(Various hates on the VIC-20...)
Kind of like finding the cure for a deadly form of cancer 30 years AFTER you get it, doesn't do you much good does it? There are plenty of systems that launched with one good title that ended up dead because of the lack of a good selection of software.
No one really remembers, given how the VIC-20 was so completely eclipsed by the C64 in popularity, but the VIC-20 had a perfectly good software library at the time. Commodore positioned the unexpanded VIC-20 essentially as an alternative to cartridge game machines like the Atari 2600 that could also teach basic computer literacy, and in that light it was a great bargain. It had a good selection of cartridge games, a decent BASIC for such a cheap machine, and with a few peripherals it could actually do some "real computer" things. Many people got their first taste of online computing thanks to the inexpensive VICmodem, for instance.
Yes, the VIC-20 looks like a toy today but a 5k computer for $299 was a heck of a bargain in 1980. About the only thing substantially cheaper than the VIC-20 on the market at the time was the Sinclair ZX-80 which cost almost $200 US and came with *ONE KILOBYTE* of RAM; A 4K TRS-80 CoCo was $399 and an 8K Atari 400 was around $500. The VIC-20 had a better keyboard than any of them, a much better BASIC than the ZX-80, and it was roughly as expandable as its more expensive rivals. For what the VIC was built for it was a success. It didn't age well (in particular its nasty 22 column screen format doomed it to be replaced en-mass when the C64 came out) but at the time it filled its niche competently. Linus Torvalds' first computer was a VIC-20 so clearly it was possible to learn *something* playing with one.
As for the TED machines: The machines themselves were capable enough in a vacuum. The graphics and sound hardware was inferior to the C64 and Atari 800-class machines (mainly because of the lack of sprites), but measured against the Sinclair ZX Spectrum, which is what Jack Tramiel was targeting with them, they look pretty good. But nothing exists in a vacuum, and by this time Commodore was clearly having "direction issues". Commodore over the course of about seven years introduced no less than five vaguely-sorta-BASIC-compatible-but-hardware-incompatible 6502 computer lines; PET/CBM, VIC-20, C64, CBM-II, and the TED line. (This isn't even counting stillborns like the TOI/Color PET.) It's really bizarre how they kept reinventing the same wheel over and over again. Commodore machines rarely evolved along a sensible "line of ascent". It seems like the engineers would cook up a random design, drop it on the market, and then simply sell it until it was played out. Outside of the C-128 it's hard to think of a case where Commodore really produced an 8-bit "successor machine" that was fully, or even very, compatible with a previous model.
(Maybe you could call the various generations of PET Commodore's best example of an evolving product line; probably a better example than the C-128, since the 128 was sort of a mishmash of random features grafted onto the C64 that couldn't really be used without switching to an incompatible new mode. There's more of a continuum between the various vintage PETs and Commodore did offer things like BASIC upgrades for older models.)
One could argue that compatibility wasn't that important if the new machine were sufficiently advanced over the previous model (take the VIC-20 vs. C64), but the TED failed that test because by the time it premiered everyone was comparing it to the C64 instead of the VIC-20. If they'd come out in 1982, possibly as late as 1983, *maybe* the 16K TEDs would of made sense. (At the cost of potentially fragmenting the VIC-20 upgrade market between them and the C64.) But the Plus/4... no. It's really hard to understand what they were thinking as a 1984 introduction. With Commodore's vertical manufacturing the RAM was the most expensive part of the C64. Thus if you put the TED in a 64k machine you've pretty much blown the entire justification for the TED chip in the first place.
If Commodore had taken better care of its designers and engineers they really might of gone somewhere leveraging their manufacturing power, but as it was they sort of turned into a home computer puppy mill. A very successful one, thanks to the singular success of the C64, but still ultimately a dead-end road.