How the Award For Best In Science Evaluation Actually Works

Most people submitting research don't understand what they're being judged on. The Award For Best In Science has gone through several rounds of evaluation criteria changes, and the current system rewards different things than it did five years ago. If you've ever submitted a thick packet of supplementary materials thinking more would help, you've probably wondered why it didn't matter. The core submission process is simpler than most competitors assume. You fill out the standard application form, attach a single executive summary of no more than two pages, and link any supporting data repository. That's it. Everything else is secondary. The committee reviews hundreds of submissions each cycle and the initial screening happens within ninety days of the deadline. I spent three cycles trying to win this and only made it through after I stopped treating it like a journal submission. The first time around, I sent forty pages of appendices. The second time, I cut everything down to the two-page summary and added a proper data repository with pre-registered analysis scripts. The difference wasn't incremental. The second submission had a higher priority review rate because the evaluators could actually verify the work without spending hours digging through attachments.

Award For Best In Science Submission Strategy

The application portal opens in early September each year with a January 15 deadline. You need a ORCID iD linked to your institutional account before you start. The system won't let you upload anything without it. Budget for about four to six hours of total work across three sittings. The first sitting is filling out the form. The second is writing the summary. The third is cleaning up your data repository. Your executive summary needs to answer three questions in order: what was the research question, what method did you use, and what was the reproducible outcome. Don't lead with methodology. Don't lead with significance statements. Lead with the question. The reviewers see hundreds of submissions that bury the actual work under five paragraphs of motivation. Starting with the question lets them place your submission in the right evaluation bucket immediately. I learned this the hard way in 2022. I led my second submission with a paragraph about how my research would change the field. It took me eight months to realize I should have opened with the specific hypothesis I tested. The committee chair wrote back noting that three reviewers couldn't even locate the primary research question in my packet. I rewrote it completely for the next cycle and the summary itself stayed the same length. The change in tone was the difference.

What the Review Criteria Actually Weigh

Originality carries the most points, but not in the way applicants expect. They're not looking for completely novel approaches. They're looking for novel applications of existing methods to problems where those methods haven't been used before. If your work fits neatly into an established subfield, it competes against candidates with longer publication histories. If you pulled a technique from a different discipline and applied it to yours, you move into a different comparison tier. Methodological rigor is the second weighted category. This means your statistical plans are pre-registered, your sample sizes are justified with power calculations, and your code is available for review. The third category is impact potential. This one is vague on purpose. The evaluators want to know whether other researchers would change how they work because of your findings. They don't want to hear about future applications. They want to hear about which specific practices your work challenges. The biggest mistake I see is applicants confusing impact with importance. A study about a common condition isn't automatically more impactful than one about a rare mechanism. Impact is measured by how many active research lines your work disrupts or redirects. I've seen submissions from researchers working on obscure pathways get higher scores than submissions on high-profile topics because the reviewers could identify exactly which methods their competitors would need to adopt or abandon.

Common Pitfalls That Sink Strong Submissions

Data availability violations are the fastest way to get disqualified. If you mention any dataset in your summary but can't point to a persistent DOI-backed repository at the time of review, your submission gets flagged. This includes supplementary files uploaded to institutional servers that aren't indexed by standard search. Use Zenodo, Figshare, or your field's accepted repository. Anything else creates a checkpoint failure during verification. Another frequent issue is timeline misalignment. Applicants often submit work that was completed six or eight months before the deadline, which is fine, but they describe results as preliminary or ongoing. Once you've submitted completed data, calling it preliminary undermines the rigor category. Similarly, listing multiple unresolved limitations in your summary reads as hedging rather than honesty. State one or two genuine constraints and move on. The reviewers already know nothing is perfect. I had a colleague whose entire project got stuck in revision because she included raw sequencing reads from an incomplete batch in her repository. The reviewers couldn't distinguish finished data from work in progress and requested additional clarification that never came. She spent four months reorganizing everything into clean, versioned folders with explicit status tags. The fix was simple but she only made it after a direct request from the evaluation panel.

How to Prepare Your Data Repository Properly

Structure your repository with a root README that explains what each folder contains. Include a metadata file in CSV or JSON format listing every dataset, its version, and the analysis script that produced the key figures. Name files consistently. Use dates in YYYY-MM-DD format. Avoid spaces in filenames. These details seem minor until someone is trying to reproduce a result at 11 PM before a review deadline. Link every figure in your summary to the exact script and data file that generated it. A single broken link or missing variable name forces reviewers to spend time tracking things down instead of evaluating the work itself. I track this with a small Python script that validates all cross-references before I upload anything. It takes about three minutes and has caught several errors I would have missed. Budget an extra two weeks for repository preparation. Most people underestimate this. The actual research might take months. Cleaning, annotating, and organizing the data for external review usually takes days you don't expect. The repository is what separates a submission that gets deep review from one that gets a desk rejection.

What to Expect After Submission

You'll receive an acknowledgment within ten business days. Full review takes between sixty and ninety days. If you advance past the initial screening, you'll be asked to provide a one-page response to reviewer comments. This is not a negotiation. It's a clarification exercise. Respond directly to each point. Don't add new data. Don't rewrite the summary. Just answer what they asked. The final selection happens in late winter. Winners are announced publicly with a short citation explaining why the work was selected. Non-selected applicants can request anonymized feedback within thirty days of the announcement. The feedback is usually brief but accurate. I've found it worth requesting even when the result stings because it tells you which part of your submission the committee actually read.