> You can literally try to clone any YC company after demo day.
Not quite true. A growing number are hardware/robotics or biotech or some other field that require significant capital investment and highly specialist Phd-level expertise to even get a proof-of-concept off the ground, so you can't just "open your favorite editor" and get started.
Then you have those which cover a niche that again requires some kind of insider knowledge or connections (it helped for example that Facebook was created in Harvard). That's not to say these are impossible barriers, but again they're not broken by hacking away at node or PHP.
The pool of ideas that a) you, personally, can execute on today and b) have a hope of generating profit or funding within a reasonable timeframe is not as big as people like to think. Much of the low-hanging fruit of profitable ideas that can be executed by the single coder in the proverbial basement has been picked long ago.