Crossgrain No.2 - Nobody Copied Anybody
PhD Statement of Purpose
A few application seasons ago, three students came to me separately. Different backgrounds, no knowledge of each other. All three were applying to the same department at Harvard. All three had written a paragraph explaining their fit with the same famous professor there.
All three cited the same paper of his as evidence of that fit.
It was the paper at the top of his profile page.
Nobody copied anybody. They didn't need to. They had all run the same procedure: open the faculty page, take the first result, wrap it in the word "aligns." Three people who had never met produced interchangeable evidence of a unique intellectual connection. That is what a template does.
As a coach, I have read Statements of Purpose (Personal Statements) since October 2008. I can usually tell by the second paragraph whether a committee will finish the page. Here is the thing I learned: every professor reading your file is asking one question. Not consciously, maybe. Not in these words. But it animates every minute they spend on your application.
If I bring this person into my group for five years, will I regret it?
Most SOPs answer a different question entirely. They answer instead: "What have you done, and what do you want?"
This is the format: a chronology of undergraduate research, a paragraph naming professors whose work "aligns with my interests," a closing statement about contributing to the "esteemed community." Every sentence is true. Every sentence is safe. And every sentence tells the professor nothing about the one thing they need to know.
But the professor already knows what you've done. It's on your CV two pages earlier. Repeating it with adjectives — "I passionately investigated," "I was deeply involved in" — doesn't add information. It adds décor. A CV narrated aloud is still a CV.
The real question sits underneath: not what you did, but what you learned that your labmate didn't. Not what you want to study, but what you see in your field that others do not see yet.
What a professor is actually reading for
Four things — which, you'll see, are one thing.
One. Is this person going to fight me, or work with me?
Not intellectual disagreement. That's healthy. The "fighting" that exhausts a professor is social overhead: the student who needs constant reassurance, who resists feedback, who turns group meetings into ego management, who blames the equipment, the cohort, the field — never the hypothesis. Every professor has hired this person once. They only make that mistake once.
This quality can't be stated. "I am collaborative" is a claim no reader can verify, and an unverifiable claim reads as decoration. It has to be specific: in how you describe your past work (do you mention your co-authors? do you say "we" when it was a team?), in how you narrate failure (do you take ownership, or does the equipment "prove unreliable"?), in how you engage with the professor's work (are you building on it, or bowing to it?).
Two. Does this person have their own research agenda, or just a résumé?
Here is the fork. Most applicants write some version of: "I am interested in Professor X's work on perovskite stability. I would like to contribute to this area." Here is the alternative:
Many labs in my field celebrate efficiency benchmarks measured in controlled environments. My question was: why do even 22% efficient devices fail within six weeks on an actual human body? After two failed hypotheses, the answer turned out to be that we were measuring the wrong variables — stability data collected flat, on devices that spend their working lives being bent. So I inverted the standard approach. Instead of eliminating crystallographic defects, I placed them deliberately, as molecular anchors at predicted stress points. The devices now hold stable performance past 3,200 flex cycles under varying humidity at body temperature. This is where Professor X's work becomes a map. Her tunneling-barrier equations predict minimum energy barriers for charged vacancies at 0.4 nm junction points. That is where my anchor defects perform best. Her models explain my results better than my own did. That is the question I want to spend the next five years on.
Same field. Same professor. Three things changed.
The fit became falsifiable. The 0.4 nm claim could be wrong. Professor X could read it and disagree. A claim that can be wrong can prove you understand her work. "I would like to contribute" risks nothing and proves nothing.
The paragraph reversed direction. Not "your work relates to my interests" — "your equations predict my data." One sentence like that outweighs three named professors, because it can only be written by someone who actually read the papers.
Two failed hypotheses were mentioned by name. Committees believe evidence that you kept going after failing twice.
Three. Will this person place well after graduation?
A professor's reputation compounds through alumni placement. Every hire is a bet on the group's future standing. The signal here isn't "I have a 4.0 GPA." It's whether your questions are aimed at the field's next problem or its previous one. A student who can articulate why their subfield's current consensus might break — with evidence, not attitude — reads as a future job candidate someone will want to hire. A student who summarizes the current consensus reads as a competent technician. The difference is one paragraph of genuine intellectual positioning.
Four. Can this person plan without reaching for buzzwords?
Read a stack of SOPs and you will find: "multidisciplinary approach," "advancing the field," "sustainable solutions," "novel insights." These words do not communicate ambition. They communicate that the writer hasn't figured out what they actually want to do. Nobody who knows their next two experiments reaches for "advancing the field." They reach for the name of the compound, the temperature range, the failure mode.
These four collapse into one: does this person have their own question?
If you have your own question, you won't fight the professor. You'll be too busy chasing the answer. If you have your own question, your placement path clarifies because the question points somewhere. If you have your own question, three professors named on a page become one — the one whose tools can actually test your hypothesis.
A Statement of Purpose that answers the real question, "Will I regret hiring this person?", doesn't need to name three professors or list four techniques. It needs to demonstrate that the writer has seen something in the field, formed a hypothesis about it, and identified the specific tools this department has that can test that hypothesis. That's the whole genre.
Most applicants never write that essay. But it can be taught. It has been.
The full record, by cycle
2023 — Arizona State U (Tempe), Caltech, Cornell, Harvard, Stanford, U of Chicago, U of Illinois (Urbana-Champaign), U of Michigan (Ann Arbor)
2024 — Boston College, Carnegie Mellon, Dartmouth, Northwestern, U of British Columbia, UC Berkeley, U of Chicago, U of Michigan (Ann Arbor), U of Virginia, Yale (two students)
2025 — Brooklyn College, Columbia, Stanford, U of Chicago, U of Illinois (Urbana-Champaign), U of Pennsylvania
2026 — MIT, U of Pennsylvania
The committee still asks the same question. The only variable is whether your essay answers it, or answers the easier one instead.
The diagnostic — $300 USD
One session, 60 minutes, on your current draft. Google Meet. Submit the draft before the session.
What the diagnostic covers:
- Which question your draft currently answers — theirs, or the easier one
- Whether a committee could tell your essay from the other applicants'
- Whether what's missing is fixable in four sessions
- How to make better decisions
The program — four sessions, $1,700 USD
One session per week, 60 minutes each, Google Meet. Total with the diagnostic: five sessions, $2,000.