> Give a large grant, wait a long time, and then call in the note
And what are the consequences if the scientist has pittered that money away? How do you decide if a scientist/manager is even deserving of that level of confidence in the fist place
However... A flaw in Thomson's idea is that some research is expensive, costing much more than leisure time.
Another approach is angel-investing / business-loan style: Give a large grant, wait a long time, and then call in the note -- to return research that justifies the investment, or repayment with interest.
Or X-prize style: Give outsize awards for outsize accomplishments, and let the researches take on the risk.