle municipal water, which is already ideal for coffee
• Bowls: some snack bowls from Target
• A dosing cup
• Some sparkling water from my Sodastream

The bowls I used comfortably hold around 180mL, so I opted to go with 150 grams of water and 9 grams of coffee in each one (to maintain Hoffmann’s suggested ratio of 60g/L). I also pre-purged the grinder with 1 gram of each coffee beforehand, as the Sculptor does tend to retain quite a bit (not as bad as most consumer burr grinders, but having a bit of purge is still a good idea).

I ground at the #7 setting on the Sculptor, which is somewhat finer than I normally grind at for pourover. I took a deep whiff and didnt notice any differences in their aromas. I then poured 150 grams of water¹ into each one and let it brew for 5 minutes before scraping off the crust and foam.

Early comparisons

It was quite striking to me just how visibly different each roast looked at each step of the process.

The non-decaf had significantly more chaff in it than the decaffeinated beans². After brewing, it was surprising to me just how different they looked; the non-decaf and EA looked very similar to each other, but the Swiss Water took on a reddish hue, and the CO₂ was significantly brighter in color.

After scraping off the crusts, the Swiss Water was darkest, EA the lightest, and non-decaf and CO₂ looked most similar to each other.

Flavor profiling

I started out by slurping the various coffees to try to determine a rubric for the standardized tasting notes. I very quickly found that it was very difficult to do this in a way where I could feel like I was being objective about the notes. Depending on the order in which I sipped and slurped, sometimes I’d notice more or less in terms of acidity, bitterness, aroma notes, and so on.

One thing I did pretty consistently notice is that non-decaf and Swiss Water tended to have the most complex flavors, CO₂ tended to have a bit more of a sharp/acrid flavor profile, and EA had the least complexity. Sometimes I would notice nutty, sweet, caramel, or chocolate notes, but I had a very hard time telling whether any of that was subjective or if it was just wishful thinking.

I do feel like I most consistently tasted caramel notes in the EA and chocolate notes in the Swiss Water. I didn’t notice any differences in the body or texture, acid was just all over the place depending on what order I tasted things in, and whether I picked up on any floral notes or not seemed like it was just random (but I feel like I got them most consistently from non-decaf and Swiss Water).

Preferential ranking

So, the flavor science felt like I was going nowhere fast with it, so I decided to just try to figure out my overall preferences for things based on doing some A/B comparisons.

For this part of the cupping, I started out each test by washing my palate with sparkling water, and then did back-and-forth comparisons between two coffees, one slurp each done twice (so, basically, water,coffee A, coffee B, coffee A, coffee B). I systematically went through every coffee combination in order, including both directions to try to limit bias, and then on things I wasnt quite sure of I went back and redid those tests in random order.

Here is the result of those tests; the row is coffee A, and column is coffee B. For the scores I gave +1 to the coffee I preferred, and -1 to the coffee I didn’t like as much; in the case of a tie I gave no points in either direction.

Summing up the two test scores, I came up with this ranking:

• Non-Decaf: 3
• Ethyl Acetate: 2
• Swiss Water: -1
• Subcritical CO₂: -6

The interesting thing from this test is that I slightly preferred Swiss Water more when it was the second one tasted rather than the first, but from both modes of testing it’s pretty clear that I like non-decaf slightly more than EA, EA somewhat more than Swiss Water, and subcritical CO₂ is by far my least-favorite.

This is a bit surprising to me because in the profiling phase I found that Swiss Water tasted the closest to non-decaf, and I still preferred the flavor of non-decaf over EA, but of the decafs I definitely liked EA the best.

That said, all three decaffeination processes resulted in quite tasty coffee, and I’d be totally happy drinking any one of them.

I also suspect I might have had some bias against the CO₂ one simply because it came last in all of the rungs of testing and by the time I got to it in the end my palate was getting a bit fatigued. That said, even when I did the random-ish re-tests at the end (not all of which involved CO₂, it was still consistently less-good than the others.

Aftermath

I put my purge beans into a pourover cone, and filtered all of my cupping remnants plus an additional 200mL of water in order to end up with the biggest darn quarter-caf pourover I’ve ever had. It’s pretty good.

Conclusions

For this specific crop of these specific beans roasted with this specific profile, my decaf choice would be Ethyl Acatate, but I’m happy to buy either EA or Swiss Water going forward.

(And, heck, I’d probably try another subcrit if the opportunity arose; I just haven’t seen any of them for sale in general.)

I’ll be interested to see if S&W’s roast profile changes my opinion. Also, for the preferential matrix I will fully randomize the order of my tests to try to avoid biasing against the version which happened to come last. I will also attempt to randomize the coffees themselves so it’s a reasonably blind test. I have Ideas about how to do this in a fair way, which I will explain in the actual test article.