An Honest Talk about Redaction

Having worked in surveillance / counter-surveillance for the bulk of my law enforcement career, I know the importance of redacting personal information from videos that will be released to the public or the judicial system. Blurring faces and other private information is absolutely essential to the safety of surveillance operators. But, in the new world of body worn cameras (BWC), redacting private information went from a once-in-a-while exercise, to a daily activity around Freedom of Information Act requests. Local / State / National laws often govern what type of information can be recorded / released and what needs to be redacted. An example of this includes the legal requirement to redact children's faces, license plates, and so forth.

Whilst we work on new technology to speed up the process, Amped FIVE is still the industry leader in terms of speed and flexibility when performing manual redactions. But even with that, redaction can be a time consuming process. Here's a look at the numbers, taking my time in LE as a baseline:

The average BWC can accurately "see" about 12'-18' within it's field of view, which tends to cover between 100-150 degrees (depending on manufacturer and model. Within this area, the viewer may be able to identify / recognize personally identifiable information.

Although the BWC's microphone tends to be omni-directional, the average BWC can "hear" about 20' in front of the wearer in a roughly spherical shape that's blocked a bit by the wearer.

Thus. the "bubble" of potentially redactionable material is roughly 20' around the wearer. Add more BWCs to the scene and that "bubble" increases in size and complexity.

With all this in mind, let's examine a Cost Per Redaction formula to determine the cost per request to an LE agency.

Based on my years of experience with Amped FIVE and other tools, it takes about 3-5 minutes of actual time to redact a one minute recording of a single object / subject - per camera view. Thus, a 5 minute interaction between a single LEO and a single citizen can take between 15 – 25 minutes to manually redact. A 5 minute interaction between a single LEO and 5 citizens can take between 75 – 125 minutes to manually redact. Add a second LEO with a BWC and the time to redact more than doubles. If the interaction takes place around reflective materials, the situation becomes entirely more complex.

Using my previous employer’s costs, here’s what the cost would look like in the previous scenario.

• Employee cost: $74/hour (salary + health care + retirement + etc)
• Simple redaction – one subject / 1 LEO for 5 minutes = $37
• Medium complexity – one LEO + 5 subjects @ 5 minutes = $150
• Maximum complexity – 6 LEOs + 12 subjects @ 30 minutes = . (30 x 5) x 12 = 1800 minutes x 6 LEOs – 10,800 minutes = 180 hours 180 hours x $74/hour = $13,320

Why so long? We cover this question in depth in our Redaction Class. In essence, there's time needed for triage, there's the time to actually redact the footage, then there's the time to do the quality assurance work necessary when dealing with CJIS / HIPAA data before packaging the request for release to the public.

Some agencies are actually billing for redactions using a similar formula, with employee cost figures based on their local salary tables.

Remember, redaction tasks generally come at the end of the workflow. Restoration and Clarification precede the redaction work as they help make the redaction easier to perform. In a single LEO / single subject scenario, some local stabilization can help take the redaction from a manual tracking exercise to a static / fixed location redaction. But, with multiple subjects in motion, local stabilization isn't going to help.

Whilst we work on updates to our dynamic / automatic redaction processes, please understand that manual redaction is a time-consuming process. It's not the type of process that lends itself to multi-tasking. When the employee is engaged in the process, that's all that they can do. Their concentration should be fixed on the task. Thus, an agency that institutes a redaction policy that specifies the blurring/pixelating of specific information (blur faces not the whole image), can expect to spend a lot of staff-hours on redaction. There's no two ways around it.