They're not THAT rare... even NiB they crop up fairly regularly; they're just so limited, so proprietary, have such a small library of programs that they aren't desirable, even as a collectable. About the ONLY time it was popular was in the pre-arduino days because the board was small and you could modify it to run off a square 9v.
The MC-10 was part of what is commonly called the "race to the bottom" -- which actually screwed over many other companies. Simple fact is the Sinclairs and VIC-20's low price put a fear into every other maker sending them looking to make low cost versions of their systems not taking into account that hardware just kept getting cheaper and cheaper. Worse, some companies (IBM, Commodore and Tandy stick out for this) didn't want these cheap models to compete with their mainstream editions and as such neutered the devices in ways that made them at best a novelty, at worst doomed to failure within a year of launch. For companies like Sinclair it worked becuase it's all they offerred, for companies like Radio Shack/Tandy or Commodore, it blew up in their face.
Marketing execs were all in a panic about things like the ZX-81 and TS1K, particularly as they were starting to show up at sub $100 prices in the glass case behind registers at chain pharmacies like CVS or Walgreens. The real shot to the gut being the 1983 price of $50 a pop.
Really the MC-10 is a poster child for this, in particular how the changes in manufacturing and cost of components doomed it from the start in a way I don't think it's creators could have foreseen.
A simple browse of the main radio shack computer catalogs from 1983 to 1986 tells the story...
http://www.radioshackcatalogs.com/computer.html
The MC-10 wasn't even listed until the main 1984 catalog, but let's start TWO catalogs before that. This was while the MC-10 was in development.
Code:
1983 RSC-8
----------
16k Standard Coco $400
16k Extended Basic Coco $500
32k Extended Basic Coco $650
The MC-10 was designed to enter the market at below $200 with the memory pack being $50... seemed like a good idea; but move forward just six months to the catalog right before the MC-10's release:
Code:
1983 RSC-9
-----------
16k Standard Coco $300
16k Extended Basic Coco $400
32k Extended Basic Coco $550
... and they all had their prices slashed a hundred bucks in just six months! This was partly due to cheaper design, but also because they were trying to unload the original Coco as at the same time they were prepping the MC-10 for launch, the Coco 2 was being slated for release... in the same catalogue and shipping schedule. I'm going to include the memory pack in the price to make it a more fair comparison by the time we get to 1985.
Code:
1984 RSC-10
-----------
2k MC-10 + 16k RAM expansion $170 ($120 + $50)
16k Standard Coco 2 $240
16k Extended Basic Coco 2 $320
64k Extended Basic Coco 2 $400
We can see the Coco 2 knocked $60 off the price of the original, had access to the full existing coco software library, a better keyboard, and was in the 16k configuration just $70 more than a MC-10 with the memory pack.
Really though that gap narrowed in the months that followed:
Code:
1984 RSC-11
-----------
2k MC-10 + 16K RAM Expansion $130 ($80 + $50)
16k Standard Coco 2 $160
16k Extended Basic Coco 2 $200
64k Extended Basic Coco 2 $260
That's some heavy duty price slashing, the 16k Coco 2 costing ten bucks less than a MC-10 plus expansion pack was introduced at!
Simple fact was the MC-10 wasn't selling, and to keep it from competing with the Coco 2 they'd have had to slash so far into it's profit margin it simply wasn't viable which is why it was killed off after less than a year of being on shelves. It reached a point where while it was never advertised in the catalogues, it was slashed down to $40 for the unit and $25 for the memory pack at the end. You look at the next catalogue, RSC-12, you can see why.
Code:
1985 RSC-12
-----------
16k Standard Coco 2 $120
16k Extended Basic Coco 2 $160
64k Extended Basic Coco 2 $260
In the span of two years we had gone from a 16k Coco 1 being $400 to a 16k Coco 2 costing less than the MC 10's introductory price with the memory pack!
In-house the numbers were worse, simple fact was that from about 1983 onward the sales of 16k models of ANYTHING up and stopped for Tandy. The C64 can be partly blamed for this as could the plummeting memory costs. 16 or 32k may have been hot shit in 1977, but by 1984 it was just plain shit -- particularly with 16 bit wading into the fray like a 500 pound gorilla leaving 8 bit looking like monkeys at the zoo. STILL many marketing execs at companies refused to acknowledge this which is why the Shack continued to offer a 16k version of the Coco 2, at least on PAPER, up through 1987.
Like a lot of products the 'race to the bottom' never ends well as there becomes a point at which you cut too many corners. In computing this is far more pronounced as the prices are ALWAYS in freefall as new technologies drive out old ones, and as new manufacturing methods allow larger and larger orders to be quickly filled.
This is even more the case when "competing with yourself" -- where instead of replacing a product with a cheaper one you try to go both directions at once. The PCJr, MC-10, Commodore 16 and Minus-60 -- all stunning examples of this. They simply crippled them too much for people to choose them over the mainstream product, despite the lower cost.
Or in the case of the Commodore Minus-60, comparable cost that despite gaining some really cool features like faster CPU, better colour palette capabilities and higher resolutions weren't worth losing things like hardware sprites, quality sound and software compatibility.
You can see how competing with themselves was a limiting factor when the same designs are released in places where there is no such competition -- the Matra Alice is basically a MC-10 and did a thriving business in France and several South American countries -- since it wasn't up against a more powerful system spitting distance in price from the same manufacturer in those markets.
The final nail in the MC-10's coffin though was that by the time it was in stores, Timex had discontinued the TS-1K in north America and Sinclair had few if any legitimate plans to continue in said market -- so the competitor it was designed to face had already left the scene.