It's because it's just plain hard. Simplicity and Flexibility often competing pulls. To make something that is both very very simple, and so flexible as to be used to create solutions to real unpredictable problems… it's legitimately hard.

But the driving principle of programming language evolution is increased simplicity with flexibility. PHP is a lot more accessible than, say, writing assembly language on punchcards as a programmer in the 1960s did. Programmers too want to make things simpler while still flexible, because that's a lot more powerful a tool and allows the programmer to do a lot more in the same amount of time.

I predict that even with the most accessible tool though, doing complicated things will take the kind of analytical logical thinking that good programmers (not all programmers are good programmers) specialize in, and that not everyone has 'naturally'. Good programmers don't worry about being put out of a job due to easier to use tools, because they know easier to use tools might make current 'complicated' problems easier, but will put current 'infeasible' problems into the 'complicated' category and people will still want to tackle those. (Imagine trying to write Windows 2000 in assembley language on punchcards! Impossible.)

And I would have thought that if end users wanted to build apps, they'd want to build them for the phone they have rather than for a particular OS.

I guess that now that people don't have to constantly tinker with their cars to make them work (thereby rendering pointless the incessant and useless automotive debates over whether Chevy, Mopar, or Ford is best), the consumer now is captivated by the fascinating world of handset OSes.

Some of these have been somewhat succesful, although usually either by ending up tricking the developer into becoming a programmer after all (Hypercard), or by limiting themselves to a very specific domain (Lego Robolab; perhaps Yahoo Pipes fits in that category too?).

It's an interesting thing to try to do, create a software construction environment for non-programmers, glad to see efforts in that direction. But whether it ends up finding a userbase or not, I'd bet the farm that it will NOT change “Programming as We Know It,” anymore than the others did. (Although some people might argue that Hypercard _did_, heh.)

The intended audience for AppInventor- the end users – don’t need or want it either.

On what do you base this assertion? I'm an Android user (though more developer-oriented than most), and I'd really like to able to have a quick way to interact with a web service or two. Not sure if AppInventor will scratch that itch, but it's not to say it's not there.