another_jim wrote:Thanks for trying this out. Can you say how relatively coarse or fine your grind was, when making these sciencey shots, compared to normal?

I can't give an accurate comparison because I didn't dial in a 20g shot first. If I have some extra time and enough coffee on hand I'll do that.

I can say that compared with the grind setting for 8.5g singles producing 17g in 35-40 seconds that I pull with the same coffee, the 15g shot was considerably coarser, five to six tick marks coarser on my Monolith Flat. Under normal circumstances, one tick mark represents about five seconds of shot time, so this correlates pretty well.

One might expect the grind setting for a standard 20g shot to be quite a bit coarser than the grind setting I use for a single, but bear in mind that I use your basket hole area formula to convert a double dose to a single dose, which normally that puts the grind setting for the single within 1-2 tick marks of the double. I'd expect the setting for a 20g dose to be a little coarser, perhaps one to two tick marks. So I'd estimate the difference comes to about 3-5 tick marks for the sciencey shot. Still quite a bit coarser on my grinder.

My sense is that the coarseness of the grind is only part of the story. I think the basket geometry is playing a big part here, especially because VST baskets don't perform like other baskets when the dose varies by more than a gram or so. I believe that's because the hole size is large, which in combination with the large amount of headroom may lead to the flow rate getting out of hand and possibly the puck breaking up. I think that's what happened with my 11.3g doses in the 15g VST basket.

Again, my gear pump might be throwing things off. I'd like to see the experiments done on a current model GS/3 MP or other machine with a rotary pump and a pressure gauge in the group head.

As James Hoffmann said, I think the value of these shots might be for people who have grinders that can't consistently produce good espresso. At least you can get a drinkable shot with some sweetness, albeit diluted, instead of a steady stream of sink shots. The use case for cafes is a stretch at best.

I'm glad to see that people are trying this out.

Dick, it's probably a little hard to tell if you hit the nail on the head without using a refractometer, since these shots are so ridiculous that they kind of look the same. I found that going a little finer for a shot that looked identical to the previous one resulted in a 2% EY bump and improved the flavour quite a bit. It does occur to me, though, that one thing that might help people may be to compare these with traditional shots by making americanos/long blacks. That way, the volume of the shots is irrelevant, since they're all topped up with water to the same volume. For what it's worth, with a filter roast that I usually use for espresso, I got pretty good shots ... I'll come back to it later, maybe with something like a washed yirgacheffe with some really delicate and fleeting characteristics in it, which will hopefully give me a good idea of whether the shots offer better or worse flavour clarity than normal espresso. Also, did you give the short shots a try, or just the fast ones?

What size basket do we use?

I asked Michael, the lead author of the paper and the guy who did the miserable task of pulling all the shots and measuring everything, about this on Instagram and his response was:

20g VST baskets were used for the main experimental data (though I collected data that wasn't included in the paper using 18g and 15g baskets, going down sequentially in dose from 20g). 15g baskets were used to collect data for the Oregon cafe section of the paper. I'd highly recommend 15g doses in a 15g VST basket.

I had a look through the maths part of the paper and it looks like they use radius and depth of the puck (not the basket, I think) as an input, but they don't have hole size, total hole area or headspace as inputs. The closest that I could see to anything to do with the baskets being an input to any of the maths was "overpressure", which is defined as the pressure excess relative to atmospheric pressure, which is applied by the pump. However, I couldn't see how this is measured or at what point in the machine. It seems like the mathematical model doesn't do anything like accounting for the different hole sizes and total hole areas that may exist in different baskets, and I don't really know what impact that might have. I also can't remember if the hole sizes and total hole area in the VST basket series are supposed to be the same, or if they were supposed to allow greater flow in bigger baskets. However, anecdotally, I do notice that I go coarser as I go up the VST basket series in order to maintain the same flow rate, but I haven't actually compared EY when doing so. So I have further head scratching to do to work out theoretically if the 15 or the 20 ought to be better or why. Opinions and speculation welcome. The simple answer is that one of the authors of the paper is telling us to use the VST 15.

What about pressure and flow rate?

Jake, all good points that I will pretend to understand to some extent, and re-read a few times until I actually do! But for people who are wondering what to do ...

Michael also said the pressure was set using a Scace 2, but he didn't know the flow rate:

Scace 2, on the Opera with gear pumps (no rotary/vibratory pump) can't recall the flow rate sorry (it was 4 years ago when we started!)

I'll have a bit more of a look at the paper when I have some time. I think there's a reference in the supplemental material to "static pressure". But if you look at figure 3, the punchline seems to be that if you extract at lower brew pressures, EY goes up. So I suspect that dropping the pressure some amount is probably going to help, so I think if you are getting a brew pressure reading that's 3-4 bar lower than whatever your normal brew pressure readings are when you are brewing at whatever you think is a standard shot, then you have probably taken steps in the right direction.

I should point out that it looks like that figure 3 is based on the maths, not actual experimental data, since it seems to have more or less the quasi-linear change between grind setting and extraction ... but figure 4 shows us that that's not what you actually observe in the real world. So there's a fairly valid question of whether or not in fact one would observe in the real world the predicted circa 2% increase in EY at the same grind setting that the maths predicts. I'd be surprised if we actually get this in practice, but I haven't made back to back shots with the same coffee to check this. There's also the interesting prediction that EY skyrockets at 4 bar, so I might try that, just for fun, too.

What about preinfusion?

I couldn't see that there's anything in the maths that uses anything to do with preinfusion as an input, but I was kind of skimming. It looked to me like the maths assumes constant pressure, however it is that pressure is measured.

I've created a profile for the DE that preinfuses slowly, pauses to allow the shot to bloom (similar start to the "blooming espresso" shot, but tweaked a bit) and then ramps up to a flat flow rate. I gave that a quick go and extraction yield seemed to jump up 1% and the shot seemed a bit heavier in body and a bit more chocolatey, but that's like an n=1, or near enough. I'm trying out some other stuff at the moment, so I might share that profile so others can have a play around with it, too. My thoughts were that the bloom start should increase EY, but the ridiculous flow rate should keep us out of the "clogged flow state" to minimise EY variability. A pressure profile would work too, but I went with flow profile because that way it's pretty easy to adjust by looking at pressure.

Why, exactly, are we reducing dose to 15g in the first place?

As I mentioned earlier in this thread, I've posed the question to Prof Hendon in the comments of the article of essentially how did they make the leaps of starting off by iterating to the same EY, which they acknowledge tastes different, then maintaining the same EY to yield 3/4 as much dissolved solids, so that they get something that is neither taste equivalent, nor dissolved solids weight equivalent (which will determine strength of any standard drinks that the shot is diluted into, such as the milk drinks). I'm very happy to have my ignorance and stupidity pointed out if that means we learn something, which seems worthwhile, since I've spoken to a lot of people about this article and no one else seems to understand this part. Unfortunately, I haven't had any responses on the article, nor on instagram, nor from Michael on instagram. In some senses, it doesn't really matter, since I'm still trying it and it will taste how it tastes regardless of the explanation.

I think that there's the question of why one couldn't make the super fast shots from the 20g dose, since that will yield the same dissolved solids weight and therefore will hopefully maintain strength in milk drinks (albeit that they will be diluted somewhat by the additional water in the shots). A good response to that is clearly that if you do that, you are just moving to a point that, by definition, the barista has already decided is worse than the barista determined tasty point (BDTP), but I don't understand from the article why there's any reason to suppose that any of the alternate shots are any better (in terms of absolute taste, at least ... I can see that they may have less EY error based on figure 4, but I'm still struggling to work out why we care about EY error unless less error means that the shots taste better).

I'd like to see the Decent blooming profile you used.

I like to find the tasty spot, where I get the highest and most even extraction, and then just stay there. Isn't that what we all do anyway?

I changed the parameters to 19g/40g in ~40s (20g VST) using 6 bars and I think my espresso tastes great. I'll order 15g VST to pull 15g shots, but for now, I'm sticking to 6 bars of pressure.

Bunkmil wrote:I'd like to see the Decent blooming profile you used.

If you go to the Decent Basecamp, I created a new thread called "Slow PI, bloom and gush - experimental profile" and you can download the profile there, or you can just program one for yourself. It's basically:

• preinfusion at a flow rate/time
• pause for maybe 20s
• 5ml/s flow rate

I discovered EY probably goes up a little and you can grind a little finer if you preinfuse at 4ml/s instead of 2ml/s. The pressure uptick may not be reliable to exit preinfusion on this profile, so don't rely on it - better to calculate the PI volume you need and set the time accordingly. 9 seconds at 4ml/s seemed about right for 15ml. 15s at 2ml/s also seemed to work for the 2ml/s flow rate ... it might not have totally saturated; maybe it needed to be 18s or something.

Anyway, on closer review of a few more shots, the yield uptick may have been due to my pulling slightly longer shots, so TBC if it does indeed increase EY.

It's been a while since I've posted here! I had a twitter conversation with the author, and wrote a blog post about my thoughts. I thought I'd link to them here in case folks are interested. TLDR; he agrees that he didn't take roast profile into account, and hopes this starts a conversation.

Twitter conversation: https://twitter.com/dylanmc/status/1230926897166475264

My blog post, a modest proposal: http://mecodegoodsomeday.org/coffee/mo ... posal.html