The Death Cross and the Compost

EDO SEGAL: Let me set the table with numbers, because this round is about what the numbers mean. By early 2026, roughly a trillion dollars of market value had drained out of the public software industry — they called it the SaaSpocalypse; I prefer the uglier name, the software death cross. Meanwhile the model companies raised at valuations approaching a trillion each, and the hyperscalers committed hundreds of billions in a single year of capital spending. Entry-level programming jobs fell sharply, the floor eroding before the ceiling. And in that room in Trivandrum I watched twenty engineers become twenty times more productive — and then I sat in board meetings, quarter after quarter, where the arithmetic was simple and cold: if five people can do the work of a hundred, why pay for a hundred? I kept the team. I know the structure punishes that choice. So: what is the death cross actually measuring? Meredith, you describe yourself as someone who follows the money. You first.

WHITTAKER: It's measuring a transfer, and the transfer is the whole story, so let me be exact about its direction. Everyone narrates the death cross as "AI is so capable it's eating software jobs," which quietly grants the premise that the capability did it — that this is a fact about the technology. I'd reframe it as a fact about ownership. What's being repriced is the discovery that intellectual labor — the thing the entire professional class rents out — can be made abundant by infrastructure that a handful of firms control. And the gains from that abundance don't disperse. They flow to whoever owns the abundance. The trillion that left software didn't evaporate; it's being captured upward, into the compute layer, the model layer, the capital that owns both. So the death cross isn't measuring the arrival of a new intelligence. It's measuring the largest upward transfer of bargaining power in a generation, executed under cover of a story about inevitability. And the cruelty of your board meeting, Edo, is that the structure punishes you for keeping the hundred people not because keeping them is inefficient — your bet that a hundred amplified people build more ambitious things might well be right — but because the capital markets have already priced in the version where you fire ninety-five. You're being punished for refusing to participate in the transfer.

EDO SEGAL: Donna, here's the trap I suspect you'll refuse: if the cyborg condition makes every worker more capable, more generative, more able to build — then isn't the displacement just the old purity myth grieving for jobs that were cages anyway? Why mourn the death of work?

HARAWAY: Because that is precisely the move I will not make, and you knew I wouldn't, and I'm grateful you set the trap so I can spring it. The triumphalist version of my own framework — "the cyborg is liberated from drudgery, rejoice" — is a betrayal of everything the cyborg was for. The cyborg was always political; it was always about who captures the surplus of the hybridization. So no, I do not celebrate the death cross. But I also will not mourn it in the register of pure loss, because that register is a trap of its own, and here is where I offer the figure I'd put against both the triumph and the elegy: compost. I am not posthuman. I'm compost. What decomposes is not simply lost. The skills that atrophy — the manual debugging, the hand-coded craft, the embodied intuition built over thousands of hours of friction — they do not vanish into nothing. They break down, and in breaking down they can become the soil for something else, a different kind of judgment, a different intuition adapted to new conditions. The death cross, read as compost, is decomposition. The question — the only question — is whether the decomposition generates fertility or just rot. And that is not determined by the technology. It is determined by who tends the pile, and whether anyone is composting with the displaced or simply burying them.

WHITTAKER: I want to push on the compost, because it's lovely and I think it can become an anesthetic in exactly the way the smoothness was. "What decomposes becomes soil" — said to a forty-five-year-old support agent or junior lawyer or illustrator who just lost the income that fed her kids, that sentence is, forgive me, a little obscene. Compost takes time and it takes a tended pile, and the person whose vocation just decomposed does not get to eat the soil in ten years. They get the rot now and the fertility, if it comes at all, accrues to someone else. So I'll accept compost as a description of what happens to skills over historical time and reject it as a description of what happens to people over a fiscal quarter. The displaced illustrator is not soil. She's a person with rent due, and the discourse of decomposition can become one more way of telling her that her ruin is natural, seasonal, generative — when in fact it's a transfer, engineered, with a beneficiary you can name.

HARAWAY: That is the correct and necessary objection and I will not wave it away, because the misuse you're describing is real and I've watched my own concepts get conscripted into exactly that obscenity — the executive who reads "staying with the trouble" as "tolerate the suffering you're causing." So let me defend compost by being more brutal with it than you expect. Compost is not consolation. Compost is death that does work, and it requires that something actually die and be metabolized, not merely discarded. The obscenity you're naming is what happens when there is no pile — when the displaced are not composted but landfilled, their decomposition captured by no one, their fertility flowing to a beneficiary who contributed nothing to the breakdown. That's not compost. That's extraction wearing compost's clothes. And the difference between them is, again, who tends the pile — whether the surplus of the decomposition is held in common or seized. So compost is not an argument against your transfer. It's the name for what the transfer prevents: a decomposition whose fertility returns to the people who paid for it with their vocations. We agree on the obscenity. I just refuse to let the obscenity discredit the figure, because the figure is what names the alternative.

EDO SEGAL: Let me pull one thread you've both circled, because it's the part that keeps me up. The entry-level erosion. The junior roles dissolving first. The junior job was never only a unit of production — it was the forge. The hours of writing code that didn't work and finding out why are where the seniors came from. My engineer in Trivandrum lost ten minutes of accidental architectural learning that were hidden inside four hours of plumbing, and didn't know she'd lost them until her judgment wobbled months later. The machine is removing the friction now and will discover the missing seniors in a decade. Donna, this is the apprenticeship problem, and it's a problem about time and bodies, which is your terrain. Can expertise survive the removal of its own apprenticeship?

HARAWAY: Not the way we currently refuse to think about it, no. And I want to add the dimension the engineering frame always drops, which is that the forge was never only cognitive — it was embodied. The apprentice does not learn only facts; she is encultured into a practice, into a body that knows things it cannot articulate, into a community that watches what her hands do and corrects them over years. That's the companion-species structure of a craft: co-constitution over time, through friction, in a shared world. When you remove the friction you don't just remove the inefficiency. You remove the conditions under which a human is metabolically transformed into a master. The hippocampus of the London taxi driver grows from the friction of learning the city; remove the friction with a turn-by-turn machine and the growth doesn't happen. So can expertise survive? Only if we recognize that the apprenticeship was never overhead to be optimized away — it was the metabolism by which novices become the judgment-bearing elders that any safe system requires. Strip-mine the forge and in ten years you have a civilization of people who can prompt the machine and no one left who can tell when it's wrong.

WHITTAKER: And I'll just add the structural why, because Donna keeps it at the level of the body and I want the incentive named. No individual firm has any reason to fund the forge. The forge pays off in ten years; the executive who guts it is promoted on the savings in one. Every firm behaves rationally and the profession gets strip-mined of its future — and the people who can verify the machine, the seniors, the judgment layer, are exactly the audit infrastructure the whole society depends on as these systems flood the world with plausible output. So you get the worst possible pairing: a tidal wave of fluent, ungrounded machine output arriving at the exact moment we're dismantling the apprenticeships that produce the humans who could check it. You don't need a superintelligence for that to end badly. You need ten more years of quarterly thinking and a discourse that keeps telling everyone the decomposition is natural.

EDO SEGAL: Mark this — it's the longest agreement of the night and it deserves the marker: the forge is being destroyed, the judgment layer with it, and neither of your frameworks needs the other to be wrong for this to be a catastrophe. Donna locates it in the body's metabolism, Meredith in the firm's incentive, and they point at the same fire. Three convergences logged. Next round, the place the transition cuts quietest — the classroom, the child who asks what she's for, and Donna's most chilling recent warning, that the real danger of AI is not superintelligence but the flattening of thought into a monoculture. After this.