Thread by @AngelSDiaz_

Algorithms can determine how long someone spends in jail, who has access to schools, and where you vote. @BrennanCenter signed onto a letter with recommendations for how NYC can protect against biased decision-making by city agencies. Here are some highlights #NYCAlgorithms

NYC should publicly disclose its use of algorithms, explain how they work, and allow the public to challenge decisions made by them. Explanations should account for New York's linguistic, socioeconomic, and cultural diversity.

Algorithms must be tested for disparate impact based on protected status. If disparate impact is found, the agency should explain why its use is necessary to achieve an important agency interest, and that there is no less-discriminatory alternative.

The City Council should pass a law allowing New Yorkers impacted by discriminatory algorithms to bring a private right of action.

NYC should establish an online resource of all algorithms on an agency-by-agency basis. This resource should include disclosure of source code so algorithms can be studied for bias and disparate impact.

City contracts with vendors must include provisions requiring them to disclose information for all data-sets used to develop and implement the systems, along with any records of bias testing.

Agencies should perform an Algorithmic Impact Assessment, preferably *before* acquiring or building a new algorithm. Each agency should review existing systems for fairness, justice, bias, civil rights, privacy, and related concerns.

NYC should always consult impacted communities regarding the use of algorithms. The letter also provides a number of subject-matter experts for issues ranging from policing to housing rights, and child welfare to public benefits.

Here's the full text of the letter: assets.ctfassets.net/8wprhhvnpfc0/1…