engineer + builder + leader
Management and Internal Reward Functions
April 02, 2023


I remember the week that I wasn’t sure I was cut out to be a manager. I had been a manager for 6 months or so. I had read all the management books and had put together a good team with a positive culture. We had just released a much-awaited new feature that was getting major uptake from the field and a ton of kudos from our internal customers. I went home that night and the thought playing in my head was “I did nothing all week, I contributed nothing at all to any of the team’s accomplishments, and my job is totally without worth.”

Since then, I have come to really love management - though it took me another three to six months of suffering through self-doubt before I began finding wins to celebrate. I’ve now mentored dozens of people as first-time managers and pretty much all of them have at some point gone through the same experience I did. This is different from impostors syndrome as objectively I could see where I was doing well (and of course, where I had ample room to grow). Instead, I think it’s a problem of recalibrating your internal reward function.


In reinforcement learning, a reward function is the function that the learner is attempting to maximize. For example, an AI agent being trained to drive through reinforcement learning might have a reward function that minimizes the number of times a human driver feels the need to intervene or disengage. I think that humans also have an internal reward function (really a set of them, but we’ll focus on a single function), which they use to understand their accomplishments. In the context of human productivity, the reward function is often trying to maximize getting something done - especially for builders like engineers. If what a person is doing every day scores high in their internal reward function, they will feel satisfied. If it scores low, they will want to make a change.

In general, as you become more senior in your career, you are working on tasks at longer time-scales and so your internal reward function shifts over time (or perhaps you decide to stay at a particular level because your reward function doesn’t shift). As a junior engineer, your reward function is maximized for nearly instantaneous tasks. Every PR you commit, every dependency you update, and every deploy to production feels like a high activation input into your reward function. As you grow in seniority, the timescale of your work changes, but the change is relatively slow and continuous, allowing your internal reward function to update as you take on new tupes of work. As a mid-level engineer you might have to wait until a larger feature ships, or as a senior engineer you might have to wait until the quarter-long feature they’re tech leading is in production. It isn’t until you become a Staff+ level engineering that the tasks you’re doing - and the tasks your internal reward function needs to activate on - are much longer initiatives like setting the architectural direction for a whole team, for which you might not see the fruits of your labor for a year+. (Sidenote, that’s also why at ClassPass, Senior engineer is a terminal level and only one of the many paths available is becoming a Staff Engineer).

However, when you perform the career change that is stepping into management, there is a huge discontinuity between the tasks you’re performing and the way your reward function works. For an IC becoming a first-time manager, on Monday their reward function was activating for pushing code to production, reviewing PRs, and for generally being in the flow of building. Then on Tuesday, they become a manager, and now their tasks are happening on much longer timescales. They’re expected to set a great team culture, which is hard to validate until you’ve seen it at work for a few months. They’re expected to build a team of people that can get the right thing done, which means that instead of shipping a feature directly, they’re helping serve a team that can ship that (and many other pieces of code). Their reward function is now not firing at all; “how can a meeting be a valuable use of time,” it says, “when it’s not producing any tangible code?” In my experience it takes between three months and a year to really rewire that reward function so that you are able to take that satisfaction from the tasks of management.


There are a few responses to this mismatch.

The first is the new manager immersing themselves back in IC work. That doesn’t just feel comfortable, it feels valuable - their reward function is activating all over the place! However this means that they’re not learning the skills to become a good manager, nor are they working on rewiring their reward function. This is one of the reasons why I rarely let a new manager be a player/coach. I have worked with some extremely successful player/coaches and think that it can be an awesome role - but you can’t do it until you’ve settled your reward function.

The second is the new manager becomes despondent and goes back to being an IC. This could be because they didn’t wait long enough, or because their internal reward function just isn’t set up right now to be rewired in a way that makes management feel rewarding (or it could be that they weren’t a very good manager)

The last is that the manager makes it through to the other side. I haven’t found a good way to skip or shortcut this painful rewiring, beyond letting new managers know that they will likely have this experience as well as spending additional time with them on where they are being effective and productive. I think it does help knowing that the feeling is common and does not necessarily mean you won’t like management or won’t be a successful manager.


I have also experienced this and seen it happen as individuals move to more senior management roles - from a line manager, to a director, to an executive. In each case, your timescale and tasks go through a reasonably large step change and it takes time for your reward function to catch up. I have found these transitions easier than the one from IC to manager mostly because you’ve already crossed the chasm once and so know it’s doable a second time. That said, each time I’ve taken a big management jump in my career, I’ve felt that icy sense of worry around my accomplishments.

This delta between your tasks and your reward function is useful to keep in mind at all times, but especially if you’re moving from an IC role into management. Your internal reward function and what you’re actually doing are going to be out of line and so you should expect and prepare for that feeling. As I said, I haven’t found a way to help new managers skip over this part - if you’ve been able to do so succesfully, please let me know!