They are absolutely here to stay in consumer products. Whether the job market will be good and constant over the next 25 years is something that I can't say, but they're likely not going away, and to the extent they do, they'll likely be replaced with something so similar that those skilled in the art will still be able to adapt.
I wonder if compilation for Harvard architectures is any more or less difficult than for Von Neumann. I'm guessing there are pros and cons with both.
Thanks for mentioning that they're here for a good bit yet, I'll definitely take a look at them.
These parts are sold by a lot of vendors but you can get everything you need from Digikey [2]. And when you're trying to build a consumer device, it's nice to buy a microcontroller that is inexpensive - you can get this [3] one for $0.38 if you buy 3000 at a time. We normally started with flash parts and committed to OTP once our firmware was solid.
[2] http://www.digikey.com/product-search/en?keywords=microchip
[3] http://www.digikey.com/product-detail/en/microchip-technolog...
I didn't think about environmental restrictions though, that's really interesting to take into account, huh.
Looks like the single-unit (or handful-of-units) price is $1.14 for the PIC10F202T, which is very nice considering that Digikey is in a bit of a different class to the Chinese OEM Arduino clones you'll find on eBay for $2-$3.
I am curious though - I remember reading in one of my old phonebook-sized product catalogs that certain variants of the PIC line were not recommended for new designs. Does the PIC10F202T fit into that category? I can't see any indication for or against on the webpage, nor any links that look like they'd reveal that info. (I ask because my memory of exactly what was marked legacy is poor, and I remember that some of the PIC10 range was listed as such, but I don't recall exactly which. I'm also curious what this marking means exactly - whether the chip will be phased out at some point, or what.)
How much of a deployed base - and perhaps more importantly, how much future demand - is out there for this sort of thing? Would it be worth it for me (as a random person on the Internet, not OP) to spend the next N years learning these microcontrollers' instruction sets, errata and idiosyncrasies?
Like OP, I'm incredibly interested in assembly language, and "really tiny microcontrollers!" is one of the instinctive responses I also gave in my own answer, but I'm not sure how widespread demand for it is, so I was somewhat more cautious.