You start by breaking it down, a lot. Into smaller problems. Analyze, and break those down too. Find edge cases. Wash, rinse, repeat. For example: "what is the

You start by breaking it down, a lot. Into smaller problems. Analyze, and break those down too. Find edge cases. Wash, rinse, repeat. For example: "what is the best way to organize a liberal democracy of >100M citizens?" - How can you tell than an organization is better than another? (i.e. what does best mean) -- This implies a set of comparison criteria, find those -- which criteria is more important -- do you get a total order or a partial order? - Who is a citizen? - What liberties (i.e. rights) does your democracy grant? -- Under what circumstances? -- Is there a hierarchy of rights? (i.e. if two conflict, which one is more important) -- What happens when there are not enough resources to grant those rights? And so on... You could spend rivers of ink writing about those. And the interactions between sub-problems can get mind boggling, but you just make a note, and make a call and keep going. In problems that involve other people, you have to include them. Ask how the look at the problem, try strategies of communication, etc. At some point, you just need to do something about one of the sub problems. The one you deem most impactful right now. It will hopefully take you closer to your end game. As you move, the problem space will move. And you'll have to revisit all your assumptions again and again. Each action you take, will have unintended consequences. Some problems will be solved, some new ones will pop up. Things that weren't important are now urgent, and things that seemed essential are really not. You deal with those as they come, but keep moving.