Transcript: Goals, Motivation, and Shipping

This is a transcript of a mentorship conversation I had with an AI Safety founder. It has only been lightly edited.

The way to manage an organization is through goals. Everybody knows that. But those goals tend to be boring and not actionable. People say we need this much in sales, we need to hire this many people. They’re boring. People get detached. Most of these goals don’t increase the energy and the motivation in a team. They decrease it.

There is marketing you use for the outside world, where you’re marketing your product. But there is also marketing you have to use internally. What are you telling yourself and your team? What is bringing you to work every day? What are the big exciting goals? There’s a lot of art to that.

The best ideas are in the room. Not my ideas. So when we used to plan on a quarterly basis, I would do a request for proposals. Everybody sends suggestions for how we could work. Anyone. Sales, the janitorial service, it didn’t matter. Anybody can send one.

But then you end up with hundreds of ideas, and you can’t just hand that to your engineering team, because it’s distracting. They’re all different. So the big part of the exercise was clustering them in my head so that I could grab twenty five ideas and give them a single name. These people are all talking about how to visualize data on the website. These other ones are talking about something almost like visualization, so I’m just going to put it there. You create the cluster, and at that point you throw away the contents of the cluster. Now you have a shell. You say: visualizations. And you give it to the right person to work on.

Then you go back to your team and say: because of all your suggestions, we’re going to work on this. And from that you develop very operational goals. I’ve always been good at figuring out the singular goal that, if you go after it, unlocks a lot of side quests along the way.

The impactful decisions are few. They tend to be not that many. As the organization grows you have fewer of them, because you have more people. When you’re small you have impactful decisions every day. But once you have ten million dollars in investment and twenty or thirty people, your impactful decisions are fewer. Call it one per month. But they’re very important, and they happen fast. You meet someone at a conference and decide you need to hire them, and then you work really hard to hire that person. You bring in an investment. You shift the strategy: we’ve been doing this, it’s not going anywhere, let’s convince the team to do this other thing, and everybody leans onto that.

There’s a reason you can only do so many of these per year. They’re sudden to the organization. If you do too much of it, you’re the kind of leader who feels like he’s distracting everybody. If you don’t do it at all, you’re a passive leader.

Elon Musk has this concept of a surge. When they’re behind on some goal, he calls a surge and moves into the factory. He’ll move into the Tesla factory, sleep on the floor, and make everybody stay. They go for a week. A very intense period to catch up to some goal. The key to good leadership is knowing the right balance of how much of that you can do. Most people don’t do it enough. Most people are very passive about their organization. They assume everybody in the org has agency, everybody is equally motivated, everybody is doing the right things.

People want to be on a mission. If you’re a kind, caring person, there’s this feeling that pushing people to work more, to do more, means you’re exploiting them. So you back off from that behavior. But people, especially when they’re young, want to go through intense periods. Nobody would volunteer for the World War Two trenches. But when they’re forced to be there and then they come out, they remember it as a very important period of their life. People want to go to war on a mission together. Your job is to constantly make people feel they’re on a mission.

My prime as an operator was at Amazon, where I had a hundred and fifty people. I operated like a conductor of an orchestra. There were people on my team who were very structured: have a plan, work the plan. That was never me. I was feeling the room, trying to figure out whether I could get more value out of it. If I came into the office a few days in a row and felt things were slow, I’d do something like call a hackathon for tomorrow, right away, just to change the momentum. Or I’d do one on ones with almost everybody, and I’d always talk to people about what they could do to maximize their impact.

Use the world as a processor. You can think of the world as working for you. You put something out there, the world processes it, and it comes back to you. There’s a version of this that’s just bias for action. You can spend cycles thinking about whether to do something, when in fact you can just do it and then decide.

Take a conference. You can sign up to give a talk three months out, on a topic you’re not prepared for at all. If the organizers say no, fine, you don’t have to worry about it. But if they accept you, now you have to do it. That tends to work. I put a lot of irons on the fire and see what comes back to me. I don’t need to keep thinking about it. I do the application quickly, and if it comes back, suddenly I need to deal with it, and I go through three or four crazy days of work and prepare myself.

This week I was working on a benchmark, and a lot of data vendors were giving us tasks that were shitty. So I started writing notes about what makes a good task. It was an internal document, just for me, for future reference. Then I thought I should write it as a manifesto on a domain name. Buy the domain, call it goodbenchmarks.ai, put the thing up. If people are into it and it gets distribution, I’ll make it better. If they’re not, fine. I turned something boring and lame with no consequences into something I could externalize and see what happens.

Think of your company as a black box that churns out artifacts. You’re sending signal to the world and getting signal back. So the question is always: what can you put out?

When in doubt, ship the closest thing. If you find yourself without a good way to prioritize, prioritize whatever is closest to being packaged and sent out. Do that over the 401k plan. People care about the 401k, but nobody on the outside is going to look at your company and say, oh, you guys have a 401k.

A lot of management is just not doing things. It’s saying no, and getting obsessed with one thing. So push the admin off your plate. A part time accountant, remote, in another country, can do your books, and then you ask them to do a few more things and pay them by the hour. It’s useful to learn about something like a 401k once, but it can eat a huge amount of time.

There are the big level goals, and then there’s every week, where you’re feeling the pulse of the company. If by Wednesday you feel like you’re not putting anything out, you haven’t written a blog post, you haven’t shipped anything, then you pause. You talk to your teammates: what’s our next milestone, what’s our next most obvious thing? And you clutch onto that.

Start with the smallest thing you can ship, then escalate. Say you want to demonstrate something. What’s the easiest way to show it? Do the cheapest, most obviously artificial version, the one nobody would ever do in real life, and put it out there. Then you escalate from there. You go from obviously a demo to something more real, and if attention is being paid, you keep escalating. Each step pulls more people in. You can get others bidding into your goal, providing the thing you need, until they’re aware of you and you have momentum.

The best teams are mission-agnostic. At the end of the day it’s about an effective organization. You could switch a great team’s mission and they’d still do well. Take a team building one thing, drop them on a deserted island where they need to build survival systems, and they’ll do it. The best teams can do anything. The intrinsic capability of the team is that it can take any objective, operationalize it, and deliver it. Because you’re an early company, that practice matters more than your strategy.

A company with five hundred thousand dollars and two employees is not going to have a huge impact. What you need is fifty employees and ten million dollars, and to get there you go through the hunger games, the trials and tribulations, and you show everybody who could support you that you’ve actually gone through the process. You have to show people, hire people.

A lot of it is performative. The mission is an excuse to show that you can accumulate resources and control and power, that you can be a good steward of some social power, that more people should trust you with their careers and more funding should be trusted to you. Once you have some momentum, the goals start becoming more and more impactful.

Some people get this operational intuition because they played sports in college or went to the military. Some are just born with it: they get anxious if they’re not delivering results. But you have to get there. It’s not something you can put in a goal. It’s something you wake up every day thinking about. What can we put out? How do you make it exciting and interesting for the team?