Responsible use
9 Foundations, Inc.

What Twin is

Twin is a synthetic population of the United States: 330 million modeled residents, each placed in a real census tract, built from federal surveys and the decennial census. It answers questions about where kinds of people are concentrated and when the conditions around them change. Its outputs are estimates about places, not facts about persons.

What Twin is not

Twin is not a list of people, a consumer database, or a record of any real individual. No Twin corresponds to a real person, and Twin outputs cannot be used to identify, locate, or contact one. Twin does not contain personally identifiable information and never has.

How we scope an engagement

Every engagement starts with a stated decision: the question the client is trying to answer and what they will do with the answer. We build the archetype and select the hidden variables against that decision, and we document both. If a request changes purpose mid-engagement, we re-scope it.

What we decline

• Any attempt to identify, re-identify, locate, or profile an individual, including by combining Twin outputs with other data for that purpose.
• Uses whose purpose is to exclude, disadvantage, or discriminate against people on the basis of race, ethnicity, national origin, religion, sex, sexual orientation, gender identity, disability, familial status, or age.
• Individual eligibility decisions in credit, housing, insurance, or employment. Twin describes places; it is not a substitute for the individual assessment those decisions require by law.
• Surveillance, or any use aimed at monitoring specific people or households.

Sensitive characteristics

Some domains in Twin describe health, disability, language, immigration, and safety. They exist because public-health work requires them, and they are subject to the rules above. We support uses that bring a product or service to the people who need it. We decline uses that would deny a product, service, price, or opportunity to people because of a health condition or protected status.

Client data

When a client overlays its own sales, engagement, or location data on Twin, that data is used only for that client's engagement. It is not added to Twin and is not shared with anyone else.

Accuracy and limits

Twin is validated against Census benchmarks, with an average Pearson R² of 0.98 across census tracts. It is still a model. Outputs are probabilistic estimates at the census-tract level and should be used as such: to prioritize, rank, and time decisions, not to assert facts about any household.

Questions

If you are unsure whether a proposed use fits these rules, ask before you start: twin@9Foundations.com.