When I was only a few years into my new life in QC, I found myself in the strange position of ‘owning’ an experiment without really having any sense of ownership over it at all. I would come in, do heroic battle with the arbitrary waveform generators, carefully tune the readout, despair at the garbage data and slowly, painfully converge on a nicely working device. At that point, the Stakeholders would be drawn in by the smell of progress in the air. You could tell they were approaching because the sound of ‘Have you tried…?’ would grow from a whisper to a dull roar1. And so they would appear with their many good suggestions2 about what knobs to turn, what to look for, how to interpret the data and so on.
The lack of agency was frustrating, I desperately wanted more time to think about what I was seeing and to explore on my own, beyond just troubleshooting and initial tune-up. There was clearly some inscrutable schedule driving my tech leads’ insatiable desire for results, but I was not privileged enough to learn what it was. I only barely appreciated how my work would be slotting in to the overall scheme of things.
No one likes to be micromanaged, or whatever the opposite is. Ignored? Undirected? Of course, not everyone can handle expansive freedom of action in the workplace, but when you’re hiring PhDs, or PhD-like people, you’re doing so because you need high-agency, high-performers to answer strange questions that quite possibly have never been asked before.
Eventually, I found myself going from a novice lab gremlin to a team leader responsible for the productivity of about a dozen people, many with PhDs themselves. How do you effectively lead such people? Well, the spoiler is “I am not sure”, but I do have some thoughts on what has made my team relatively high performing compared to others in the same organization. As far as I can tell, the ingredients for a successful team include, but are not limited to Freedom, Vision, Narrative, and Boundaries.
Freedom
This one is pretty obvious, and was the thrust of my opening anecdote. The point of having a high performing team is that they collectively do what you cannot. Thus you must trust them like you would trust yourself to make good decisions about which tasks to pursue, which tasks to abandon, and how to structure their work. Note that this requires that you’ve assembled high-agency individuals who can use the freedom you offer productively. Not everyone can, and that’s ok. Depending on the team’s composition and goals, those who can’t operate in a relatively unstructured environment must either be removed from the team, or managed closely by a delegate of your choice.
While I offer my team very little in the way of specific instructions, that doesn’t mean they are directionless. Their efforts should be informed by adherence to The Vision.
Vision
In my head, there is a grand, overarching plan for the whole team and what their work means for the superconducting qubits group as a whole. In fact, based on feedback from my team, one of the things I need to do better is actually share the vision out loud more often. It helps people to know that they’re not working on something no one cares about. Similarly, sharing the vision helps the individual team members make good decisions, especially when I’m not around.
If everyone understands The Vision, then they understand where we’re going. They know the motivating goal of the team. Thus, since they’re the experts on where we stand relative to that goal, they should be able to devise some sort of thru-line from here to there. If you’re ever wondering whether your team understands The Vision, just keep asking them what they’re planning to do after $CURRENT_TASK is completed. Keep going until you’re either satisfied, or they get annoyed with you because there’s too much uncertainty to forecast for sure.
Narrative
One of the most useful tricks I have picked up from my mentors is to ask, semi-rhetorically, “What’s the story here?” whenever discussing a new result. It’s a question that’s almost fundamentally at odds the reality of the scientific endeavor. I have never done an experiment that followed any kind of narrative arc. It’s always an iterative process of adjustment, measurement, and analysis. Nonetheless, once all of the data are recorded, analyzed, and understood, it is useful to repackage it into a narrative. This is (part of) the reason good scientific talks are good. They don’t chronologically follow what the grad students did in the lab, that would be insane3. Instead we start with some overall context, followed by a motivation, which defined our actions (the experiment), which produced our data, which we analyze to come to a conclusion that answers4 the motivating question.
The same trick works great in industry, and is an important counterpart to The Vision. The narrative also helps to succinctly describe your many successes to executives who may or may not know anything about what you do5.
Boundaries
With great freedom comes great responsibility. It is your responsibility, as team lead, to answer questions like- When is a task considered done? How long to grind away at a problem before re-evaluating? How long to spend on interesting side problems? The PhDs on your team run the risk of falling down rabbit holes chasing minutiae, because they lost sight of The Vision in favor of a cool, weird problem. Or they might spend weeks reading obscure texts about Green's functions to craft the perfect solution, when a Good Enough6 approach could be crudely implemented within a day. You need to be able to prevent this, or identify it and provide feedback to refocus the effort. Yes, there can be great value in rabbit holes7, if you accidentally stumble across the right one, but you also have deliverables and a schedule. There are people counting on you.
There is a divergence here between Industry and Academia, though the extent will vary from lab to lab. Since the whole point of a graduate program is to learn, I would expect a relatively wide latitude given to graduate students during their education8. So when you are thinking about reasonable boundaries for your team, don’t lose sight of your local context.
There is a secret fifth ingredient. It is the most powerful and impactful of them all, but few are truly prepared to wield it. Nonetheless, mastery of this one thing can set you apart from your peers. It is simple to state:
Shut the Fuck Up
For those of us who find ourselves in positions of leadership and power, especially over novices with little formal work experience, talking too much can be a failure mode. Statements that are intended to be idle speculation can be taken as suggestions, or worse, commands to perform this or that experiment. A few seconds of thought might trivially show that the ideas were silly or unworkable, but that’s a few seconds for people who are masters of the subject. Novices don’t have the benefit of that experience, especially if they’re not being closely mentored by more senior members of the team.
I’m not saying you should stay mysteriously silent like some sort of inscrutable deity. You should just talk less and understand that, due to your position, your words will carry more weight than you intended or expected. In my view, most of the talking you do should be in the form of questions.
Conclusion
I don’t believe that physicists have any higher rate of naturally good managers, or particularly bad ones. Successful physicists, be they academics or in industry, are plagued by the same problem that every successful professional faces. What got you here, won’t get you there9. The technical excellence we develop and hone through grad school, post-docs, and subsequent industry jobs is necessary, but not sufficient to guarantee success as a leader of a high-performing team. The job is fundamentally different, but it will (hopefully) build on your technical skills. If I could write down a checklist for how to do it well, I would10. Alas, all I've got are some things I've noticed and thoughts I've had so far. Good luck!
Postscript- Individual Contributions
Common wisdom suggests that it’s not appropriate for people in leadership positions to do much individual contributor (IC) work, if any. This is mostly true, your focus should be on ‘high leverage’ work11- tasks which will enable more of your team to work faster/better. It has been a long time since I've worked on a fridge or personally run an experiment. However, I still engage in regular simulation work and data analysis, often in parallel to whomever is actually assigned to do so. I do it to close gaps in my own knowledge, or improve my understanding of some new (or old!) measurement. I usually find that such work helps develop my intuition better than simply reading papers, which is critically important in this line of work. You've got to be careful here, it's easy to start with a few lines of code and end up deep in the rabbit hole of obsession. It would have been your job to watch for this in your team, but presumably your boss trusts you not to fall prey to this yourself.
It’s hard to really blame them, their responsibilities had grown so great that they’d stopped taking data long ago. They were in withdrawl from the scientific life they’d known and just looking for their next data fix.
I’m not being facetious here. They had many more years of experience than I did, and, in the end, really improved my understanding of superconducting qubits, and superconducting electronics in general.
Or an ethnography of science, I guess.
Suspend your disbelief.
This probably depends on whether you work at a ‘pure-play’ QC company or not.
Good Enough is highly context dependent. In highly-nebulous environments, I consider good enough to be something that gives us a clue about what to do next. Sometimes, that clue points to 3 weeks of Green’s functions, but usually not.
I like to let rabbit hunting go for a day or two, at most, then refocus the team if it doesn’t seem particularly productive. Similarly, if all of the high priority tasks are covered, I think encouraging exploration of weird questions and little details is beneficial. You never know what you might find (spoiler: you find more problems to fix).
This is a little worrisome to me. When the sheer financial gap between industry and academia starts to make grad school seem like an insane decision, we are losing something valuable. Something that cannot be easily replaced inside of corporate structures that are fundamentally motivated by (eventual) return on investment.
Also the name of a decent book by Marshall Goldsmith. It could be condensed to a long pamphlet, but I found it useful to read when I was nearing the IC → team lead transition.
I’d also be filthy rich from my well-attended leadership seminars.
I think I first encountered this term in Andy Grove’s High Output Management, which I think is useful even for individual contributors.