Obviously you gave the subject a lot of thought before posting this very level-headed writeup, Dan. Kudos to you.

HB wrote:Subsequently I would dock its Convenience / Features score due to the ungainly layout of the steam arm / steam arm toggle / brew button array. I'm hopeful that a smart product ergonomics design engineer will revisit the prototype control panel layout too. It's bank of same-sized buttons offer poor visual clues and no tactical clues to their different purposes.

Dan, thanks for sharing your experience. I wonder if you would have applied the same criticism to the current control panel (shown here)?



I also wonder how many other current changes would have affected its ease of use and performance and how many lie ahead before production begins.

Marshall wrote:I wonder if you would have applied the same criticism to the current control panel (shown here)?

First of all, let me state that everyone is an expert on ergonomics (or at least thinks they are).

I may be slightly more attuned to the subtleties because I worked for several years in a usability lab studying human/computer interaction (video cameras, two-way mirrors, pre and post-test interviews, etc.). Our company hires PhD research scientists to study these questions because literally millions of dollars are at risk if a product suffers poor usability (and yet it still happens all the time, but don't get me started). I was a programmer, designer, and tester working under the supervision of usability experts and some of their critical analysis skills rubbed off on me. Obviously computers and espresso machines are different product domains, but when it comes to control panels, I've seen what works well and what tends to test poorly.

Now ask yourself a few easy questions:

• Could someone with a visual impairment use the product with the same efficacy as someone with no impairment?
• Is there a clear visual hierarchy? Looking at the product, can you tell at a glance what's important and what's infrequently used?
• What are the common tasks and what areas does the user's line of sight and hands traverse to accomplish them?
• When a control acquisition failure occurs, how negatively is the user's experience impacted?

Back to the GS3, ponder the questions above while visualizing yourself at the helm. Where do you see a problem? I see many. None are even close to what one would characterize as a fatal flaw, but there's no risk of its ergonomics winning any awards. To answer your question directly, the new panel design doesn't appear to address any of my egonomic concerns. On the positive side, it is pretty and I bet the glow looks really cool in a dimly lit kitchen.

The GS3 is very tempting, and I may even end up skipping a vacation or two and getting it. But in the end, owning one wouldn't help me answer my espresso questions.

I think at this point the research boundary has moved to exploring and then testing specific questions about within-shot profiles:
-- pressure profiles, whether preinfusion followed by declining pressure, a la lever machines, is better, and if so how?
-- temperature profiles, is the HX hump better than the dual boiler straight line, and if so how?
(and if one more person says "smoother," I'll barf; even though I've said it enough times myself.)

Andy has handled the pressure part closed loop -- would a less exotic setup with an accurate gauge and well tapered motor dimmer be good enough, albeit not as accurate?

Sean is thinking about systems that can create any type of in-shot temperature profiles; and these may turn out to require fairly exotic gear too. Would an old-line thermosyphoned HX, modded to control the HX and group head temperatures at independently settable values, get us started? We already do this informally by varying flush regimes, but it requires several temperature monitoring points, and maybe PIDed or front panel adjustable thermosyphon restrictors. Such mods don't require nearly the same level of technology as fully programmable temp profiles, and they could provide several different, with luck, reliably repeatable, curves.

What I'm saying is that a lot of our purchases are not just driven by getting great espresso, but also getting enough control to see exactly how great espresso happens. Since our conversation has moved away from getting precision and on to getting different types of profiles, the GS3 may not be able to play a part in it.

another_jim wrote:I think at this point the research boundary has moved to exploring and then testing specific questions about within-shot profiles:
-- pressure profiles, whether preinfusion followed by declining pressure, a la lever machines, is better, and if so how?
-- temperature profiles, is the HX hump better than the dual boiler straight line, and if so how?

We all know that the humped profile makes smoother espresso.

Just kidding.

another_jim wrote:What I'm saying is that a lot of our purchases are not just driven by getting great espresso, but also getting enough control to see exactly how great espresso happens. Since our conversation has moved away from getting precision and on to getting different types of profiles, the GS3 may not be able to play a part in it.

Bill Crossland said that the GS3's brain board has the extra capacity required for pressure profiling control. But since the number of people who would be willing to pay the upcharges involved are low, it may be a while before that happens. Unless...researchers find it is useful and customers start to demand it. LM listens to customers.

But concerning the temperature profile, that ain't gonna change much.

Marshall wrote:Dan, thanks for sharing your experience. I wonder if you would have applied the same criticism to the current control panel?

I also wonder how many other current changes would have affected its ease of use and performance and how many lie ahead before production begins.

From your picture, it look like the current panel uses pretty much the same layout as the old one (shown below). In my experience, it is easy to mix up the buttons. Of course, that tendency goes away after a while.




I was hoping LM would keep the revisions I made to the display on their final production versions, but alas, I don't think they will.

AndyS wrote:From your picture, it look like the current panel uses pretty much the same layout as the old one (shown below). In my experience, it is easy to mix up the buttons. Of course, that tendency goes away after a while.

It's the same sequence you normally see on automatics: single ristretto, double ristretto, single, double. The steam, water sequence on the right might differ from some. I thought that moving the icons directly to the buttons, as LM has now done, might be less confusing to someone who was completely unfamiliar with commercial machines. I didn't notice any confusion among the Homecoming consumer users (apart from the ones who had no idea how to make espresso). It's child's play compared to any Sony remote control.

In any event I suspect anyone paying $4,500 for an espresso machine will read the part of the manual that explains what the 6 buttons do. Actually, I have two of the manuals (Installation and Software Programming) for anyone who has questions (sorry, no pressure programming).

Marshall wrote:In any event I suspect anyone paying $4,500 for an espresso machine will read the part of the manual that explains what the 6 buttons do.

I agree wholeheartedly, anyone with a triple digit IQ will figure it out in seconds.

Mini-rant: As part of my work in user interface design, I debate this point frequently and your comment typifies the attitude of most software developers - "they'll learn how to use it." Of course, the purpose of good product design is to enable the progressive discovery of the system's capabilities, not rely on the user's accommodation of its shortcomings. I'll accept excuses for products that are price performers, but those who wish to claim they're at the top of the pile must not only provide superior performance and gee-whiz features, but a superior user experience. Delivering the GS3's performance surely required lots of "out of the box" thinking; I assume the decision to start with GS1/GS2 components led to the "in the box" issues I'm quibbling about.

Does this look familiar? Over thirty years since its introduction and nobody could improve on its layout? Right...


Photos of 1970ish GS1 courtesy of elysian (flickr)


Paul Pratt pulls shots from his restored GS1 at the SCAA conference in Charlotte (flickr)

AndyS wrote:Bill Crossland said that the GS3's brain board has the extra capacity required for pressure profiling control. But since the number of people who would be willing to pay the upcharges involved are low, it may be a while before that happens. Unless...researchers find it is useful and customers start to demand it. LM listens to customers.

If you want to experience a vast universe of dashed hopes, take a survey of all the unused I/O ports on CPUs and unimplemented functions buried in the code of software packages. If sensors and actuators had followed the same price/performance path as processing, we'd all be robots by now.

another_jim wrote:If you want to experience a vast universe of dashed hopes, take a survey of all the unused I/O ports on CPUs and unimplemented functions buried in the code of software packages. If sensors and actuators had followed the same price/performance path as processing, we'd all be robots by now.

True, no doubt, but Bill C has said, "We are working on the controls and software for controlling pressure profiles." So in the end, this particular hope may not be dashed. Maybe.