About the project

Who runs it

The American Mainspring Project is founded and edited by a single private citizen with a scientific background, working independently. There is no staff, no board, and no organization behind the name; it is one person's project, built openly and with AI tools described below. One disclosure on this page is deliberately incomplete, and we say so rather than hide it: the founder's name is withheld at this stage and will be published when the project moves fully into the open. Everything else here, the funding, the legal status, the conflicts policy, and a working contact address that reaches the founder directly, is disclosed now, and until the name is, the fair response is to weight this project by whether its sources hold when you click them, which is the same test it applies to everyone it covers.

What this is, and why it exists

This is an independent policy and accountability platform, not a party, campaign, PAC, or advocacy organization. It exists to do a few connected things: state what the evidence actually supports, pin every figure to a source anyone can check, hold both elected officials and the agencies that carry out policy to account for what they do, and lay out a concrete roadmap that officials and citizens can read, argue with, and sign onto. It is meant as a platform for a politics not owned by either party as it stands now, and a home for people who feel unrepresented by the current choices. The larger bet behind it is that artificial intelligence, used carefully and under human judgment, can raise the quality of government accountability and help unlock human potential rather than replace it.

Where its conclusions land

Honestly: left of center, more often than not. The founder's view is that on most of the questions here, the weight of the evidence points that way, and the project does not pretend to a false neutrality about it. But left-of-center is where the conclusions land, not where the method begins. The method is evidence first: the platform credits good policy wherever it comes from, including from conservatives and red states, concedes weak cases in print, and changes any position when the evidence changes. If that ever stops being true, the Corrections page is where you will catch it.

How the work is governed

The project's real governance is its published method, which binds the authors the same way it binds everyone it covers: every figure traces to a source on the Evidence page; contested numbers use the official scorekeepers; weak cases are conceded in print; a weekly automated hostile review attacks the project's own pages from the right, the left, and a technical angle; and every substantive fix is logged publicly on the Corrections page. Where a judgment is editorial, the pages say so and publish the rules.

Made with AI, governed by humans

This project uses artificial intelligence, openly. Claude (Anthropic's AI) helps research sources, draft text, run the weekly agents that scan Congress and propose evidence, and even attack our own pages in the weekly hostile review. Nothing AI produces publishes itself: every proposal goes through human review against the published evidence standard, every figure gets pinned to a source a person can check, and every error that slips through gets logged on the Corrections page like any other.

We consider this an ethical use of the technology, and the test we apply is the same one this platform applies to everything: tools exist to serve people. We are not against AI, the way we are not against markets. We are against AI at the expense of humanity, the way we are against an economy that serves its own numbers instead of the people inside it. A tool that helps a small project hold power accountable with evidence is the technology working the way the innovation brief says it should: AI handles the paperwork, people make the decisions.

Who pays for it, and legal status

The project is self-funded by the founder and takes no donations, sells nothing, and has no clients, sponsors, or outside money of any kind. Costs to date are small and personal: registering the domain, hosting on free service tiers, and metered AI usage for the research and review agents. There is no budget, payroll, or fundraising. Legally, it is an unincorporated personal project, not a registered nonprofit, company, or political committee. If that ever changes, for example if it incorporates or begins accepting support, this page will say so before it happens, and the change will be logged on the Corrections page.

Conflicts of interest

The founder holds no financial stake in any policy outcome tracked here, represents no client or employer through this work, and receives no money from any person or organization the project covers, endorses, or opposes. The one material disclosure is the heavy use of AI tools, described above and used under human review. If a genuine conflict ever arises, it will be disclosed on the relevant page rather than managed quietly.

Contact and your data

The data controller for this site is the project's founder-editor, reachable directly at the address below. Form submissions (signup, volunteer, endorsement, corrections) are handled as described on the Privacy page; you can ask what is held about you, or ask for it to be deleted, at any time. Reach the project at contact@american-mainspring-project.org. For corrections and challenges specifically, the Report an Inaccuracy form is fastest; for endorsements and volunteering, see Get Involved.