Better living through job sharing

Most jobs-- especially most leadership jobs-- would be done better if shared

I was reminded to write this post by Tyler Cowen’s interview with Claudia Goldin. Goldin is well known for her research on the causes of gender pay gaps: in particular, for the finding that the more flexible the hours are in a given job, the smaller its gender pay gap. Cowen presses her skeptically on the extent to which high-productivity, high-responsibility jobs— the jobs that tend disproportionately to drive pay gaps— could feasibly be made more flexible.

I think most such jobs can and should be more flexible, because most such jobs can and should be shared: they should have a team of 2-3 peers and equals doing them jointly rather than a single person. Moreover, I think this becomes more true the further you go up leadership hierarchies. Finally, though gender equity and improved work-life balance would be important benefits, the case for job sharing doesn’t depend on those benefits: we can expect that institutions of all sorts would be more productive, more reliable, and more ethical if most of their jobs were shared. Since building higher-quality institutions is key to a better future, this should motivate us to push harder for broader job sharing.

What a job sharing world would look like

Job sharing exists now but is generally an unusual accommodation for employees with especially strong preferences for part-time work, and the positions open to it are, as far as I can tell, disproportionately lower-end individual contributor roles. What I’m envisioning here, by contrast, is a world where:

• Most educational classes, at all levels, would be team taught by 2-3 teachers or professors rather than a single one.

• Most companies would be run not by individual CEOs but by co-CEOs.

• Most countries’ chief executive authority (presidency, prime ministership, etc) would be vested not in a single person but in a team of 2-3 people.

• More generally, most people who now report to a manager at their job would instead report to a team of 2-3 co-equal co-managers.

• Even most major musical ensembles, for example, would be co-directed.

None of this is unprecedented: a couple of my favorite high school and college classes were team taught; I knew of some manager roles at Google which were shared by two people, and famously Larry, Sergey, and Eric operated as a triumvirate for many years; the co-CEO model is used by some pretty successful companies and the Roman Republic famously had two consuls at any one time. But making it the default practice would be a big enough change to constitute a social revolution. Why might such a revolution be desirable?

Shared jobs mean fewer SPOFs

In my CTO consulting practice I’ve found that many co-founders of small companies struggle with their own indispensability. It’s not just that they find it hard to take a vacation and are stressed out by the burden of constantly being the bottleneck for others; it’s that they realize the company they’re pouring heart and soul into is less nimble and less sustainable because its “bus factor” is 1. So we spend a lot of time talking through the path to giving others the knowledge, access, trust, and responsibility they need to share the load, and the management practices needed to build a company in which no single person is indispensable.

Any engineer can tell you

¹

about the importance of redundancy to systemic reliability: this applies to human systems as much as to mechanical ones. That goes double for higher-ranking leadership roles in a hierarchy, since more people depend on each such role being performed well: the huge effect of founders’ deaths on business productivity is a marked example. Introducing explicit redundancy to job roles is an obvious way to build institutional reliability. And arguably co-equal job sharing is an even better way to do this than the more common understudy/vice-president/succession-plan models, because being a designated backup is a poor substitute for actually taking a day-to-day part in the performance of the role.

Shared ideas are better ideas

Leadership always, and even more so in our time, is disproportionately thought leadership. The most productive work, likewise, is disproportionately knowledge work. So anything that improves the quality of thought and knowledge generation is, all else equal, very good for productivity.

And having to constantly share your ideas with a peer and equal is a powerful tool for making those ideas better! That’s why pair programming and adversarial collaboration are known-good practices. It’s also why chavrusa is a central tradition of Jewish religious study. Job sharing would generalize the advantages of those practices.

Shared authority is less abuse-prone

The principal-agent problem is among the oldest obstacles to efficient and ethical institutions, and the further up a hierarchy you go, the worse it gets. Anyone whom any institution entrusts with great power will have numerous ways in which they can use that power for their own benefit to the detriment of the institution’s effectiveness and values. Innumerable accountability and monitoring mechanisms, all with their own downsides and inefficiencies, are designed to keep CEOs from skimming from the corporate coffers, presidents from lining their pockets, teachers from sexually exploiting their students, and so on.

Having a co-equal share a job’s power with you isn’t a guarantee that you won’t abuse that power, but we should expect it will help considerably. Such a co-equal has both incentive and means to function as informal auditor and watchdog toward your official actions, and you toward theirs. We realize that, and distribute authority accordingly, when the cost of abuse is particularly clear and large: think of the protocols that require two soldiers to simultaneously turn two keys to launch a nuclear weapon. But using similar protocols more broadly would not only protect against catastrophe but diminish inequity, since people already socially marginalized will tend to be more vulnerable to abuse by superiors with sole authority over some aspect of their lives. If you’re inclined toward anarchist/radical-democratic/anti-hierarchical ideals, you can think of sharing authoritative jobs as a way to move toward those ideals, reducing domination and exploitation, without giving up the practical advantages of hierarchical power structures.

Shared roles are more “formative”

Yuval Levin, in his excellent book A Time to Build, makes a distinction between what he calls “formative” and “performative” institutions. A formative institution is one that works to mold the character of those who hold responsible positions within it, to make them part of something they know to be larger than themselves, to guide them to serve a greater mission. A performative institution is one that individual actors use as a platform to build personal brands and/or social capital for their own selfish ends. Obviously in the real world this isn’t a clean binary— all minimally functioning institutions contain both formative and performative elements— but Levin’s thesis is that our institutions are becoming less trustworthy and less effective because they are moving more toward the performative side.

A shared job means less spotlight on an individual, thus less opportunity to performatively build a personal brand, and the effect of this is bigger for the “top” jobs. Job sharing also institutionalizes key formative value judgments: that the mission is greater than any single individual and should survive beyond any individual’s tenure, that you should care more about a thing getting done than about you yourself being the person to do it. It encourages a humility that most institutions could use a lot more of.

The future is more pro-sharing

We have several reasons to believe the advantages of job sharing will grow over time:

• Knowledge work will become even more predominant.

• Reliability risks generally, and abusive-agent risks specifically, will become more severe as we rely on complex institutions to sustain a higher level of technological civilization.

• More prosaically, if “the robots” really do “take” most jobs, job sharing will help mitigate the effect of that transition on unemployment, and simultaneously enable greater service availability and shorter working hours.

Problems and mitigations

So what’s most likely to go wrong if more jobs are shared? A few possibilities:

• It can be harder for even the smallest group to be decisive than for an individual to be decisive, and that lack of decisiveness has a cost. There’s a reason “designed by committee” is a pejorative. Hopefully keeping sharing-group sizes to two or three people will limit that cost, but clear norms for tiebreaking and timely dispute resolution that don’t just boil down to “split the difference” are still needed to keep it manageable. Luckily, these kinds of norms are already necessary for the orderly functioning of collaborative organizations, so solutions to this problem are already evolving.

• Labor costs will be higher for shared jobs. Much of this is likely for any future in which people work somewhat fewer hours on average, and that future would all else equal be a good one, so we should be willing to bear those costs. But there are probably also significant search and matching costs unique to the problem of finding 2+ people to share a job.

• Some people have very idiosyncratic skills and talents, such that it would be exceptionally hard, if not impossible, to find a good sharing partner for them. In some such cases, poor collaboration skills are part of the flipside of their skill or talent, and leaving them to work as a lone genius is a necessary mitigation mechanism. I think this is true mostly in individual contributor roles; generally lone geniuses don’t make good leaders, though they often seem to, at great cost to those who follow them. It’s also more often true of independent contractor and artisan types, for whom competitive market incentives solve many of the problems job sharing otherwise might. So focusing job sharing on organizational leadership and other deeply organizationally-embedded roles makes it less affected by this tradeoff.

• Humans may have deep-rooted desires, whether cultural or biological or both, for deference specifically to individual personal authorities. Having (I think!) much less such desire than most people, I find it hard to empathize with this, and thus hard to understand it. But it’s worth more psychological and social-scientific investigation about how job-sharing in powerful roles might interact with this, and how feasible cultural change to mitigate it might be.

What did I miss?

Other advantages of job sharing? Other downsides or challenges? Books, articles, other references that explore these issues? Let me know!

1. Many engineers, indeed, delight in nothing more than telling you about this at great length.