b) Architectural changes that will require hardfork better to implement before launch
I largely disagree on this point, and I think it is fairly important.
It is better to be able to manage hard forks effectively with thoughtful design, good communication, broad buy-in, and successful execution (many of which ultimately reinforce the others). If hard forks become a "torment" (or worse are viewed as something that can't be done at all) this is a symptom of a much more serious problem with the development process and community that indicate paralysis, stagnation, and, likely, death.
Not that there was anything wrong with launching a CryptoNote fork with a different blockchain design from the start though (separate question).
There is a likely point of extreme stability, maturity, and near maximum adoption where further hard forks become both impractical and unnecessary. But as long as coin is young and in active development you had better plan for hard forks or you can't adapt and will likely be left behind. No one can make every important decision in advance.
This is theoretical though, as Monero hasn't delivered any hard forks yet. So we will see. If we fail to pull them off I think we will be in trouble.
I understand what you mean, it's extremly important to be able to fix issues even if it need's hardfork. And this is exactly why i implemented alerts - that make users aware of critical updates asap.
But i still very disagree with you "disagreement".
Hard fork is very painful especcialy for big network - painful and slow process, since you need to wait until about 95% of network gonna update sortware. And unfortunately you don't have any reasonable way to force them to do that.
And i talk now about the case when everything is okay with community. The worse case will be if some part of comunity(or some silent miners) will disagree with hardfork changes and they would try to ingnore that - this may do hardfork impossible at all.
I have now doubt that you aware of how difficult it would be to make a hardfork for bitcoin now(if they want), and how much time it will take.
I meant it would be perfect for developers to be able make hardforks easily when they want - but it has some contradictions with crypto-currency nature (in my vision): it's a basement as a set of known rules and technologies. And when you change a basement it's very risky to ruin the whole construction.
So my point that hardfork is a bad case for currency, especcialy for widespread currency.