A first list of moves
This is a first ranking, drafted to be argued with. The order reflects two things: how much the move helps across all the worlds the argument allows, and how little it costs if the argument is wrong. Moves that are good in every world come first. Moves that only pay off in specific worlds come later, weighted by how likely those worlds are and how much they would hurt.
1. Decide what you believe
Read the argument, the resources behind it, and the reviews against it, and arrive at your own probability for each link in the chain. Write it down. Everything below depends on the numbers, and borrowing the authors' numbers or a critic's is the most common way to get the rest wrong. This costs time and nothing else, and it is the only move that improves every other one.
2. Do what is good in every world
Health. Relationships that would survive hardship. Work you would choose even if nothing were at stake. Savings and low fixed costs. Skills that transfer. These are the same in a world with no AI risk at all, which is the point: their value is unaffected by the outcome you cannot hedge, so they are the best use of effort under uncertainty about it. If taking the argument seriously changes anything here, it should be to do more of these, sooner.
3. Get on the right side of the tools
Use the systems, competently, now. Know what they can do and what they cannot. In the dislocation and disempowerment worlds, the people who come through best are the ones the tools amplify rather than replace, and that is determined early. Where possible, own rather than rent: capital, equity, a business, anything whose returns do not depend on your labour holding its price. This is the move most specific to the non-binary bad worlds, and those worlds are likely enough that it earns its rank.
4. Position for the bad-but-survivable worlds
Live somewhere with institutions that work and a state that is neither failed nor total. Keep some assets and relationships that do not route through a screen. Have the classical resilience basics, at the level of a sensible household rather than a compound: supplies for weeks, redundancy in what you depend on, neighbours you know. This is cheap, it is what any careful person does anyway, and it is the correct response to the partial-catastrophe and misuse cases specifically.
5. Contribute to avoidance in proportion to your belief
For the binary world, prevention is the only survival strategy, so some effort belongs here for anyone whose probability is not negligible. For most people that is voice and money. Voice: the march pledge, letters to representatives, saying what you think in rooms where it is unusual to say it; the book's Keep the Coalition Large is good advice on how. Money: to the organisations working on the governance and verification mechanisms that the Avoid section argues are load-bearing. If you have the relevant skills, career, but that is a small group. Decide the amount from your number, give it, and stop.
6. Set an attention policy
Decide now how often you will revisit all this, and hold to it. Quarterly is plenty. Between reviews, act on the last decision. The previous page explains why; the short version is that the question returns almost nothing for continuous attention and can take everything.
What not to do
Do not build a bunker, buy remote land, or stockpile against the binary outcome; it does not have a survivable margin, and the money is better spent on items 2 through 5. Do not treat the argument as licence to abandon long-term commitments; the future is still the most likely place you will be. Do not bet a career on a task the systems are visibly approaching. Do not confuse being informed with being prepared: reading about the risk is not a move, and after item 1 it is mostly a cost.
What is missing
This list is generic by design, because it was drafted without knowing the owner's circumstances, and the right list is specific: to age, dependants, work, location, and the actual numbers from item 1. The revision this page is waiting for is the one that makes it particular.