Qualities of a successful open crime data set

Hundreds of agencies are making their crime data openly available. We wanted to make sure we highlighted the great qualities of an open crime data feed and continue to push agencies to make their crime data feeds as good and complete as possible. 

We’ve found that sharing data openly is a good indicator that a police agency has control over their data, they are not afraid of accountability, they can manage confidential and non-confidential elements, and they are a good player with state and national law enforcement. Remember, ‘open’ means the data is available in machine readable format for anyone to collect, use, and share without restrictions.

Good Qualities

RMS and CAD data - both tell different stories and are both important.

Lat/long coordinates - helps with geocoding accuracy.

Machine readable - PDF is NOT machine readable.

Follows a standard - ex: SOCS 

Up-to-date frequency - hourly or daily 

Accountable - Contact information for questions. This opens up a feedback loop with consumers.

Continuity and Consistency - keep the data the same unless you are adding something important (like lat/long or descriptions). It’s hard to process data that keeps changing location or structure. Make sure to let consumers know when changes are made.

Uniformity - same feed that’s being shared with one is shared with all. This helps avoid confusion. For example, if a crime mapping vendor is given CAD data but the media data file only holds RMS data it can cause confusion and may even lead to people drawing the wrong conclusions.

Accessibility - data is easy to find by anyone.

Bad Qualities

Varied - Changes format and location frequently.
Incomplete - Missing data points like a location, date, or time.
No contact information listed.
Low accessibility - Emailing out data is good, but it’s more accessible to anyone when it’s published directly on the website.
Preferential access - some agencies give their private vendors early access to public information over the public. Don’t do this. Unfairly distributing public data by giving preference to one company generally seems like a bad idea for transparency and democracy.
Low frequency - Monthly or quarterly updates. Yearly data is cringeworthy.

Are we missing something from our list? Let us know!!