Optimizing Resource Allocation: A Proposal for Utility-Based Funding Applications
Coverage of lessw-blog
In a detailed proposal published on LessWrong, the author introduces a methodology and accompanying software tool designed to upgrade the fidelity of funding applications by replacing static monetary requests with utility functions.
In a recent post, LessWrong explores a structural inefficiency in how grants and funding are typically requested. The standard model involves an applicant requesting a specific sum of money to achieve a specific goal. However, this format compresses a complex reality into a single data point, resulting in significant information loss. The author proposes a shift toward sharing a "utility function over money"—a graphical representation of how much value a project generates at various funding levels—and provides a web-based tool to facilitate this.
The core context here is the "information asymmetry" inherent in philanthropy, research grants, and venture capital. An applicant understands the marginal utility of their project better than the funder. For example, a project might be viable at $50,000, twice as effective at $80,000, but see diminishing returns beyond $100,000. A single "ask" of $100,000 hides the fact that $80,000 is the optimal efficiency point. By forcing applicants to pick one number, funders lose the ability to optimize their portfolio across multiple projects.
The post argues that the ideal information exchange is a normalized utility function. This allows the funder to see the "shape" of the project's potential. The author introduces a custom tool that allows applicants to draw these functions directly in a browser. The tool supports features like enforcing monotonicity (ensuring the curve doesn't irrationally dip as funding increases) and exporting the data as CSV for analysis. This enables a workflow where funders can aggregate multiple utility curves and mathematically determine the most effective distribution of capital.
The tool itself is designed to be intuitive for non-economists. It allows users to manipulate points on a graph to represent their project's specific needs—such as a "step function" for minimum viable funding or a "logarithmic curve" for projects with early diminishing returns. It also addresses the technical challenge of normalizing these curves so that different projects can be compared side-by-side, specifically by framing the utility in relation to the project's maximum expected value (EV).
This methodology addresses a critical friction point in enterprise workflows and ROI analysis. While currently a theoretical proposal with a prototype tool, it highlights a path toward more data-driven, algorithmic decision-making in sectors that have traditionally relied on intuition and static spreadsheets. By moving from discrete asks to continuous functions, decision-makers can identify "low-hanging fruit" (high utility for low cost) and avoid over-funding projects that have hit a saturation point.
For grant-makers, research directors, and applicants interested in increasing the efficiency of capital allocation, this proposal offers a compelling alternative to the status quo.
Key Takeaways
- Traditional funding applications suffer from information loss by requesting single numerical amounts rather than conveying marginal utility.
- Applicants possess unique context regarding how specific funding increments impact project outcomes, which is lost in standard applications.
- The author provides a web-based tool allowing applicants to draw, normalize, and export utility functions over money.
- Utility functions allow funders to optimize capital allocation based on the marginal value across a portfolio, rather than binary funding decisions.
- The proposed workflow includes technical considerations such as enforcing monotonicity and normalizing to maximum expected value (EV).