I don't know if the alternatives would be better - I don't believe patents are bad in every case. But I do know that if there is one place where the benefits from patents seem to outweigh the burdens, it is with drugs. At least, as long as politicians don't put grandstanding ahead of saving lives.
What I think this would do is create a powerful incentive to hocus one's R&D figures. In the drug industry, each extra year of patent protection can well be worth hundreds of millions of dollars in revenue, which is an incentive indeed (with hot fudge and extra sprinkles on it.) It's not like our drugs come cheaply to us now, but I can see that for an anticipated blockbuster, ways would be found to ensure that the patent racked up the longest possible life. You already see that happening at the back end of patent terms, as the constant battles with generic companies illustrate. Every possible reason is trotted out for why a given patent is invalid, and every possible means to preserve or extend the patent's term is invoked in response.
I am inclined to agree with Tabarrok that "one size fits all" is a poor approach to intellectual property. There is no clear-cut economic solution, because of the familiar problem that the activity that generates cost (research) differs from the activity from which one can extract revenue (sales of a product or service). My thoughts (which are not developed in any depth) are in this essay.
For Discussion. How could you design a system that varied the length of patent according to R&D input in a way that would be less likely to be gamed?