A set of questions for scoping a problem, six habits that show up in legislation that actually moves, and examples of both. Use them in order or steal the parts that help.
Section one
Twelve questions to ask before you write anything
These work in a notebook, in a meeting, or as prompts in any AI tool. The order matters.
01
Has anyone solved a problem like this before? What did they do, and which parts transfer?
02
Has anyone tried and failed? Where exactly did it break down?
03
What is the real bottleneck: money, law, technology, or one person's veto? Name the body that could remove it.
04
Can you describe the solution as a story? Narrating how it happens is usually easier than designing it in the abstract.
05
Who is the first person who has to say yes? Write down their name and title.
06
How could your solution fail? List the failure modes, then attach a response to each one.
07
What is one obviously good thing you can add that makes the proposal strictly better?
08
What would have to be true for this to be easy? That "if only X" is your actual to-do list.
09
What technology doubles the speed, halves the cost, or widens the reach of the current approach?
10
Can the internet do the work? Anyone on earth can reach anyone, which makes submission, review, and payment systems cheap. Let people send fixes, sort them with software, and pay for the ones that hold up. Google's six-month cybersecurity certificate is one example of training that can be delivered this way.
11
Could a federal bill solve this? Could a treaty? Who has to be in the room, and how do you reach them?
12
What would you do with an unlimited budget, and which parts of that can be done now with what exists?
Section two
Two things that are true about almost every large problem
Most problems have a workable solution. The obstacle is political will.
If a bill is good but costs an industry money, buy the opposition out. An amendment can offset the loss directly, or fund the transition. It is an ugly way to legislate, and it is how a great deal of legislation actually passes.
The technology usually arrives before the policy does.
We have vaccines for diseases that still kill millions. We can produce meat without animals. The gap is rarely invention. It is price, distribution, and someone with authority deciding to move. That is a political problem, not a scientific one.
Section three
Six habits from bills that actually move
Each one is a drafting technique, not a philosophy. The examples are real proposals on this site.
01
Getting a bill past the industry that would fight it
If a bill is good but costs an industry money, offset the loss in the bill itself. Our factory farming proposal does this: it pays farms to move to humane systems, so the profitable choice becomes the humane one.
A more formal version: a company that considered lobbying against a bill and chose not to can recover the expected value of the lobbying it gave up, if the bill passes. A firm with a 20 percent chance of blocking a $1 billion loss would collect $200 million for standing down. If a bill creates $10 billion in value and costs one firm $1 billion, paying $1.2 billion is still a good trade.
This matters because a large share of good bills die for one reason: they cost a specific company money and the benefit is spread across everyone.
Write the conditions into the text. A policy that only helps when inflation is under 3 percent should apply only when inflation is under 3 percent. Then the objection, "in some scenario this hurts," has a direct answer: in that scenario, it does not apply.
Most bills are argued about in the abstract. Conditional language turns the argument into an engineering question with a testable answer.
03
Write the exception before someone else writes it for you
Take the specific scenario where a bill goes wrong and legislate around it. Say the bill lets anyone sue a company for failing to do something clearly beneficial, and the obvious failure is a partisan jury. Add a requirement that juries be balanced. Then the objection is already handled.
Keep doing this for each failure mode you can name. The proposal gets narrower, and harder to attack, every time.
04
Start from why it is good, then maximize that
Democracy is good for three reasons. People feel heard. They actually are heard. And the government has an incentive to improve lives because that is how elections work.
Any change that increases one of those three is a democratic improvement. A system where anyone can send an idea, and a good idea gets implemented, increases all three at once.
Name the underlying good, then optimize for it directly.
05
Design the failure, then build the backup
List every way the policy could fail. Attach an automatic response to each one.
A subsidy meant to move farms off factory systems might be too small in year one. Attach a clause: if uptake is below the target after twelve months, the subsidy doubles. The policy stops depending on the first estimate being right.
A legislature can hand implementation to an outside body while keeping override authority. Congress could give a technical institute authority over AI policy, subject to approval by both chambers within 240 days, with a vote called automatically.
Then the body that is actually good at the work controls it, and others have to compete to do better. Governance can use the same pressure that makes markets work.
Section four
Three tools for agreements that are hard to reach
01
Leaders whose goals are the problem
Some conflicts persist because the person in charge wants something other than peace. Russia's war in Ukraine is a clear case: the goal is territory and influence, not security.
A summit cannot change a leader's goals by itself. What it can do is create a room where the cost of holding those goals becomes visible, and where a different set of goals, solving shared problems, has momentum and public attention. Two to four weeks of working sessions on problems every country wants solved changes what a leader is seen to be obstructing.
This may accomplish nothing. The cost of trying is low compared with the cost of another year of war.
A country can sign a treaty alone by making its own law conditional. Pass a bill that says: we pause frontier AI development if and only if China does the same. Or: we will not let AI make lethal decisions alone, if and only if China accepts the same limit.
It is a treaty written as domestic legislation, with the other party's signature implied by their behavior. It does not require a negotiation to begin.
03
Treaties that shift the equilibrium
Countries can pre-commit to joining a treaty when others do. India says it will join the IAEA if and only if Pakistan joins, and will leave if Pakistan leaves. Pakistan says the same in reverse. Neither has to trust the other first, and both end up inside.
Written this way, a treaty can create a stable outcome that did not exist before anyone signed.