Ideas · Attention · Part one of two
Nobody Falls in Love With the Dissertation
The greats were never in love with the process. They were in love with an idea that had them. The process was built for a reader with almost no bandwidth, and that reader is about to stop needing it.
The part of you no plan can reach
E = P + iI
I is the idea. It is already in the denominator.
In the Reality Equation, Expectation has two parts: Prediction, what habit and history expect, and the Ideal, ideation, your orientation toward what is not yet actual. The idea arrives on its own. Nothing you plan, envision or schedule can add to it. All you can do is answer it with action.
1:40 a.m. · Week 31 of the writing year
You have the graduate school’s formatting guide open in one window and chapter three in the other. The left margin has to be an inch and a half. The table of contents has to match the headings exactly. Somewhere under the stack of forms is a notebook with the idea that brought you here, the one you used to think about in the shower. You have not opened it in a month.
I have a PhD. I was never in love with the dissertation. I was not in love with the coursework, the qualifying exams, the committee meetings, the proposal, the defense, or any other part of the academic process. I was in love with the idea. Deeply. Profoundly. Most people who have been through it will tell you the same thing if you ask them late enough at night.
That is not a complaint about universities. It is an observation about what the process is for, and what it is not for. Once you see that clearly, you can see why it is about to change.
The idea arrives whole
Look at the people we hold up as the greats, and notice what they were actually in love with.
Andrew Wiles, ten years old
In 1963 a boy in Cambridge found a library book about Fermat’s Last Theorem and was captivated. Decades later, as a professor, he spent seven years on it in secret, in the attic office of his Princeton home, telling almost no one. His reason was attention. He said you cannot focus for years without undivided concentration, and too many spectators would have destroyed it.
Stephen Hawking, without a pen
By the mid-1970s Hawking had lost the use of his hands for writing equations. Kip Thorne described what came next: Hawking developed an extraordinary ability to see and manipulate geometry in his head, and much of his breakthrough research relied on it. The process of physics, the blackboard and the page, fell away. The idea did not.
Srinivasa Ramanujan, failing everything else
Ramanujan lost his college scholarship because he would not stop doing mathematics long enough to study anything else. He passed mathematics and failed his other subjects, left college in India without a degree, and kept working alone until a letter to G. H. Hardy brought him to Cambridge.
Grigori Perelman, skipping the journals
Perelman posted his proof of the Poincaré conjecture as preprints on arXiv in 2002 and 2003 and never submitted it to a journal. He declined the Fields Medal in 2006 and the million-dollar Millennium Prize in 2010. The institution came to the idea. He did not go to the institution.
Four very different lives, one pattern. The idea had them first. The process was something they routed around or endured. There is a line, often credited to Carl Jung, that I keep coming back to: people do not have ideas; ideas have people. Every one of these four was had.
The process is written for a narrow reader
So why does the process exist at all? Because the institution that grants the degree cannot see your idea. It can only see artifacts.
An idea arrives whole. It is high-dimensional, simultaneous, felt all at once before it can be said. A committee is five busy people. A graduate school is an office that processes these by the hundred. A human mind, on its best day, holds about four things at once. Most of us cannot drive home and remember six items from the grocery store.
So the institution asks for the idea in pieces small enough to check: a proposal, a qualifying exam, a chapter, a progress report, a signature page, a formatting review, a defense. Each one is an artifact. Each artifact lets the institution update what it expects of you. That is all a milestone is: the only way a low-bandwidth reader can learn, slowly, whether the thing you said was real.
None of that is the work. It is the work translated for a reader who cannot take it in whole. The dissertation is a compression format. It is necessary for credentialing and irrelevant to the idea.
Figure 01
Two bandwidths
The idea
Whole, simultaneous, high-dimensional. Felt before it can be said. It arrives complete.
↓ serialized for a reader who holds about four things at once ↓
The process
- Proposal
- Coursework
- Qualifying exam
- Chapters
- Progress reports
- Formatting review
- Signature page
- Defense
An idea does not need a plan
Here is the part that took me longest to understand. Nothing about the idea is waiting on your plan. Nothing is waiting on your vision. The vision is already there. The blueprint is already there. That is what it means for an idea to have you: it arrived complete, and your bearing keeps returning to it.
In the Reality Equation the idea lives in the imaginary part of Expectation, the Ideal. And the framework has one rule that matters here more than any other: no one can touch either term. You cannot reach into your own Expectation and add more idea to it. You cannot plan your way into the denominator. The only thing anyone can act on is the moment in front of them. Action leaves artifacts, artifacts become history, and history is the only thing that ever teaches Expectation anything. As I put it in September, consciousness does not manufacture the future. Consciousness manufactures history.
That is why the only answer to an idea is to act. To do. Not to plan, not to vision. A project plan is a story about the future, and the framework already has names for stories about the future: hope when they run better than expected, fear when they run worse. A Gantt chart is hope with columns.
It is also, I think, why Elon Musk seems to compress time. Walter Isaacson’s biography records the five-step algorithm Musk repeats to his teams until they mouth along: question every requirement, delete any part or process you can, simplify and optimize, accelerate cycle time, and only then automate. Notice what is not on the list. There is no step for planning the work. When xAI built its Colossus supercomputer in Memphis, Musk said it went from start to finish in 122 days. You do not get there by managing each task on a two-week timeline. C. Northcote Parkinson explained why in 1955: work expands so as to fill the time available for its completion. A two-week timeline is a request for two weeks of work.
What AI takes, and what it leaves
Now change one input. Everything in the lower panel of that figure is the kind of work AI does well and does tirelessly: formatting to a style guide, assembling a literature review, tracking citations, filling forms, drafting the progress report, writing the status email to the committee. Serializing an idea for a narrow reader is exactly what a machine with no attention budget is for.
The non-obvious consequence is not that research gets easier. It is that the process stops being able to hide the absence of an idea. For a century, a student could spend five years busy with milestones and come out the other side with a degree and no idea that had them. When the milestones cost nothing, the only thing left to examine is whether the idea is real and whether you are its person. The bottleneck moves to the one part that cannot be copied: you, in love with the thing.
May 2029 · A defense
The candidate’s AI assembled the document overnight from three years of lab notes, code, emails with her advisor and the papers she argued with. Nobody on the committee asks about formatting. Nobody asks about chapter two. They spend three hours on the idea itself, pushing on it from every side, and what they are really examining is the person: whether she can see it whole, defend it live, and say where it breaks.
Five predictions
The defense becomes the degree.
As written documents become cheap to produce, the live examination becomes the only part of the doctorate that tests a person rather than a pipeline. I expect leading programs to move their weight decisively toward oral examination before 2032, and some to write it into their rules.
Time to degree falls first where paperwork was the bottleneck.
Fields whose delays were writing, review and administration will see the doctorate shorten first. Lab and field sciences will move slowest, because cells, soil and telescopes do not care about AI.
“What idea has you?” becomes the admissions question.
Programs and employers will care less about whether you can produce the artifacts, because everyone can, and more about whether an idea has you. The candidate who can say what it is in one sentence, at a dinner table, wins.
More landmark work arrives the Perelman way.
Posted openly, checked in public, recognized by institutions after the fact. Journals and degrees become ways of certifying what already happened rather than gates the work must pass through first.
The advisor’s job shrinks to the part that matters.
Recognizing whether an idea is real, and whether this person is its person. That has always been the rarest skill in academia. It is about to be nearly the only one.
Where I am in this
I have lived both sides of it. I did the process, all of it, and the only part I remember with any love is the idea. Today this blog is how I answer the ideas that have me, and I publish almost every day. This piece began as a paragraph I sent at two in the morning, unplanned, because the idea showed up and wanted to be written. My two favorite instructions to my AI are four words long: You do it. You decide. Everything that is serialization, I hand off. What I keep is the part nobody can do for me: paying attention to the idea, and acting on it.
The ledger
- Already true
- Every step of the written process, from literature review to formatting to the status update, can now be drafted by AI. And the greats have always routed around the process when the idea demanded it: Wiles worked in secret, Perelman skipped the journals, Ramanujan failed everything but mathematics.
- What has to happen
- Institutions have to admit that the document no longer proves the work, and move their weight to the live examination of the person. That is a cultural change, not a technical one, and universities change slowly.
- Where I am probably wrong
- The process does something I have undersold here: it forces a finished argument out of a beautiful feeling. Plenty of ideas that feel whole at two in the morning fall apart in chapter three, and the discipline of writing them down is how you find out. If serialization is where an idea gets tested, handing it to a machine could produce more confident nonsense, not more Perelmans. Then the answer is a harsher defense, not a softer process.
Go back to the student at 1:40 a.m. The margins can be somebody else’s job now. The notebook was always the point. The idea is still in it, still whole, still waiting for a mark.
It never needed a plan. It needed her.
Background
- Stephen William Hawking: A Biographical Memoir (Royal Society memoir, with Kip Thorne’s account)
- Abel Prize 2016: Andrew Wiles biography
- NOVA, “The Proof” (transcript, with Wiles on undivided concentration)
- Encyclopedia.com: Andrew J. Wiles
- MacTutor: Srinivasa Ramanujan
- Physics Today: Mathematician turns down $1 million prize for proving the Poincaré conjecture
- Walter Isaacson, Elon Musk (2023), on “the algorithm”
- C. Northcote Parkinson, “Parkinson’s Law,” The Economist, 1955
- Nelson Cowan, “The magical number 4 in short-term memory,” Behavioral and Brain Sciences, 2001
- Ideas Have People
- You Do Not Choose the Future. You Choose History.
- Nobody Wants to Move the Card (part two)