> Summary: Certain (not all) Amazon warehouses seem to have per-employee injury rates that are significantly higher than the industry average, as in twice as high or more. Apparent reason: It’s not they’re actually dangerous places to work, it’s just that they’ve maximized efficiency and reduced waste to the point where people are picking and packing and shipping every minute they’re working, never stopping. And a certain proportion of human bodies simply can’t manage that. They break down under pressure.
Wouldn’t a more parsimonious explanation be that the job previously consisted of a safe part and a dangerous part, and it was easier to automate the safe part, so now more of a human's time on the job is spent doing the dangerous part?
If you imagine the "safe part" of the job as "slack" (i.e. imagine that the employees were previously half working, and half doing nothing—not that they actually were, but it makes the model clearer), then they've effectively just gotten rid of all the slack, and so made the employees do twice as much work per work day. If there's a constant small probability of injury per item packed, then of course doing twice as much work per work day, will lead to twice as many injuries per employee per work day. You'd have gotten the same number of employee-injuries out per items-packed in, no matter what time scale that packing occurred on.
> If there's a constant small probability of injury per item packed, then of course doing twice as much work per work day, will lead to twice as many injuries per employee per work day
This assumes injury-risk is a linear function of number of items packed. That's not true. Fatigue increases injury-risk - that is how tired you and your entire team is while performing the task. If the person sorting the packages up the line is tired and making mistakes, the person binning it is more prone to mistakes, leading to an injury. In summary, there is a "fatigue-threshold" after which injury-risk increases non-linearly (say exponentially) with every item packed.
If fatigue is a simple threshold (i.e. you "become fatigued" after a certain number of units of work done), then the obvious solution is to work the employees at 100% utilization for half as long (at which point they should be roughly as fatigued as the employees worked half as hard for twice as long), and then swap them out for a second shift. (I.e., instead of having 10 employees with 40hr weeks, you now have 20 employees with 20hr weeks.)
This is the approach I believe is also advised for improving productivity/decreasing errors in the medical field (where doctors have a hard time not pushing themselves to 100% all the time, because of their personalities): hire more doctors per hospital, such that each doctor can be cut down to a shorter shift (and thus maybe lower salary, too.)
If, on the other hand, fatigue is about the lack of micro-rests between individual units of work, then the appropriate solution is more subtle:
1. replace the low-quality rest of doing "slack" work, with high-quality rest of doing nothing-at-all (or even "actively" resting, the equivalent of an athlete doing cool-down stretches between sets of an exercise), so that employees can cool down more quickly and/or return closer to peak productivity from each rest;
2. have slightly more employees (hopefully fewer than double), working off branching lines, such a way that units of work are directed to go to whichever employee is "fresh" rather than "stalled." Like processors with pipelined executions and multiple APUs per core, directing instructions to the APU that isn't currently blocked.
> If fatigue is a simple threshold (i.e. you "become fatigued" after a certain number of units of work done), then the obvious solution is to work the employees at 100% utilization for half as long (at which point they should be roughly as fatigued as the employees worked half as hard for twice as long)
I am skeptical. Fatigue rate would be the delta between exertion and recovery rate. Recovery rate is far from negligible.
Try running up several flights of stairs as fast as you can, vs taking them at a leisurely pace. Most people would be winded after the former, but not bothered after the latter.
EDIT: It's made more complicated by the fact that we have many different "recovery rate"s.
When talking exercise, there's the obvious "V02 max", or the maximum rate at which we can absorb oxygen. So long as our muscles have sugar/fat, they should be able to work if we keep our exertion below this point.
If we exceed this, our muscles begin working anaerobically; recovering energy there takes much longer. That's the difference between running and walking up the stairs.
I imagine there are other types of fatigue, but I am not a medical/health/exercise professional.
Mental fatigue is different again as well.
But there are probably background "fatigue"ing rates that are important, like hours-since-you've-eaten. In that case, doing the same work in half the time may be better.
It's easy to talk myself into realizing I know nothing.
Wouldn’t a more parsimonious explanation be that the job previously consisted of a safe part and a dangerous part, and it was easier to automate the safe part, so now more of a human's time on the job is spent doing the dangerous part?
If you imagine the "safe part" of the job as "slack" (i.e. imagine that the employees were previously half working, and half doing nothing—not that they actually were, but it makes the model clearer), then they've effectively just gotten rid of all the slack, and so made the employees do twice as much work per work day. If there's a constant small probability of injury per item packed, then of course doing twice as much work per work day, will lead to twice as many injuries per employee per work day. You'd have gotten the same number of employee-injuries out per items-packed in, no matter what time scale that packing occurred on.