There’s an aura of shoplifter interrogation room about my surroundings, and though the people are nice, they’ve all got dark circles under their eyes.

“They didn’t cover this in orientation,” I say. “Kristi, you may have noticed they cover almost nothing in orientation,” Arjun says. It’s true. In mine, there was a weirdly disproportionate amount of time spent promoting the company picnic and almost none on how this place actually works. “And that is because no one knows what is going on,” he says, looking right into my eyes. “This is the most important thing to understand about Amazon. No one knows jack shit.”

Chuck called this [meeting] to talk about CRAP, the new acronym for Cannot Realize Any Profit, focused on items where the accumulated costs of Amazon buying and storing them outweigh any money to be made by selling them. A week ago, I’d never heard of CRAP, and now the need to get rid of CRAP is everywhere; I hear people talking about it in the elevators and in line at Starbucks. This is what happens when Jeff B. has a new idea, Andy explained. It propagates in hard-to-trace ways.

It turns out none of my projects is the right size to actually get done. The double-click issue has to wait for the next bug-bash session, when tiny but annoying stuff gets fixed, and there’s no time for a bug bash this quarter. As for the data dashboard, everyone agrees it needs to happen. “But just high-level scoping the work will take two weeks,” says the guy on the dev team in charge of knowing such things, and even then the work itself will be in line behind other projects like, oh, making sure the site doesn’t crash during holiday peak. Everyone already knows the merchandisers can stay miserable and inefficient and Amazon will survive.

There are so many men here, men from Sloan and the University of Michigan and McKinsey and Deloitte. They’re transitioning to barefoot running. They bought Vibrams last month, and a sous vide machine. They like Big Hairy Audacious Goals, and in college they once saw Modest Mouse five times in a year. They have three kids and a wife with an expired law license because it just made more sense for her to be the stay-at-home parent. They work standing up. They’ve slowly come around on Belgian ales, and Tim Ferriss’s book really made them think. They wish they had more time to read. ... Can they just play devil’s advocate for a second? Can they just pressure test your idea? Can they just push back on that a little? These last three are them saying you are wrong. Sometimes they say it in an Amazon way and sometimes in a man way, though already the difference is getting pretty hard to discern.

“Chuck,” I say, “I didn’t mean to have that effect at all.” “Then can you help me understand why you did it?” “Help me understand” is Amazon-speak for “you are an absolute fucking moron.”

“We probably couldn’t prove with data that a typo hurts the business,” she says, “but they’re still wrong, because they make the brand look dumb, and dumb isn’t trustworthy.”

“I send a monthly update on my team’s activities to twenty of Amazon’s top executives around the globe, and some of them seem to actually read it. Maybe it should worry me that writing that update takes a full afternoon, so broad is my team’s scope. Maybe it should freak me out that my job is so amorphous, and largely unmeasurable, and that as the only person with this exact job I have no peers to learn from or be compared with at promotion time. But I’m too busy grabbing at more, more, more to think about it, my current career strategy being to hit people over the head with the sheer volume and diversity of work I can manage, like a well-spoken, harried octopus.”

I’ve elevated merchandising to less bothersome background noise, and I have enough ideas to keep edging it forward indefinitely. But none of it’s going to add up to a story that will get me promoted, or even necessarily thanked. Granted, that’s the Amazon way: promotions, most notoriously the one I’m in line for, are scarce and mysteriously granted, and still having a job counts as thanks.

Sally carries around a steno pad with a running list of tasks and questions and worries, and during our one-on-ones she flips from page 2 to 6 to 1 to 9, looking for things to ask me about. The sight of her hands fluttering reminds me of fifty-two pickup and my heart begins to flutter along with them, and I feel as if I were doing to her what Coco does to me, as if an endless chain of Amazon women were clutching at the jeans hems of the one above, desperate for reassurance.

Everyone stares silently at their handouts for at least fifteen seconds. Finally, a Web Services director speaks up. “Pipeline,” he says, hands up in a what-else gesture. “Start girls coding as young as possible.” Yesss, I’ve won my bet with Brian that the pipeline would come up right away. One hundred percent of the gender discussions I’ve witnessed at Amazon have involved men agreeing we need to teach girl fetuses to code so we can hire them seventeen years later. This is one area where “move fast” does not apply.

Also, the hard truth is that saying yes to every outlandish request is Amazonian. It may be ruinous and unsustainable, but Amazon as we know it wouldn’t exist without a thousand tiny acts of self-destruction every day.

“Oh, wow,” the PM says. “I had no idea it would take that long.” We have had versions of this conversation many times; my theory is that he gets a factory reset in his sleep each night.