4 ms·

I wish more people knew the story about UPS telematics. You're absolutely right, but it's not just speeding. UPS trucks these days have about 200 sensors, cover

I wish more people knew the story about UPS telematics. You're absolutely right, but it's not just speeding. UPS trucks these days have about 200 sensors, covering everything from speed to rear-door position to seatbelt buckled-ness. As I've heard it, the story goes something like this: UPS set delivery targets for drivers that were too high to consistently achieve. When drivers fell behind, they sped and caused accidents, so UPS installed speed monitoring. Drivers switched to leaving the rear door open between nearby stops, so UPS installed door sensors. Drivers started leaving seatbelts unbuckled, so those got sensors. UPS essentially frames this as fighting Goodhart's Law with brute force, trying to track so many metrics that they can't be gamed. Except, of course, that the targets aren't realistic. What actually happens are results like drivers sitting on buckled seatbelts, moving heavy packages in injury-causing ways, or marking "signature required" packages as no-shows without knocking to shave off time. None of that is permitted, UPS fires people who are caught doing it, but the rest of the time when it isn't caught, it's all convenient cost-savings. There's nothing wrong with holding individuals accountable for their response to incentives, but it's possible to do that while also holding people accountable for the incentives they create.

naveen99 8y ago
I don’t know about ups drivers, but in many jobs, there is a huge difference in efficiency between workers. the least efficient workers protect the more efficient workers when the bosses want to squeeze the more efficient workers also.
- Bartweiss 8y ago
  This is a really good point. My understanding is that UPS has partially bypassed this by setting per-route targets which update independently, but within a multi-driver route it could still apply. FedEx dodged it by franchising so that slow workers outright went into debt to the company, but then got sued on employees-as-contractors grounds. Overall, though, I think it's a huge and underrated effect limiting Taylorism. Designing strictly around efficient workers isn't viable if you can't find a lot of them, and using multiple workflows is rarely cost-effective. There's a story in Rivethead about a car-assembly stamping job that was assigned to two people splitting tasks, but could be done by one person working very smoothly. New guys would do it in pairs, but experienced workers could each do a half-shift solo and duck out to the bar for the other half. As far as I know, time-motion studies never put an end to that, because it was too scarce a skill to schedule and organize for. (Of course, the inability to squeeze out more productivity has consequences for automation, but that's another story.)

naveen99 8y ago

I don’t know about ups drivers, but in many jobs, there is a huge difference in efficiency between workers. the least efficient workers protect the more efficient workers when the bosses want to squeeze the more efficient workers also.

Bartweiss 8y ago
This is a really good point. My understanding is that UPS has partially bypassed this by setting per-route targets which update independently, but within a multi-driver route it could still apply. FedEx dodged it by franchising so that slow workers outright went into debt to the company, but then got sued on employees-as-contractors grounds. Overall, though, I think it's a huge and underrated effect limiting Taylorism. Designing strictly around efficient workers isn't viable if you can't find a lot of them, and using multiple workflows is rarely cost-effective. There's a story in Rivethead about a car-assembly stamping job that was assigned to two people splitting tasks, but could be done by one person working very smoothly. New guys would do it in pairs, but experienced workers could each do a half-shift solo and duck out to the bar for the other half. As far as I know, time-motion studies never put an end to that, because it was too scarce a skill to schedule and organize for. (Of course, the inability to squeeze out more productivity has consequences for automation, but that's another story.)