The US and world economies are incredibly complicated systems, possibly as complicated as any global "natural" systems like the weather.
With that in mind, how long do you have to let a change in policy play out to see if it is a "success" or "failure"? This question is on my mind as the US campaign cycle is entering the "intense rhetoric" stage.
The primary economic legacy of the Bush administration, at least in terms of policy changes, is likely what are called the "Bush tax cuts". That restructuring of the tax code turned 10 years old recently...taken in isolation, looking at the state of the economy then and now, would anyone consider the tax cuts a "success"? (this is rhetorical, I guess...I'm sure that a large number of people who paid significantly less taxes over those 10 years would find that a success).
But there are other tweaks currently under evaluation that have had far less than 10 years to play out; namely, the Obama administration actions are being described as "failures" by several Republican and Independent candidates for the coming presidential elections. What are the metrics we (the voting public) can use to know if something like a Federal stimulus or tax cut or trade agreement is working? This part is not rhetorical - I sincerely want to know how we are supposed to conceptualize evolutions in a system as complicated as our economy.
What I have observed is that governments of any ideology tend to reach for the biggest tool in the box every time and adjustment is needed, at least when that adjustment is part of the national conversation. Tax policy is extremely powerful and creates both immediate shifts in behavior and long term drifts in outcomes. And massive nudges to monetary supply (throw stimulus or "tightening") is like the fat kid cannonball at the pool party - there are both predictable and highly unpredictable consequences that can really affect the crowd's enjoyment of said party.
If the philosopher poets were in charge, would there be a rule that said "any change in policy must come with a period of X years to study outcomes"? Wouldn't the presence of such a rule create more robust debate prior to any meaningful changes?