Case Study 2 — Two Proposals, One Panel: Why the Better Science Lost

A composite, fictional-but-realistic scenario. The point of Case Study 1 was the anatomy of a single great page. The point here is comparative and uncomfortable: when two proposals compete, the one with stronger science does not automatically win. Often the one that's easier to fund wins — and "easier to fund" is a writing property. This case also shows the Specific Aims logic transferring to a non-NIH context (an NSF-style Project Summary with explicit Broader Impacts), so you see the structure is portable.

The setup

A review panel is scoring proposals for an early-career research program. Two land in the same reviewer's pile.

Proposal A — Dr. Tomás Herrera. A genuinely original idea: a new computational method for predicting how materials behave under extreme pressure, with implications for battery design. The underlying science is, frankly, the most novel thing in the reviewer's stack. Tomás is brilliant. He is also a first-time proposer who wrote the document the way he thinks — discovery-order, complete, dense.

Proposal B — Dr. Wei Zhang. A solid, somewhat-less-novel idea: improving an existing class of materials simulations to handle a specific industrially important case. Good science, not earth-shaking. But Wei has written three proposals before, lost twice, and learned from the critiques.

Same reviewer. Same fifteen-minute window each. Watch what happens.

Proposal A: how strong science gets triaged

The reviewer opens Tomás's Project Summary.

"The computational modeling of materials under extreme conditions has been an active area of research for several decades. A variety of approaches have been developed, including density functional theory, molecular dynamics, and machine-learning interatomic potentials, each with its own strengths and limitations. In this proposal, we develop a new method that addresses several of these limitations through a novel combination of techniques, which we describe in detail in the following sections…"

The reviewer, twelve proposals deep at 10 p.m., reads this and learns nothing in the first 80 words — it's a survey of a field she teaches. She keeps reading, hunting for the actual method and what's new about it. It's there, eventually, three paragraphs down, stated with hedges: "may offer improved accuracy," "could potentially enable." The aims are phrased as activities: "We will implement the algorithm," "We will benchmark against existing methods," "We will apply the method to several systems." The Broader Impacts paragraph is one generic block: train students, publish, broaden participation — no specifics.

The reviewer's honest reaction is not "this science is weak." It's: "There's something good buried in here, but I can't quickly tell what's new, the aims are vague, and I'd have to fight the document to advocate for it." In the panel, when her turn comes, she struggles to summarize it crisply — and a proposal the reviewer can't summarize crisply is a proposal that doesn't get championed. It scores in the middle. In a program funding the top ~12%, the middle is a rejection. The most novel idea in the stack does not get funded, because the reviewer could not extract it fast enough to defend it.

Proposal B: how solid science gets funded

The reviewer opens Wei's Project Summary.

"Lithium-metal batteries could roughly double the energy density of today's cells — but they fail unpredictably, because no existing simulation can model the dendrites that short them out at industrial timescales. We have built one that can. Our prototype simulator reproduces measured dendrite growth in three test systems to within 8% (preliminary results), at 50× the speed of the standard approach. This proposal will harden that prototype into a validated, openly released tool, establish its accuracy across the materials industry actually uses, and demonstrate it on a battery system our industrial partner cannot currently model."

The reviewer, same hour, same fatigue, reads this and gets it in fifteen seconds. Problem (batteries fail, can't simulate why), pivot ("we have built one that can"), data (8%, 50×), and a three-part plan that maps to three aims. Wei's aims are objective-first: "Establish the simulator's accuracy across the five most-used electrolyte classes…," "Determine whether the method scales to industrially relevant cell geometries…." Each has an expected outcome and a pitfalls-and-alternatives note. The Broader Impacts are concrete: the tool released open-source under a named license, a partnership with a specific industry consortium, two students trained in skills in shortage, and an undergraduate research module piped through the PI's existing summer program.

Is Wei's science as novel as Tomás's? No — and the reviewer knows it. But she can summarize it in one sentence, she trusts the plan, and she can defend it in the room. It scores near the top. It gets funded.

The lesson, stated plainly

This is not a story about gaming the system, and it is not a claim that writing beats science. Over many proposals, strong science wins more than it loses. But on the margin — and funding is all margin, because the cutoff is brutal — the proposal that's easier for a tired reviewer to understand and champion has a real, repeatable advantage. Tomás didn't lose because his idea was bad. He lost because he made a brilliant idea hard to receive, and a reviewer can only advocate for what she can quickly grasp. That is the chapter's thesis with names attached: ideas rarely get rejected; writing about ideas gets rejected.

Run the diagnostic on Tomás's proposal and every failure is a writing failure you now know how to fix:

Tomás's good news is the same as everyone's: every item in that table is fixable in revision, by a writer who reads his own draft the way a tired reviewer will. If he rebuilds the Project Summary around the four moves, that brilliant idea becomes the easiest in the stack to champion — and then its novelty, finally legible, works for him instead of hiding from him.

The takeaway to carry forward. When you finish your proposal, do not ask "is my science good?" Ask the reviewer's question: "Could a tired stranger, reading this once at 10 p.m., state in one sentence what I'm proposing and why it matters — and then defend it to a panel?" If the answer is no, the best science in the world won't save you. Make the idea receivable, and let the science do the rest.

Back to: Chapter 17 · Case Study 1 · Exercises · Key Takeaways