Epistemic status: Quick write-up to get the idea out there.

“Plans are worthless, planning is everything” - Dwight D. Eisenhower

Many governments say they want to be agile, I’ve seen this in many different policy documents, but words are cheap. If you want to be agile, you can’t just say you want to be agile, you need to design mechanisms to make it happen. I don’t want to be too prescriptive about the exact mechanisms here. Honestly, I suspect that it’s a bit like homework, a lot is lost if you just copy someone else’s answer. It’s far more valuable if people in government seriously think about how to make it happen.

You do want to include a few things though to guard against things going wrong. You want the plan to consider both how to respond in an agile fashion, but also how to avoid this devolving into counter-productively thrashing around. You want the plan to be formed by looking at the current rate of progress of capabilities and how fast that government needs to be able to respond. You want the plan to be published, along with details about how the plan was formed (though it might make sense for these details to be released *after* the initial plan so as to not slow it down). You want a review cycle to determine if the plan seems to be working, but you don’t want the review cycle to slow things down. This means that you need a mechanism to accelerate review if the circumstances call for this (along with some forethought about what kinds of circumstances might warrant this). The project would need independence to protect it from political pressure to pretend that things are going better than they are. The project should also seek to be apolitical.

I think this could be an easy win. I think it could be a gift that keeps on giving. It’s also exceptionally robust to different scenarios/worldviews.

Main caveats:

• It feels as though this ought to have positive impact, but this isn't guaranteed. We've already noted a number of ways in the main text that things could go poorly.
• Perhaps you think you already know what concretely needs to happen (ie. AI pause). In that case, an agility plan might be a distraction—or worse, an excuse to delay.
• If timelines are short, then perhaps there simply isn't time for this proposal to provide much benefit.