How much slower is remote work?

post by jsteinhardt · 2021-10-08T02:00:17.857Z · LW · GW · 11 comments

Contents

  Some Data
  Benefits of Remote Work
  Getting People to Work In-Person
None
11 comments

I'm a bit of a productivity nerd, and I like to keep track of how long tasks take. Sometimes this involves actually tracking everything I do all day, although most of the time I just keep rough mental track.

Given that lockdowns forced everything remote for about a year, and that some people are still remote, I was pretty interested in how much remote vs. in-person work affects productivity. I have some pretty direct comparisons, because I can observe meetings with the exact same set of people on the exact same project, both before and after it switched from zoom to in-person. I also have some harder-to-compare anecdotes on long-term effects.

My bottom line is that remote work is sometimes significantly less efficient (by factors of 2x-5x), but that this effect is felt most for people who are new to an organization or who are switching fields or roles. In particular, many senior employees won't experience a productivity hit, so the relevant decision-makers might systematically underestimate the costs of remote work.

For context, I'm a professor at Berkeley, so my data is from managing students and post-docs, but I think the trends probably generalize. Overall, I think you will take the largest hit from remote work if you:

• Are a junior employee, or are switching to a role that requires different skills or organizational knowledge.
• Work on more "abstract" or "conceptual" topics where it is useful to draw on a board and have shared visual context with collaborators.

Conversely, you will not notice much drawback from remote work if you:

• Are an experienced employee
• Are mostly doing "execution-oriented" work (writing code or papers)

Managers are a special case here because they are senior and mostly execution-oriented, but some management (such as giving difficult feedback or propagating social norms) is much easier in person. As a manager, I also found myself spending more effort on monitoring and improving people's mental health, although that might have been because of Covid-19 and not remote work itself.

Some Data

Here's some actual data to back this up. For more math-oriented projects, when we switched from meeting over zoom to meeting in person, I found that our previous 60-minute meeting finished at the 40-minute mark, because we had discussed everything we wanted to. The speed-up was a combination of:

• Fewer conversational pauses / interruptions, because it was clear from body language when someone was done talking.
• Shared visual context: you could easily tell what someone was attending to (e.g. an equation on the whiteboard) because of their gaze.

For more experiment-oriented projects, the gains were less consistent. For some that involved looking at lots of graphs each week, the shared visual context was also important and led to a ~33% speed-up similar to math. For others, the experimental results were relatively clear and most discussion was brainstorming next steps, which wasn't much slower over zoom.

While this was the most measurable change, I suspect the biggest effects by far are around onboarding new group members / spinning up new projects. I had 2-3 projects start during Covid that were qualitatively new compared to what that person had done before. In each of these cases, the first week that we met in person, we immediately made progress on questions that had been stuck for 2-3 months. This seems huge to me.

I also notice instances of people picking up new skills (i.e. a more theory-oriented person getting better at programming) via osmosis from other people in their office. In some cases they had been trying to do this for a while remotely, but picked things up much faster after switching to in-person. This has even been true for me, where I've picked up valuable research/advising tips from hallway talk with other faculty (I just finished year 2 as a professor, and would have been pretty happy to get the same tips a year earlier!).

Finally, the social cohesion of my research group has unsurprisingly skyrocketed since switching back to in-person.

Benefits of Remote Work

The two main benefits of remote work, from a productivity standpoint, are avoiding an open office environment and having more control over who can interrupt you.

Open office environments, despite the hype, are pretty terrible. There is a significant minority of people (including me) who mostly can't get work done in them, because there are too many distracting visual and auditory stimuli. For this minority, and potentially for others, working from home is great because anything other than an open office is great.

In some workplaces with bad productivity cultures, it is common to be interrupted when you would otherwise be doing focused work (this need not coincide with open office environments---some people get interrupted even in private offices). Working from home removes this possibility, and so can also be an improvement.

Finally, some people prefer to work from home for other reasons not related to productivity (e.g. more flexible schedule), but that's beyond the scope of this post.

Getting People to Work In-Person

After noticing these benefits, I've wanted to encourage people to switch to in-person. Of course, this is a bit touchy because some people may prefer not to do so for health reasons, or are (quite reasonably) averse to wearing a mask all day (Berkeley still requires this). Prodding people only had some effect, but offering events with free food had a huge effect---I immediately got ~100% in-person attendance at group meetings, for instance, such that we stopped having to offer a zoom option. We currently hold the food part of the meeting outdoors. This works well, except that sometimes there aren't tables, so we stand in a circle. It turns out that a lot of people's social habits are built around sitting at a table, and in a giant circle it's harder to get free-flowing conversation. But this seems fixable, and most places have tables anyways.

11 comments

Comments sorted by top scores.