Designers loved Flash because it was WYSIWYG. It's also why PDF lives. You usually only had to test it on one computer, not different versions of different browser brands. The client-side "auto-flow" of HTML browser drives designers batty. I can relate: I make CRUD GUI's and different browser variations butcher their placement to hell and back, requiring me playing Whack-A-Mole to fix it. The online forums say, "Shut up and learn layout rocket science!" A few snotty Sheldons mastering auto-flow math doesn't scale well.
The browser should have a WYSIWYG mode for when your application needs it. The server could re-scale as needed for different devices based on a layout engine of the designer's choosing. That way one is dealing with one layout engine instead of say 50 (browser brand/version combos). Seems more logical to me. We don't need auto-flow rocket science if we just partition layout engines logically.
I believe Flash and Java Applets both made the same fatal mistake: tried to be an entire virtual OS. That made them too complicated, which invites hackers. If they had focused on just UI "serving", they could have been small enough to be manageable.
WYSIWYG mode, unfortunately, ignores the needs of screen readers and other methods of accessing content. The risk of WYSIWYG tools is they lie to you; not all your users can actually see what you see. And that's to say nothing of the less exotic scenarios like different screen sizes or devices.
... unless you're working in a space allowed to discriminate against users that need assistive access to your content of course, in which case you can spring for the WYSIWYG dream.
One should not go using terms like "discriminate" against people who are clearly not doing so.
Assistive access to content is an afterthought for almost everyone in software, and that's perfectly fine. It's an extremely uncommon and unnecessary use case for nearly the entire device-using population.
So by that logic we shouldn't bother with wheelchair ramps because nearly the entire population has working legs. Nor should we bother with Braille on signs, since nearly the entire population can see just fine.
Any software professional knows just as well as I that these concerns are specialized and people with a particular interest in complying with these special use cases are paid for their particular interest and capability in worrying about, and complying with, regulations within the realm of assistive access.
Most software developers think about getting their software to work correctly, not about edge-cases dealing with the UI for tiny, specific disenfranchised groups of people.
EDIT: Even if we were just to stick to the US, only 2.7 million people use a wheelchair here. That's 0.9 percent of the US population. That's really, REALLY freakin' specialized. Don't stoop to heaving virtue my way when the facts hurt.
EDIT#2: Perhaps you can quantify your ridiculous statement by telling me how the developer of this [0] application should be punished for not building assistive access into his GPLv2 project, and how he should have & could have actually done so.