I agree with everything in the above post, with the following caveats.

One of the Thummer’s design goals was to minimize weight and bulk, so that it could be conveniently strapped to one's forearm and waved through space to activate internal motion sensors as expressive controls. If your instrument's weight is all the way at the end of your arm, and you have to accelerate and decelerate it frequently, its weight (mass) matters a LOT. Weight and bulk are also the leading contributors to shipping costs, which matter if an instrument is being sold primarily online.

Likewise, if an instrument is going to be strapped to a forearm, then that arm's hand can't move freely across the keyboard's surface via rotation of the elbow or shoulder; only finger and wrist movement is possible. That means that the keyboard layout has to be optimized to put as many tonally-important intervals within the hand's fixed span as possible. This affects the choice of note-layout, the size/spacing/shape of the buttons, the number of octaves, and the number of notes per octave. It also biases the keyboard towards an interval-based system, rather than a pitch-based system, so that the notes of the current key are always closest to the center of the keyboard. NB, the hand of the non-strapped arm CAN move freely across the instrument. Which hand should be free, and which one strapped, depends on the way the note-layout groups the tonality’s melodic and harmonic intervals. On the Wicki note-layout, it is more convenient to play harmonic structures with the unstrapped hand, whereas the playing of melodic structures is convenient even with the strapped hand, so a Wicki-based jammer should b strapped to the melodic hand, all else being equal. Generally speaking, the melodic hand should be the dominant hand; that is, the right hand for right-handers, and the left hand for lefties. For left-handers, being able to play melody with their dominant hand is a significant advantage for the jammer. (Now, many lefties, trained from birth on the piano keyboard, are likely to respond by saying “being forced to play melody with my non-dominant hand never slowed me down any;” I say, don’t kid yourself. As a child, my father was required to place his left arm behind his back whenever writing in class, to force him to learn to write right-handed “like everyone else.” Was he able to learn to write this way? Of course. But it wasn’t natural, and it put him at a distinct disadvantage relative to his classmates. The design of human tools should embrace diversity, not enforce conformity. Doing so is not “just” egalitarian; it also accelerates and simplifies learning, by leveraging the way diverse people’s minds and bodies already work before music lessons begin.

Button travel is also affected by button-density. Imagine that you're stretching your finger out flat to reach a button that's far away from its hand's palm. When that finger's tip presses down on that button, the rest of the finger is moving downward, too, with the tip-ward portion of the finger travelling the full length of the button travel, the middle portion travelling about half that, and the palm-ward portion travelling very little. If the buttons are densely packed, then the tip-ward portion of the finger may unavoidably press a palm-wardly-adjacent button downward, too. To avoid this, either the buttons need to be spaced further apart, or the buttons need to travel less. At Thumtronics, we looked at the button-travel of a range of acoustic instruments (clarinets, flutes, recorders, trumpets, etc.) and found a huge range, dictated more by the mechanism of the instrument rather than by any ergonomic considerations. It's not at all clear that there's any universal ergonomic optimum, independent of instrument mechanism. So we chose a button-travel of about 3mm, which was consistent with many woodwind instruments’ button travel.

Sensing key velocity is not essential to expressiveness. The trumpet, flute, clarinet, saxophone, etc. are all quite expressive without sensing the velocity with which their respective keys are struck. They get their expressive power through means other than key velocity. Key velocity is essential to the expressiveness of the piano keyboard because it is the *only* expressive variable that the piano has (a couple of binary foot-pedals aside). By providing other expressive controls, such as internal motion sensors and thumb-operated joysticks, a jammer can be much more expressive than a piano even if it does *not* sense key velocity.

Let me say that again: sensing key velocity is NOT essential to a jammer’s expressiveness. If Thumtronics had taken this into account sooner, Thummers would be on the market today. The pursuit of key velocity and polyphonic aftertouch were expensive and pointless diversions.

Measuring a button’s velocity is difficult. Electronic pianos do it by having two separate contacts on a piano key, and measuring the time between the first and second contact. This approach works because piano keys are levers; buttons, as on a jammer, are not levers. The two-stage measurement approach to deriving key velocity is not available to jammers. Instead, jammers need to measure continuous pressure, and derive velocity from pressure. This is computationally expensive; it makes the physical buttons expensive; and it strongly affects the feel of the buttons. None of these are good side-effects. It is better, then, for a jammer to have simple on-off buttons. AFTER those are working and selling well into the market, then it would be nice to add non-button sensors that measure the pressure on each keyboard – what’s called “channel pressure” in the MIDI world. (Roger Linn suggested this to me, BTW.) Having just a few of these pressure sensors per keyboard dramatically reduces the cost and complexity of the instrument (relative to per-key pressure sensing), while still giving the performer a significant degree of key-velocity-like expressive power.

Regarding foot-pedals…given that the jammer is a digital (MIDI/OSC) controller, it can work in cooperation with a number of other digital controllers to affect the current sound. For example, I see no reason to incorporate foot pedal jacks into a jammer. The signals from the pedals would pass right through the jammer, unaffected, anyway. It would be better to have the foot pedals—or bagpipe bags, or elbow angle sensors, or sphincter-pressure gauges, or whatever—implemented as individual and independent control devices. The ideal of cables connecting foot pedals to a jammer—which could be in constant and complex motion—fills me with horror. If the pedals communicate wirelessly, then they can communicate with the synthesizer directly, rather than communicating with the synth via the jammer. The jammer’s thumb-operated joysticks and internal motion sensors are different from the proposed foot pedals; they must be integrated into the jammer. Foot pedals, on the other hand, are inherently external to the jammer.