"there's an incentive for readers to withhold more valuable information to others (instead of spreading them and making money with Google Adwords)"

There are two problems. The first one is that you make an assumption here, that individuals choose what's in the best interest of the group (if it's only about the individual, the Adwords money might outweigh the information value...). Secondly, even if the assumption holds and people will behave according to this rationale, then still it takes only two or three odd, less-rational individuals to redistribute the information, thereby devaluating it back to free.

That doesn't mean your model won't work. It seems it will (and does already) work very well for something like The Economist - perhaps it simply requires a certain type of readership...