Kendall is still working on the BET analyses of the four K10-Lundgaard samples. The BET process coats the coffee samples with nitrogen at low temperature and pressure, and accurately measure the amount of nitrogen used. This gives a highly accurate mass to surface area ratio for the sample (don't ask me how. If you can make sense of of the explanation, tell me instead).

In the mean time, he took some SEM photos of each grind sample. Kendall cautions us not to read anything into differences -- the way the samples are prepared can introduce large biases. The photos merely show how far from smooth spheres ground coffee is.

The sample the diffractometer measures as coarsest, Hopper-18:


The other three, roughly the same on the diffractometer, Hopper-15:


Single-18


Single-15

Are those mostly fines in the SEM pictures? Looking at the original graphs, I'd expect to see a lot of coarse particles (or at least a large volume of coarse particles) in the 400-600 micron range, most of these look like fines which are around the 50 micron size.

Also, just out of curiosity, what's the background, a filter or something?

The background is a porous, double sided, conductive carbon tape. Samples were prepped by homogenizing the coffee ala WDT then dipping the tape into the coffee. The stub was then blown off by a can of air. Whatever stuck is what you see. The idea was to not apply pressure and further damage or align the coffee with a spatula. So the sample may be biased by what happened to stick as well as by what happened to remain after dusting off.

Jim et al.,
I am extremely curious to see where this all ends and hats off to your approach to validity versus statistics.

One point in the debates so far i do not understand and maybe someone can explain to me.
I see the test/retest "noise."
I also read a debate about differences between grinders and if they are big enough to be meaningful.
So, it was said the differences between grinders are smaller than the test/retest noise.
Two examples then were mentioned: Kony and VersaLab M3.

I copied both Kony and VL graphs, making sure they exactly were the same size and then traced created a copy with handmade traces of the Kony and VL graphs that plot the 15g single.

To my surprise, in the light of discussion above, the difference between the two is rather big and LOOKS bigger than the test/retest and certainly big enough to potentially be meaningful.

Now the problem with the particle shape I had speculated about when you were on page 1 of this thread, and actually the way beans break in a grinder may be influenced more by their genetics, their origin (climate, terroir) and/or roasting than the grinder, but I would still assume the grinder to have some impact.

Anyhow, curious to see the outcome this next Titan project.

Regards
Peter

The point about the test/retest is that it showed when you compared the same grinder's samples, say 15 and 18, or hopper and no hopper, they were usually no larger than the test/retest of the same sample. You are comparing two different grind samples from different grinders and machines.



Here's a graph of all the samples. Pick a high one and a low on, and they'll be very different. But if you look at the entire set, you can see they come from the same population. Given the grind to grind changes, there's no basis for claiming large differences between grinders.

I see Jim's point that he underlined in a PM. Even though the data shown so far do not support the statement that variance inside sample classes per machine is as big as the entire sample population's.

(I myself) Having called "espresso" a multi-variable optimization space in which we are on the quest for god coffee, and such spaces usually are characterized by several regional optima that however are suboptimal at the global level, I now wonder if Jim is deconstructing all sorts of supposed myths. That would be a good thing in itself, if alone from a scientific point of view if each myth would be a hypothesis that needs repeated attempts of refutation.

In this light, and the research questions in this thread, mythbusting or attempts at positive proof, I feel a look is needed at this so-called optimization space (picture below).



This is just a 5 minute attempt to add perspective to the debate and my points above. By no means meant to be sound and complete, just a start of a debate.
Particle shapes (caused by grinder/burrs or not), tamper (tamping and tamper profile), filter basket (nr of holes, hole size, hole size distribution, basket shape, etc) or water quality are but a few variables that have been discussed in these fora and are not in the model picture.

The question is if myths are to be busted, or indeed have been, what are the most important variables in this optimization space, and, are god shots just lucky shots within the noise space of this implied mutli-variable model?

KR
Peter

Great flow diagram

CafSuperCharged wrote:
This is just a 5 minute attempt to add perspective to the debate and my points above. By no means meant to be sound and complete, just a start of a debate.
Particle shapes (caused by grinder/burrs or not), tamper (tamping and tamper profile), filter basket (nr of holes, hole size, hole size distribution, basket shape, etc) or water quality are but a few variables that have been discussed in these fora and are not in the model picture.

The question is if myths are to be busted, or indeed have been, what are the most important variables in this optimization space, and, are god shots just lucky shots within the noise space of this implied multi-variable model?

KR
Peter

It's really quite simple, at least in my view. Unless you start out with high quality, fresh coffee, all the rest of it is process without a purpose.

90% of the work is going to be done before the green beans are roasted, in growing (and sourcing) the beans and processing them in a way that does not damage them. The beans must be transported to the end market so that they can be used while fresh, or they must be preserved in some fashion (notably freezing) before they are to be roasted and used.

Bad roasting can ruin good beans, as can staleness, however the ability of good or gifted roasting to improve what was in the beans to begin with is limited; I'd give this 5% of the remaining 10%.

All the other process variables that have to do with producing the finished beverage are not worth more than 5% of the total (sorry, gifted baristas ). Bad preparation of the beverage can ruin what is in the beans, but the potential for "improvement" is limited. This does not mean that these process variables are unimportant, merely it means that they are more likely to subtract from the end product then they are to add to it.

Exactly how much one attributes to such variables as the machine or grinder used, or the techniques used to pull the shot, pale in comparison to the importance of the raw material (coffee).

In summary, everything that follows the arrival of the green beans in the country where they are to be used has much more potential to diminish the end product than they do to "improve" it. This is one reason why I find obsessing about process and equipment to be relatively unimportant in the greater scheme of things.

ken

@Ken - yes and no.
I subscribe to your analysis in general and that is why I set up the model the way I did.

The question however is, do you start with the most delicious roasted beans (100% quality level) in the world (probably very expensive) and next do not do a great job as a barista with a few chains of variables that each, because of sloppiness, score, say, 95% of what is attainable, so the end result of five variables at 95% is 77%.
Or do you start with less good roasted beans (e.g. because the region/terroir is not the best there is) and blend a few different characteristics to something that is perceived 90% quality.
And next as a barista you attempt at 99% perfection, so the end result of the five times 99% barista controlled variables times the 90% of the coffee results in 85.5% perceived coffee quality. I think the latter, my personal prejudice, is more representative of the Italian espresso market (and no, they generally do not work at the 99% level and the end result is rarely over 70%. IMHO)

And in this light all the discussions we have here at HB are, or could be, very meaningful indeed.
Or, let's see where Jim et al.'s work gets us in terms of this model/reasoning.

KR
Peter

I think any "chicken or egg" discussion that tries to put the steps from bean to cup into a hierarchy of importance is doomed to become a freshman dorm mental exercise.

I have had too many good beans, properly roasted, which were obliterated in the hands of unschooled baristas, and too many commercial-grade bags from Italy, rendered as very drinkable espressos by skilled baristas, to take these discussions very seriously.