Evidence Based Policy-making

Newspapers, TV journalists and other commentators love political u-turns. A big change of policy by a prominent politician, ideally after a long and costly defence of said policy by said politician, makes a great headline (or, if you're lucky, several great headlines, followed by a resignation). The media as a body seem to love criticising politicians for the policies they promote and then, if they actually persuade someone to change their mind, they attack again over the apparent u-turn.

Let's be completely clear. This practice is utterly stupid and it acts against our interests as a society and as individuals. It deters politicians from changing their minds even when it becomes clear that their ideas were wrong; the result is a political system where politicians can't admit mistakes for fear of public humiliation and retribution. Politicians are forced, if they wish to retain their position, to advance either very safe policies with no possible risk of failure or, if they advance grander schemes, to pursue them even when it becomes clear they will not deliver the anticipated benefits. 

This is the precise opposite of what should actually happen. We want evidence-based policies that, whoever has suggested them, can be tested and proven before being rolled out, sometimes at enormous expense, across the nation. I see it working like this:

1. Politicians suggest their policies based on their individual expertise and ideology, just as they do at the moment. They use their reputation and position to describe the advantages, costs and risks of their Big Idea to a grateful nation.
2. Civil servants, having (ideally) rectified the most obvious flaws and failings before the initial policy announcement, then design scientifically valid trials to determine whether or not the policy will actually deliver the intended effect. In discussion with the politicians, civil servants also define the criteria by which success of the policy will be measured.
3. Parliament then enacts legislation to begin a limited trial, as designed by the civil servants, to test the policy and prove its effectiveness.
4. The National Audit Office, or some other independent body, gathers and evaluates the evidence from the trial and, using the pre-agreed criteria for success, announces the success (or otherwise) of the policy. 
5. Finally, if the trials suggest the policy will work, the politicians return to Parliament to begin a nationwide roll-out, claim the glory for their initial insight and generally get on with whatever self-congratulatory activities they deem appropriate. Alternatively, if the trials suggest the policy will not deliver the anticipated benefits, it can be quietly shelved, without undue embarrassment, and everyone can get on with their lives.

This system would give us better outcomes (because only proven policies would be implemented) while encouraging cross-party support for successful legislation (because nobody wants to argue against independently verified evidence) without ever preventing politicians from attacking each their opponent's ideas, just as they do at the moment. It would be a first step to a better, more rational, legislative process.