These are judgments, not facts, so they are set out here once rather than argued again in every piece. If a verdict seems to break one of them, that is worth pointing out, and the corrections page is where a change would be logged.
Say where the money comes from
Every spending promise is paid for by higher taxes, lower spending elsewhere or more borrowing. Every tax cut is too. A politician who rules one of those out should say which of the other two they would lean on. I hold every party to this, and a promise that skips the question is judged weaker for it.
Cost every proposal
When I say what I would do instead, I say how it would work, what it would cost, who it would affect and what could go wrong. If something can't be costed yet, I say what the cost depends on. A proposal with no downside isn't a real proposal.
Name the loser
Almost every policy leaves someone worse off. Saying who is part of being honest about it. A plan that claims to hurt nobody usually means the cost has been moved somewhere harder to see, such as a frozen tax threshold or a levy on bills.
Check what has already happened
Before proposing anything, I check whether it has been announced, tried or already done. Many good ideas already exist in a past Budget, a consultation or another country. Building on them is better than presenting them as new.
Prefer rules and decisions to one-off money
A grant makes a good announcement, but a rule change or a published decision usually lasts longer. I look for the mechanism: who does what, by when, using which power.
Show the working
Governments should publish the reasoning behind big decisions, not just the decisions. That includes the numbers they relied on and the options they turned down. Uncertainty created by silence is a cost, and I count it as one.
Be fair to every side
I write each side's strongest case as its supporters would. I criticise the side my verdict favours at least once. I judge the argument, not the person making it, and a claim repeated loudly is no more true than one made once.
Give consent a stake
New homes, power lines and energy projects are easier to build when the people living near them share in the benefit. I favour giving communities a direct stake over asking them to put up with the downside for the national good.
Be honest about uncertainty
When the evidence isn't in yet, I say so once and use "too early to judge" rather than guessing. Then I say which figure or event would settle it, and it goes on the track record.