Two new features come to haveibeenflocked.com:
Downloads
You can now download data directly from reports and agency views in CSV or JSON format.
It is important to note that these downloads are NOT (excerpts from) the original source files!
The files are generated from the haveibeenflocked.com database, after they have been processed and imported, a process which could introduce errors.
While we strive to represent the source information accurately, and we hope you will be able to use the downloaded for auditing and investigating use of the Flock system, they are not intended to be a substitute for original, agency-provided sources.
A few notable differences between these downloads and original source files:
- They contain a license plate hash instead of a license plate (see this post for details). This allows for detailed analysis while safeguarding privacy.
- They contain a “best name” and “best name confidence” column. If we found a name that matches a UUIDs or a short names like “J. Smi,” these columns will have those resuls. The provided confidence score is a 0-1 score reflecting the algorithmic certainty of the match.[1]
- PII in the
reasonfields is redacted (see the page on privacy and data redaction). - Redaction results are included (for both haveibeenflocked and agency redactions).
- Information about the source(s) for each record is included.
Sources
Which brings us to the next new feature:
This new table is fairly straightforward — it shows information on the files imported into the haveibeenflocked.com database.
Name Resolution
Not a new feature, but there have been a few, largely under-the-hood, changes to the name resolution process. Results should now be more accurate.
This is still a very fuzzy process, so future changes to the algorithm are likely.
These should always be considered a “best guess” and should be manually verified. ↩︎