Not in any way shape or form. It's an analogy.
Shoddy design is shoddy design.
Not at all. A designer could be saying, "hey, the DM is the referee and if you cannot trust him to be fair that's not a game design problem, it's a personality problem."
Either way it's insulting and condescending, and an excuse for shoddy design.
Speaking from personal experience, I find more work in tedious columns of rules rather than a quick explanation of what someone is trying to do and quickly adjudicating the probability and letting the action move on.
You may enjoy the extra work, you may be good at it, but you're still doing off the cuff design when you do that.
Truth is, I run that way, myself, frequently. I have a talent for it, and it's easier than learning the minutiae of a not-so-good game that I'd just have to heavily modify, anyway. It's also a great technique to fall back on when the rules fail you. But, it's what I'd have to consider an 'advanced' technique, it's not for everybody, and, taken to the extreme, it's freestyle RP, at which point you need few if any rules, anyway.
it's easier to make a rules light(er) system and add complexity with modules for just the type of game you are describing, than it is to build a rules heavy core and remove stuff and still keep it balanced.
I think what we're both casting about for here is an 'elegant' rule system. Not rules heavy, not incomplete, but spare and efficient and intuitive. It's a great ideal to strive for. The 5e playtest is definitely not there, nor even trying to get there, though. It's 'rules' light by excision of complexity, not be reducing the need for complexity.
Adding to is always easier than taking away if you don't want to break a game.
I think of it as breaking the game is always easier than fixing it. But, adding is not that easy, because it's both a creative and a design task. You have to dream up the stuff you're going to add, and balance it. Banning undesireable things - taking away what you don't want - can be quite easy, if the game is robust enough and modular enough.
And, that's a point, too, because a modular game is both a simple game (the core, with no modules) that you add to, and a complex game (using all modules) that you deduct from. To be really good in either case, it needs to be designed and balanced as the complete game, and the potential combinations of modules accounted for - a very substantial undertaking for the designers. I can't see doing it piecemeal (publish a simple system and add complexity to it after the fact) as a good alternative. But then, that's what playtesting is for.