Work is a Verb

A newsletter by

Knowledge work has a problem. Remote work got the blame.

One number went viral at the start of June. Remote work, the headlines said, explains 64 percent of the rise in unemployment for young college graduates. A companion paper tied remote work to a third of the post-pandemic rise in mental distress. The prescription floated through the coverage like a given: bring everyone back.

Most of the coverage skipped the fine print.

The 64 percent is a preliminary estimate from a working paper that hasn't been through peer review yet. The researchers describe it themselves as "back-of-the-envelope." That doesn't make the research wrong. The problem it points at is real. Unemployment among recent grads climbed from 3.6 percent before the pandemic to 5.6 percent this spring. Something changed. The question is what caused it.

The study doesn't actually measure who works from home. It sorts occupations by whether the work could, in theory, be done remotely. Engineers who were marched back to the office five days a week still count as "remotable." The category is much larger than the behavior. Remote work surged in 2020, then leveled off while grad unemployment kept rising. The thing being blamed stabilized while the problem it supposedly caused kept growing. That is a strange way for a cause to behave.

Then there's AI. The jobs labeled "remotable" are nearly the same jobs most exposed to generative AI, a rank correlation of 0.77 across occupations. The two variables move nearly in lockstep. "Remote work, not AI" and "AI, not remote work" leave the same fingerprint in the data. Stanford research finds workers aged 22 to 25 in the most AI-exposed occupations saw a 13 percent relative employment decline since generative AI spread. Big tech quietly gutted the entry-level door around the same time: by 2024, new grads were down to 7 percent of Big Tech hires. The door didn't close because people went home. It closed because companies decided to stop opening it.

The policy implication floating through the coverage was clear: end remote work. But researchers at the University of Pittsburgh tracked 54 major tech and finance firms that ran return-to-office mandates. Time-to-fill rose 23 percent afterward. Hire rates fell 17 percent. The companies that ran the implied experiment got slower at hiring, not faster.

Knowledge work does have a problem, and it's falling hardest on people just starting out. The honest read is just different from the headline. What develops a young worker has nothing to do with the commute. It comes down to whether someone owns your first 90 days. Whether you have a real mentor or just a Slack channel. Whether the team writes down how the work actually gets done.

When people feel isolated, that's usually a design problem. You don't fix a design problem with a commute.

Remote work is the simplest story we could tell about knowledge work. The simplest story is rarely the right one.

The full breakdown, including the AI entanglement problem, the mentorship evidence, and what to ask before accepting any role, is on the site.