How this works

Who is behind this

I'm Derek Miller. I've spent my career making complicated systems easier to understand and easier to use. I started out teaching English abroad for two years, then spent a few years in digital marketing, before moving into technology and modernization. Most of that work has been federal: years leading IT programs, including large efforts to move infrastructure to the cloud, refresh tens of thousands of workstations, and stand up remote work for thousands of people in a matter of weeks. Today I work in sales operations, focused on how organizations adopt new technology and automation without losing the thread of why they wanted it.

ATTRACTOR grew out of the same instinct. Policy is not as impenetrable as it is made to sound. It just gets explained badly, usually by people who want you to pick a side before you understand the question. I wanted a place that did the opposite: lay out how things work, show the sources, and let you decide.

ATTRACTOR is an independent project. It is not funded by, affiliated with, or speaking for any party, campaign, candidate, or organization.

How proposals are researched

Every claim on this site is meant to trace back to a source you can check. We hold to a tier of sources, and we prefer the strongest one available for any given number.

1. Tier 1. Always preferred.

   Government and academic primary sources. The Congressional Budget Office, the Bureau of Labor Statistics, the Bureau of Economic Analysis, the Federal Reserve, the Census Bureau, federal agency data, and peer-reviewed academic studies. When a number can come from here, it does.

2. Tier 2. Acceptable, with the lean disclosed.

   Established research institutions. The Tax Foundation, the Peterson Institute, the Urban Institute, Brookings, and the National Bureau of Economic Research. We use these when no primary source answers the question. Where an institution has a known lean, we say so.

3. Tier 3. News, not analysis.

   Major newsrooms. Reuters, the Associated Press, and the news desks of major outlets. We cite these for what happened, not for what it means.

4. Never.

   Partisan advocacy sites. Anonymous claims. Single-outlet reports we cannot verify anywhere else. These do not appear in our evidence base.

Estimates and projections are always labeled as such. A projection is not a measurement, and we do not present it as one. Where a cost has been formally scored, we say who scored it. Where it has not, we say that too.

Working estimates and unresolved costs appear on The math. It keeps historical portfolio figures separate from current healthcare comparisons. The complete cost and funding gap are still unknown. A related official score is not a score of our exact design.

Revision policy

These proposals are living documents. Evidence improves. Programs get scored. Circumstances change. When that happens, we revise.

Each domain shows a revision date; source years can be older. Our correction note explains changes to the funding claim. The open-question register distinguishes evidence gaps from policy choices. If we get something wrong, we change it. We would rather be corrected than repeat a mistake.