This is why most people advocate learning python or ruby. You don't have to deal with the underlying manipulation of the computer until you've decided you actually like programming.
And there's a certain concreteness to such a machine. The mental model has no vagueness at all. You could imagine building a physical copy of it and playing the role of CPU--adding small numbers, moving data between registers and memory, etc. I've played with this idea myself before of using an invented 16-bit machine as a teaching device.
Maybe it's just a case of personal differences, but as a starting programmer, I would have been extremely excited about a simple programmable thing that I can fully understand and use to do things within a game world. Python is cool too but it's overwhelming. There are learn-Python books that are thicker than most of my university textbooks.
Personally, I think Python is a great way to teach programming, because it allows us to get to concepts quickly. Accidental complexity is at a minimum, and the inherent complexities of programming become the focus. But, even thought that's my personal preference (based on some experience; I've used Java and C++ to teach programmers and had to explain away accidental complexities), I hesitate to say it's best because I have no evidence backing up that claim.
What's important is learning certain concepts, perhaps the most important of which is algorithmic thinking. That is, knowing what result you want, knowing how you start, and being able to reason through how to get from the start condition to your desired result, step-by-step. What helps is when the result you get is something you care about. I can easily see that for some people, getting something to behave a certain way in a videogame will be a more compelling result than others.
An MMIX like ISA would open up the world for compilers much more. With this, it looks like people should be confined to writing simple stuff - which is kinda the point but still rather sad.
Now for practical usage? Sure Assembly is probably a waste of time. I still haven't found a real way to implement that wouldn't be. But the fundamental understanding of how a computer works, how decisions are made so close to the machine enhanced my understanding about programming which in turn enhanced my understanding about how I wanted to develop and engineer, even for the web.
My .02 cents anyway.
And in my mind, "serious about programming" extends beyond building web apps.
