Respectfully, it is NOT relative to my knowledge or usage. If a simple device does not *require* including a micro to be done as simply as possible, then what is the point of adding one?

If you want to make several thousand TOH, you can absorb the cost of software development (good luck finding the volume of customers to make the economics hang together). If you want to make 1 of something, then writing two sets of software, learning yet another new device, sdk and environment, is an overhead I can't afford. I will simply use that rpi or STM32 directly and leave the 'chenphone' un-purchased.

Yes, I could design in your STM32F0, I already use them in several of my commercial products. It's just not going to happen, it's unnecessary for simple devices and I don't see enough of a reason to invest my time on complex ones.

Within some huge error bars, component costs are largely only relevant in higher volume products. Development time (hardware and software) is similarly only relevant in low volume ones.

Chenphone would be useful to me as a simple(ish) portable linux box with decent connectivity, useful for prototyping and one-off sensors/controllers.
For anything commercial (ie anything I might sell as a product) I wouldn't start from here.

As you suggest, if there's an (ideally otherwise uncommited) i2c bus available, then providing landings for pogo pins is nearly free, it's just a mechanical consideration. The cost of providing connection on the TOH side is irrelevant. Likewise for an interrupt pin.

Frankly, I would probably buy Chens Lauta concept sight unseen just for the 2017 N900-ness of it, regardless of the wired connectivity. No point wishing for features I never stated a case for though, is there?