The author claims that proximity to a job — ranging from those that do the jobs being close and those that the assign jobs as far — is a good proxy for knowing whether a job is pointless. If the person doing a job thinks it shouldn't have to be done, then it probably shouldn't.
And when those people are asked, a lot of them do think their jobs are pointless, and that if they were to stop performing them, nothing would measurably change in the world.
Do the pistons in an engine need to know about the tires on the pavement in order to justify their work? Does the entire car need to know anything about the intentions of the driver for it to be useful? No. And yet, according to Graeber, if a corporate lawyer thinks that it's pointless for them to review a sub-department's accounts for tax implications, then it must be so. It doesn't matter if this isolated contribution helps, indirectly and incrementally, to push the company further.
Regardless of whether a company is producing insulin for third world countries or fattened geese livers for ostentatious wannabe aristocrats, there will be components of that company whose contributions are limited in piece, but vital in the whole.
Also, most progress is made through creative experimentation and failure. We try 100 things, 99 fail, and then one turns out to be penicillin. Do the 99 failures then count as bullshit? Should we have stopped trying things because people working on one of the failures were, at some point, demoralized and disillusioned? Fine, fair enough, if you can predict the one success before slashing the doomed projects, but to do so you'd need a functioning oracle — a crystal ball. Short of that, you have to try everything...
And what about redundancy? Another example: five people are hired to work an assembly line. These five people ensure the voltages are sufficiently within expected boundaries on, what seems to them to be, a random circuit. The voltages are checked in sequence; the first worker finds bad voltages all the time, but the fifth worker — last in line — has never seen an incorrect voltage, and that's after working at the assembly line for a whole entire year. This fifth person thinks their job is pointless. But the random circuit turns out to be a critical component in the number one selling brand of pacemakers. Removing this fifth person has been shown to increase failures in one out of 100,000 pacemakers, per year. 3,000,000 people have pacemakers. So, if we were to follow Graeber's theory, and follow the worker's intuition, 30 people's pacemakers would fail a year.
Or what about general risk management? The actuary whose work prevents bad things from happening never has anything active to show for their work. At best they can say, "look at all these horrors that did not happen!" Well the list of horrors that haven't happened is infinite. Did you prevent the moon from cracking in half as well, some cynical critic might ask? And so perhaps this actuary also thinks their job is bullshit, too; that their pouring over spreadsheets at the paid demand of some brokerage house whose automotive insurance subsidiary provides a cashflow for future investments is for not. And yet, take this person out of the picture, and it's suddenly discovered their supposed bullshitting was a linchpin brick in a dam holding back a reservoir of hurt.