I'm going going to try and argue the value of a flat brew temp profile. Honestly... I don't know if it's the right answer and, in fact, doubt there is a true "right answer" that can be generalized across all coffees and all desired flavour profiles.
But the reality is that a huge number of serious baristas work on machines that are designed to deliver a brew temp profile that is as flat as possible -- and now the machines are actually getting close to truly delivering this profile. In addition, we're now starting to see machines that can do this in a controlled, repeatable and accurate fashion.
The trouble is that many (if not most) espresso blends are not formulated to take advantage of this profile and, in fact, in many cases are exposed by this brew methodology.
I had assumed, as I'm sure many people had, that if you were to evaluate an espresso blend across a range of brew temps you would see the quality of the shots at these temps distributed on a bell curve. With the arrival of the prototype GS3 I finally had the chance to validate this belief. To my shock - it turned out to be untrue with the first coffee I tested. And with the second. And the third.
Now - in fact, it is true at a very high level. If you look at the distribution on a 1-2F scale you see a rough bell curve for all these coffees. The curve shows different forms, but it is a bell curve. But if you evaluate the coffee at tighter granularity - for example at a 1/3F scale - you no longer see anything like a bell curve. What you see instead is a saw-tooth peaky distribution across the top of the bell curve and then a progressive drop-off on both sides from there.
At first I doubted my results but after repeating them - across multiple coffees - I had to accept them.
It actually took a vacation on the coast with friends, Valerie and Bronwen to figure this out. It was actually Valerie who figured it out. It's wave forms - intersecting.
Okay, from here on out we're talking true theory - and you need to take it all with a grain of salt unless or until I get a chance to test it.
Let me explain... espressos are made of of multiple beans in a blend. Each one of these beans, I think, has an optimal brew temp and from there a distribution for quality that is, in fact, a bell curve. The trouble is that these curves are all different.
The reason there is the saw-tooth distribution at the top of the bell curve for the blend is that you are seeing peak and trough intersections between the various beans - across all the various attributes (body, clarity, flavour, aroma, finish, etc). This is creating positive interaction points as well as negative one.
The above chart is purely fictional and incredibly (over)simplified. But none the less, it should illustrate the point. Obviously, the chart really should be multi-dimensional to represent all the attributes and characteristics of each coffee and should accurately represent the value of each bean rather than making each one a perfect 10.
But even in this simplifed example you can see the trouble. The fictional blend illustrated here would probably have shown no problems or issues on older machines. The lack of temp stability intra-shot (no flat line brew temp) as well as inter-shot would result in a variance in the brew temp that would hide the differing brew temp profiles.
On one of the new and up-coming very temp-controlled this will not be the case. Sure - you'll be able to produce good espresso assuming the beans are good and the blend is good. In fact, you'll be able to produce 3 or 4 different tasting good espressos from this depending on your brew temp. But you will not be taking advantage of what the machines can deliver here.
Now, imagine that you were to evaluate your coffees as single origin espressos at various degrees of roast. Imagine that you were to determine the bell curves for each coffee along with the flavour profiles and flavour attributes for these beans. Now create a blend where the coffees are combined not just for flavour components and desired final flavour but those combined with intersecting brew profiles.
In the past we've had to pre-determine a brew temp (at a very gross level) and then evaluate experimental blends and single-origin components based on that one temp. This doesn't have to be the case anymore. Now, we can open it all up and let the beans tell us what to do.
Here is another fictional and oversimplified example.
Note that this blend has only three beans rather than five as with the previous example. Odds are going to dictate that it's going to be increasingly difficult to find beans that match an existing profile (flavour and brew temp) with each additional bean you add to a blend. As a result, I think we may see a decrease in the number of beans in a blend if people pursue this route.
This does not mean a decrease in overall quality or complexity of the resulting espresso. The myth that single-origin espresso was inherently inferior to blend; inherently less balanced and even inherently less complex has been pretty conclusively discredited. Using these high-quality single-origin coffees for your espresso blends should allow for the creation of coffees that take advantage of the strengths of flat brew profile, temp stable machines.
If I were a roaster/retailer with a cafe using Synesso machines or looking to buy a next-gen machine like the upcoming improved GB5 I would seriously consider looking at creating a blend based on the realities of this brew temp profile.
And if I were a serious barista competitor -- I would be working on with a competition blend based on this idea.