New users drive the rewards per person down. New investors drive the rewards per person up. My expectation is that new users seeking rewards will eventually outweigh new investors, so the likelihood and difficulty of earning rewards will follow a long term curve something like the bitcoin mining difficulty.
As a result, the first scenario might work today, but I'm not to sure that it will still work 5 or 10 years from now. I don't think a company can count on it for a long term business strategy.
Your second scenario isn't so different from mainstream stock pickers. How many people, for example, are buying Jim Kramer's picks before he releases them to the public? The main difference is probably that the cryptocurrency market is smaller, so more easily influenced, but as it grows, that will change.