Commodore 16
Lets's compare the C16 to its intended competition:-
Mattel Aquarius:- 4k RAM, 38x24 text in 2 of 16 colors, no bitmap graphics, 1 bit sound, no joystick port, no disk drive port, horrible rubber keyboard. At least they had the sense to include a full expansion port, but apart from that it was dire. Amazing that a famous company like Mattel would put their name on such a POS.
Timex Sinclair 1500:- effectively a ZX81 in ZX Spectrum style case with rubber keyboard and 16k RAM. Monochrome 32x24 text, no bitmap graphics, no sound, no disk drive port, CPU runs at ~1/4 speed when the display is active. Not surprising that it was discontinued only one year after launch.
Ti-99/4A:- 16 bit CPU based on minicomputer architecture, same video chip as MSX with 2 of 16 color 8x8 tiles making 256x192 bitmap graphics, plus sprites, 3 channel sound chip, more than 16k of RAM, cartridge and expansion ports, mechanical keyboard, joystick port - and a tragically flawed design. You see, that 'more than 16k' of RAM was actually 16k of video RAM behind the VDP, plus a miserable 256 bytes of main memory (which acted like the zero page does in 6502 machines). This made loading machine code games from tape impractical without an expensive RAM expansion. To make matters worse the BASIC ROM was accessed serially, so it was very slow.
Of the 3 intended competitors the Ti-99/4A was the only one which could better the C16 (for cartridge games only, forget about tape). However it was only 'competitive' because Ti slashed the price below cost in order to match the VIC20.
Wikipedia continues...
The C116 had a horrible chiclet rubber keyboard. Only 51,000 units were sold. If that was Tramiel's vision...
One could argue that Irving Gould should have axed the C16, but I for one am glad he didn't. It was a much better machine to learn on than others in its class, and many people got their introduction to programming from it. And it's different enough from the C64 to be interesting to retro computing enthusiasts.
We will never know if Trameil could have pulled it off, but...
(Wikipedia again)
I suspect a 'Commodore 264' for $80 was unachievable, even if Tramiel hadn't gutted Commodore on his departure. The main reason the C16 'failed' was that the competition failed earlier, so it had nothing to compete against. It was then seen as wanting compared to the more expensive but more capable C64. The same situation applied to the C65, which Gould
did axe (a pity, but the 'right' decision as it would have failed too - just like the
Amstrad CPC Plus range failed. by 1991 the 8 bit home computing era was over).
Compared to other machines in its class, the C16's 16k of RAM was a lot. If that wasn't enough then it could be upgraded to 64k via the RAM expansion port, making it equivalent to the Plus4. Note that the ZX Spectrum originally only had 16k RAM, as did the Atari 600XL.
I think the C16 was a good machine for the price, but rather than being taken on its own merits it was seen as a cut down C64 with 'essential' features missing and an incompatible BASIC. Not being able to play C64 games made the C16 a non-starter for C64 fans. They didn't care that it was faster, or had a better BASIC, or more colors. It had to be both cheaper
and highly compatible.