This piece was originally published in the Good Science Project newsletter on December 14, 2024.In 1999, Google’s founders couldn’t sell their company for $1 million.1 In 1995, Katalin Karikó faced demotion at the University of Pennsylvania, her mRNA research repeatedly rejected for funding.2 But today, Google’s parent company is worth over a trillion dollars, and Karikó’s work formed the foundation for COVID-19 vaccines that helped end a global pandemic.These stories illuminate a fundamental paradox in how we approach scientific progress and innovation funding: the highest impact often has nothing to do with predicted impact at the time.Our current system of scientific funding operates on what might be called the “train schedule” model of progress: we expect innovations to arrive on time, following predetermined routes, with clear destinations. This manifests in several ways:Funding agencies demand detailed roadmaps and virtually guaranteed outcomes before work beginsGrant proposals must specify exact deliverables and timelinesPolitical pressure pushes agencies toward “safe” incremental researchResearchers even need to demonstrate preliminary results before receiving funding to generate those very resultsThe irony is stark: we’ve created a system that would have rejected many of history’s most transformative breakthroughs. Consider examples like these (there are many more):Douglas Prasher’s work on fluorescent protein, crucial to Nobel-prize winning research, went unfunded, leading him to drive a courtesy car for a living3Robert Langer at MIT faced rejection on his first nine grant proposals for work on biodegradable polymers4The team that discovered how to manufacture human insulin was rejected because their work seemed “extremely complex and time-consuming”5Craig Venter’s proposal for whole genome shotgun sequencing was rejected with claims it wouldn’t work—after he had nearly completed the genome6The story of American innovation reveals a stark contrast between