The conversation has moved on from this, but allow me to address this with an analogy.
I have invented bitping. I provide a vat of 1 million ping pong balls. 10,000 are red, the rest are white. Only the red are valuable. To mine for bitpings, you reach into the vat blindly and pull a ball out. If it is red it is yours to keep, if white, you toss the ball back in the vat.
So, at first the mining goes fast - roughly every 100th try yields a red ball. But as time goes on the # of red balls declines, while the # of white balls stays the same. Before long it takes 1000 tries before you get a red ball. After awhile it takes 2000 tries. Eventually, when the number of red balls is very low, it takes close to a million tries to get a red ball. Finally, no matter how many times you reach into that vat, you will never pull out more than 10,000 balls.
This is an analogy: of course it would be easy to manufacture a red ball behind the scenes. That is where the analogy fails.
bitcoin is a numerical algorithm that performs a seach in a defined numerical space. There is no known way to 'instantly' generate 1,000,000 answers in secret. Quick, find me the 1,345,0454,3453,455,545,389,432th prime! Quick! What's taking you so long? I gave you the algorithm to test if a # is the nth prime (let's pretend),so why can't you instantly tell what the 1,345,0454,3453,455,545,389,432th prime is? Because a test for validation is often completely different than generating the result to be tested.
That is not the bitcoin algorithm either, but you see my point I hope? It's a mathematical search in a space that no one knows how to mathematically cheat. Well, it's related to cryptography, so maybe the NSA does, but no one outside of the super secret part of cryptography does. It's costly to compute, but cheap to check for correctness.
Make sense?