another_jim wrote: Just because the authors hired a PR firm to hype this paper, doesn't mean it's science. Colbert coined the term "truthy," this paper is "sciency." If you read Science News or better yet Science Daily, you'll be surprised at the number of sciency papers that have gone through PR departments you get nowadays.

What is the name of the PR firm in this case? I haven't seen evidence this study has a PR firm promoting it so if there is one, please tell us what it is. And if there isn't, stop with the "fake news" claims.

Sify.com, Cnet, IFLScience,CNET again, Food & Wine, The New Indian Express.Yahoo, Popular Science, CBS, Smithsonian Magazine, SBS, The Shropshire Star, Fairfax Good Food, USA Today, Ars Technica, CTV News, etc etc etc.

I'd say that there was a PR push...

Not necessarily. The news often works where one media outlet picks up a story, others like it and just copy it. If there is a pr push it should be possible to find a press release online. If there is one, why not find it and post a link?

I suggest leaving finding the real pedigree of the 'study' alone and focus on what content there is, aside from an arms length of impressive looking math...
Should I add I'm not too convinced about the content? Catching something as complex as flow (air or fluid) in a system with variables changing as much as in espresso is nice to keep folks busy on a rainy afternoon but hardly meaningful IMO. Should the article have been written around extraction yield and taste as evaluated by a validated blind tasting panel (as in able to detect any difference in a meaningful way) in relation to duration of extraction and grind size it might have been an interesting read.

OTOH; they did what they say to have set out for: find a method that produces consistent output.

The point of this paper was to give people a map for making an espresso beverage that they like and then be able to make it 100 times in a row."

with the underying reasoning that IMO is not that far off, when looking at how many 3rd wave shops make their espresso, cramming 20 g in 18 gr baskets etc.

Most people in the coffee industry are using fine-grind settings and lots of coffee beans to get a mix of bitterness and sour acidity that is unpredictable and irreproducible," said Hendon. "It sounds counterintuitive, but experiments and modeling suggest that efficient, reproducible shots can be accessed by simply using less coffee and grinding it more coarsely."

here I differ with the authors; I find that most coffee shops are very predictive in producing bad tasting espresso in a reproducible way...the real topic that they should have investigated and should have been their primary outcome measure IMO is quality, as in taste and vs reducing waste.

As far as I'm concerned they have now succeeded in very predictably and consistently producing nothing but waste, small amounts of spent grounds ending up with an end product that is not palatable means more waste.
I predict that any coffee shop would be more profitable when they find a way of getting repeat business, IMO the best way of doing that is quality; stellar coffee, the convenience customer will return to charbucks.

another_jim wrote:Yes, and that's what I find hard to believe. By all means try it for yourself -- Take a 20 gram VST basket, and dose it at 15 grams. What kind of grind will it require not to get a complete melt down shot? We're talking VST baskets here, where even a 1 gram under dose is hard to handle. Now, I don't know the espresso machine they used. If it's got a really tight gicleur, all the extra head space may get a 10 second dwell time and, with the coarse grind, a 5 second melt down pour, so technically, a 15 second shot. But in that case, the extraction and taste claims become hard to believe. But in eiher case, something doesn't add up.

I did try it, but with a Decent basket instead of a VST, since that's what I had in at the time. I had to go coarse, I'm pretty sure, but I didn't take notes. The nice thing about the Decent is that it's boop beep boop and the machine is programmed to preinfuse at Xml/s until the puck is saturated, then give flat flow at 6 bar until you extract Xml/s, so it's pretty easy to do. I will do my homework and report back with numbers!

another_jim wrote:Just because the authors hired a PR firm to hype this paper, doesn't mean it's science. Colbert coined the term "truthy," this paper is "sciency." If you read Science News or better yet Science Daily, you'll be surprised at the number of sciency papers that have gone through PR departments you get nowadays. But if you get practiced at reading the abstracts; the distinctions between science and sciency are obvious.

I think that readers really need to be careful to distinguish what is backed up and supported by rigorous experimental work and what is subjective opinion.

All of the stuff about extraction yield and extraction yield variability absolutely looks to be backed up by a lot of data. Prof Hendon and Michael gave some sense of the work that went into it at the talk in Melbourne that I went to last year, and it sounded like an utterly thankless chore. The paper explains it. Michael even said that he didn't get good data several times, so he had to repeat some of it a few times. We're talking lots of weekends of making shot after shot and measuring everything. I feel that everything that stems from that data needs to be understood and respected.

There are a few things that the paper seems to scientifically show quite well:

• As you drop extraction pressure, EY increases.
• At typical recipes, we must expect a certain amount of EY variability.
• If you want to extract more, you can't simply go finer. There is a limit.

All of the stuff about taste and preference is an entirely different matter. The paper is very clearly written not to be asserting that the taste outcomes are supported by any sort of rigorous testing whatsoever. In fact, it refers to the "barista determined tasty point". The paper takes as its predicate that a taste judgment has already been made. The paper then talks about how to achieve shots of the same extraction yield as the shot at the BDTP, but with lower extraction yield variability. The paper clearly acknowledges that the fast/long shots and the cut short shots (the two lower variability methods that it proposes) will taste different from the shot at the BDTP, and the paper suggests that mixing together the two types of shots may regain complexity, but there is little discussion in it about how each of the three shots (or the blended shot) will taste different from each other. This is left for us to do. In fact, there's some discussion in it about how the EY of the different shots will consist of different compounds and therefore taste different.

There is clearly a lot of stuff that could be put to taste tests:

• The paper seems to start from the position that the EY variability at the BDTP is a taste problem. Well, the test for this would be to do a whole bunch of triangle tests of BDTP shots with the actual EY measured, to determine that people can in fact distinguish between them. If people can't distinguish between at least shots at the extremes of the EY whisker for that data point, then all of the stuff about worrying about EY variability goes out the window.
• The paper says that the two alternate methods result in more consistent EYs. Again, triangle tests. People shouldn't be able to pick the odd one out.
• The shots will taste different. Do they? Triangle tests. (Well, these should be super obvious, I would have thought.)
• What do people actually prefer?

I'd have to read the paper again; I'm really not sure to what extent the authors are trying to assert things about what tastes better. There are some media quotes about the article where they seem to say that the different shots taste better. To the extent that they do say that, isn't it just subjective opinion? That's not to say it should be dismissed - readers should treat it like they treat any other subjective opinions; if they respect the author's opinions they should just give it a shot. To the extent that the authors are being taken out of context, it would be nice if the people reporting and commenting on the article didn't.

There also seems to be a step that people are kind of not focusing on; I think people who are trying this out are probably comparing 6 bar 15s shots against 9 bar 2:1 shots. I don't think you can do that. People should probably compare 9 bar (or whatever they're used to) 2:1 vs 6 bar 2:1 at the same shot times, then compare that 6 bar 2:1 against the 6 bar 15s shot. Also, people who are trying it should probably try the cut short shots that the authors recommend in the paper as well. So far, looks like everyone has just tried the long fast shots.

There's another aspect that's interesting and that's how roast level relates to all of this.

--

EDIT: I screwed up the following. The point was to test 15g in the 20g basket, but I didn't actually swap the baskets. I am posting this here as a shameful mea culpa and admission, since I think this sort of transparency is important for people to have confidence in what they read. What follows is correct as it relates to the 15g basket.

another_jim wrote:Yes, and that's what I find hard to believe. By all means try it for yourself -- Take a 20 gram VST basket, and dose it at 15 grams. What kind of grind will it require not to get a complete melt down shot? We're talking VST baskets here, where even a 1 gram under dose is hard to handle. Now, I don't know the espresso machine they used. If it's got a really tight gicleur, all the extra head space may get a 10 second dwell time and, with the coarse grind, a 5 second melt down pour, so technically, a 15 second shot. But in that case, the extraction and taste claims become hard to believe. But in eiher case, something doesn't add up.

Jim, after many years of reading and enjoying your content, the following is the least I could do to try to provide some amusement ...

"STAND BACK, I AM ABOUT TO SCIENCE", I screamed.*

*Disclaimer 1: This is not science. Also, I didn't really scream that.

So, I had my grinder dialled in for a 15 g shot in the 15g decent basket to yield about 37g over a fairly typical brew time ... I don't know; probably 25s ... using a declining pressure, lever-type profile. It's not a flat 9 bar reference type profile, but I wasn't about to waste coffee to dial that in. Grind setting was 2.25 on the Monolith Max. That delivers a flow rate that starts at about 1ml/s and then peaks at around something like 2.2ml/s.

I asked my machine to deliver me the following shot conditions:



(The volumetric stop is a new feature that John recently added; I thought I should give it a go.)

I fitted the portafilter with the 20g decent basket, dosed 15.3g, and got 35.3g in 29s. I was surprised that this was a little slow, but then I remembered that I'd just dropped the pressure to 6 bar, so probably not all that surprising that it was a little slower than I had expected. Also, in hindsight, I think I bumped the temp up 1C, which may also have slowed the shot.

I then guessed what if I try more or less where I'd expect the 20g basket to sit and maybe add 0.25, so I adjusted to 3.5, dosed 15.3g and got 44.9g in 13s.

Here's a photo. Red = temp, blue = flow rate (probably the more important one; it's the dependent variable), green = pressure. Pale lines = first shot; more solid lines = second shot.



Disclaimer 2: Don't judge me by the shots that I got; first they gushed a little, then they gushed a lot.

So I ran the numbers and got an EY of 20.5% on the first shot and 19.7% on the second shot.

Now I suppose that actually none of this is especially relevant, since I suppose I should be comparing this with a dialled in 20gr shot. But I guess the subject matter of this post is really supposed to be how coarse a grind do you need. Seems to me that the grind setting that you need is in the ballpark of the grind setting required for 20gr in the 20gr basket.

Guess I should try 15gr at the same setting in the 15gr basket and see what happens.

As for what's in the cup, no surprises that the gusher was thin - they told us it would be. Both shots were worse to my tastes than the original shot. They both seemed to have less clarity of flavour than the 9bar/decline shot; pretty muddy. The fast shot marginally less so, and probably a little less bitter. I'm currently using a light roast kenyan, and suspect that it can just hack much more aggressive extraction conditions. Maybe if you use a more traditional espresso roast, there's more bitterness and the bitterness reduction is therefore really obvious and significant. But this is pretty n=1, so take all of that with a large boulder of salt.

Well, a true expert will be testing the study's conclusions soon enough. I saw this Tweet from James Hoffmann:

@jimseven wrote: Jan 23
I have seen the paper on espresso.

I have read it, but only once.

I will read it more.

I will test its recommendations.

I will make a video about it afterwards.

I've had too many emails, @'s and DMs to respond to, for which I apologise.

Thanks for trying this out on the Decent

luca wrote: Now I suppose that actually none of this is especially relevant, since I suppose I should be comparing this with a dialled in 20gr shot. But I guess the subject matter of this post is really supposed to be how coarse a grind do you need. Seems to me that the grind setting that you need is in the ballpark of the grind setting required for 20gr in the 20gr basket.

Guess I should try 15gr at the same setting in the 15gr basket and see what happens ... But this is pretty n=1, so take all of that with a large boulder of salt.

Yes it was just one try; but what you saw concurs with general experience -- under dosing a 20 gram basket with 1 5 grams of coarse grinds doesn't add up to a 15 second shot.

I played around with a 18 gram VST and a 15 gram coarse-ish dose on the Bianca (adjustable needle valve machine). I could slow down the initial water debit to get a 7 to 10 second dwell time. But I couldn't get the shot pressure above 3 bar for the extraction. To get up to 6 bar, I needed a fairly fine grind, not much different than a regular 18 gram dose for the basket required. Given adjustable pressure and water debits, you can even brew regular coffee on an espresso machine, and get high EYs, but it won't be espresso.

another_jim wrote:Thanks for trying this out on the Decent



Yes it was just one try; but what you saw concurs with general experience -- under dosing a 20 gram basket with 1 5 grams of coarse grinds doesn't add up to a 15 second shot.

I played around with a 18 gram VST and a 15 gram coarse-ish dose on the Bianca (adjustable needle valve machine). I could slow down the initial water debit to get a 7 to 10 second dwell time. But I couldn't get the shot pressure above 3 bar for the extraction. To get up to 6 bar, I needed a fairly fine grind, not much different than a regular 18 gram dose for the basket required. Given adjustable pressure and water debits, you can even brew regular coffee on an espresso machine, and get high EYs, but it won't be espresso.

Hi Jim,

Your thanks and confidence in me were totally unfounded - I buggered up and didn't actually swap the 20g basket in, which I found out when I went to try a 20g shot with the same grind setting this morning. The grind setting that I found (3.5) yielded 35g in 12s at 6 bar with 15g in the 20g basket, but it tastes markedly different from last night's shot. As you say, it's sort of like a filter coffee. It's maybe a little lacking in complexity of aroma and flavour, but it has no bitterness and it does indeed have typical Kenyan coffee flavours and an intense sweetness and juiciness. I don't have time to do refractometry this morning, so I'll do that and report back properly next week.

I've been using 15g/37.5g yield for a while, at fairly usual brew conditions. I've also tried the allonge type technique, with great success. But I haven't tried this. And the cups that this has produced are super delicious, so I'm definitely going to look into this further.

Cheers,
Luca