Issue 19  January 9th 2026

The theme of this issue is Variance.

Modern science is extraordinarily well organized. It is professionalized, credentialed, and scrupulously filtered. Careers follow regimented ladders. Funding flows through standardized mechanisms. Evaluation processes are explicit, legible, and widely shared. And yet, despite all this order, something feels off. The system produces a great deal of activity, but less surprise than one might expect given our collective capabilities. What's missing is variance.

We didn’t always lack variance—professionalization and standardization whittled it away. Reforms like standardized training, credentialing, centralized funding, and formal peer review rightly improved rigor, but they unintentionally pushed out the messy, unconventional attempts that used to broaden the space of exploration.

Those forces persist today: funding favors projects that look plausible ex ante, career incentives reward coherence and continuity, and review processes converge toward consensus. As a result, variance is filtered early—entry points narrow, expectations are standardized, and risk is shifted onto individuals rather than absorbed by institutions. This heuristic-driven filtering tends to produce safer science, not necessarily better science.

Why does this matter? Scientific progress is not linear. Major advances emerge from a wide distribution of attempts—most of which fail, stall, or lead somewhere unexpected. The role of a healthy ecosystem is not solely to predict which ideas will succeed in advance, but also to ensure that enough different ideas are tried in the first place. Variance is the raw material from which discovery is selected.

A higher-variance scientific ecosystem would look different—not chaotic, but more permissive. What a high-variance system might look like:

• Lower the barrier to attempt. Let more people try things earlier, before their ideas are fully formed or credentialed.
• Relax categorical demands. Don't require new work to fit neatly into existing fields or frameworks from the start.
• Plan for failure. Treat a higher rate of failure as a feature, not a flaw—it's the cost of sampling widely.
• Decouple early failure from career damage. Make it possible to try something that doesn't work without permanent professional penalty.
• Reward high-quality failure with status. As in Silicon Valley, celebrate ambitious attempts that didn't pan out as valuable contributions to collective knowledge, not as marks of shame.
• Defer return-on-investment demands. Apply outcome-based evaluation later, once variation has had room to express itself—not at the point of entry.

This doesn’t mean a return to amateurism. We can address the danger of excessive homogeneity through a system that engineers variance—mechanisms that invite bold, unconventional work without sacrificing scientific rigor or professional standards. If we want more breakthroughs, we need institutions that tolerate apparent disorder—platforms that host variance responsibly, deliberately, and at scale—rather than trying to eliminate it.

With the theme of variance in mind, we're excited to announce that Project Aeon has been selected for an Emergent Ventures grant to help build the platform. Given Emergent Ventures' role in creating Fast Grants and shaping the field of metascience, their support means a lot to us. Learn more about Emergent Ventures here.

Finally, if you are enjoying this newsletter, please forward it to someone else whom you think might benefit. And please consider joining our community, where we discuss ways to invigorate science and innovation, share job openings and funding opportunities, and more.