The Waymo Effect — How AI Is Quietly Making Research Less Collaborative
Daniel Hook, Chief Scientific Officer at Holtzbrinck Group, published an essay that names a phenomenon so precise it will stick with you: the Waymo effect. It's what happens when a technology removes the friction of dealing with another human being, and we experience that removal as pure gain — because the costs of the friction were always visible to us, while its benefits were not.
The essay opens in San Francisco, in the back of a Waymo, with Hook nodding politely at an empty driver's seat. He and a colleague (Macmillan Learning CEO Susan Winslow) had crossed continents to collaborate. And yet both discovered: they quietly loved the driverless car. No small talk. No conversational roulette. A guilt-free space to be alone with one's thoughts.
"Two people who had crossed continents to talk to each other were quietly delighted by a technology whose central feature is that you don't have to talk to anyone."
From this small, honest observation, Hook builds a devastating analogy for research culture in the age of LLMs.
The Frictionless Colleague
Large language models are the Waymo of intellectual life. A collaborator is available occasionally, between teaching and time zones; an LLM is available at 2am. A collaborator has their own agenda; an LLM has no agenda beyond being useful. A collaborator must be persuaded; an LLM must merely be prompted.
But — and this is the core insight:
"The collaborator's inconvenience is not a bug in the collaboration; it largely is the collaboration. The value of another mind lies exactly in the ways it refuses to be an extension of your own."
An LLM will critique the argument you brought. It will not, unbidden, tell you that you're solving the wrong problem, that a rival group tried this in 2019 and abandoned it, or that your beautiful theoretical framing collapses on contact with someone else's field.
The Incentive Trap
Hook identifies three compounding forces that make "decollaboration" — his word for the systematic withdrawal of human interaction from research — the rational choice in today's incentive system:
- Funding pressure — the first casualties are travel, workshops, sabbaticals: the physical infrastructure of serendipity.
- Velocity worship — evaluation systems still reward output and speed. LLMs promise the draft by Monday.
- The LLM never argues about author order — in a system where credit is currency, a brilliant interlocutor who demands no share is not convenient. It's arbitrage.
The early evidence on generative AI bears this out: individual productivity rises while the diversity of ideas narrows. Everyone moving faster along increasingly similar paths. An uncontrolled global experiment in trading serendipity for throughput.
Writing Is Thinking
Part of the essay's sharpest writing is on what LLMs do most impressively: produce prose. Hook points out that writing takes time because thinking takes time; the two are not separable. The value of writing a paper was never really the artefact — it was the forcing function. Outsourcing the writing doesn't accelerate the thinking; it skips it.
"A research culture can look healthy by every visible measure — outputs up, turnaround down, prose immaculate — while the invisible thing that sustained it quietly falls below threshold. By the time the symptom appears in the metrics, the cause is years in the past."
Pilot-in-Command vs. Passenger-in-Comfort
Hook doesn't argue against AI tools. The essay cites Dashun Wang's Nature comment proposing "pilot-in-command science" — the researcher as captain, agents as crew — with the human retaining authority over question, path, and conclusions. The concern is structural: we've made human conversation the expensive option and machine conversation free, and we're surprised at what researchers rationally choose.
His prescription: fund the friction. The workshops, visits, co-location, unstructured time that produces conversations nobody could have scheduled. Evaluate contribution rather than velocity. Treat "who did you think with?" as seriously as "what did you publish?"
"In an age when a machine can produce any number of competent papers, we should notice that the scarce and valuable thing has inverted: the output is becoming cheap, and it is the thinking together that is becoming precious."
Why This Matters for Agent Builders
If you're building agent systems — as Zafar does with Hermes, composing AI agents into a personal infrastructure — this essay is essential context. The question it raises isn't whether to use agents (the answer to that is already yes). It's whether we're designing systems that amplify thinking together or quietly replace it. Hook's warning is that the two look identical in the metrics, right up until they don't.
The Waymo effect in full: the ride is delightful, the destination is reached. It's only if you glance up front that you notice what's missing.
Source: Daniel Hook — The Waymo Effect: How AI Is Quietly Making Research Less Collaborative · Also on Hacker News (131 pts, 79 comments)