Every conversation about video in a defender office ends at the same question: how many hours. Not how many gigabytes, and not how many clips. Hours, because hours are what the office does not have.
Here is what the public record says, and what it means for a calendar.
What arrives
By 2016, the last year in the Bureau of Justice Statistics survey, 47 percent of general-purpose law enforcement agencies in the United States had acquired body-worn cameras; among agencies that had them, 60 percent of local police departments and 49 percent of sheriffs’ offices had fully deployed them [1]. That was a decade ago, and it was already the direction of travel.
The RAND national workload study put the consequence in a defender’s words. Its expert panelists reported that in many jurisdictions nearly all police interactions are recorded, that defenders receive body-worn and dashcam video in almost every case, and that police video has essentially replaced written reports as the primary record of searches, witness interviews and scene information [2]. The panel also named the multiplier that makes the arithmetic hurt: if five officers respond to a scene, the defense receives five videos, from five angles, catching five different conversations [2].
What it adds up to
Colorado Public Radio reported figures presented by Jefferson County District Attorney Alexis King to other district attorneys: her felony attorneys carry roughly 100 cases each, and each case has at least four to six hours of body-worn camera, which she put at 400 hours of video per attorney to review within 21 days of a case being filed [3]. Her office handled more than 36,000 videos totaling 24,000 hours in 2022, and more than 67,700 videos totaling 41,000 hours in 2025 [3]. A forensic investigator at the Denver District Attorney’s Office told the same reporter there had been a 600 percent increase in audio and video evidence in five years [3].
Those are prosecution numbers, produced by an office with investigators and staff. The defense receives the same discovery with fewer people. James Karbach of the Colorado State Public Defender’s Office described the shift on the defense side plainly: “They might be able to prepare something in two hours that now takes us 40 hours” [3].
Now set that against the time the work is supposed to take. The RAND panel of 33 defense practitioners estimated 35 hours of attorney time for a low-severity felony and 13.8 hours for a low-severity misdemeanor, for the entire case: client, investigation, motions, negotiation, court [2]. Four hours of footage in a misdemeanor is not a line item in that budget. It is a third of it.
Why keyword search is not review
Transcription and keyword search are useful, and most offices should have them. They are not review, for three reasons.
The first is that a transcript indexes words, and much of what matters on a body camera is not a word: where the officer’s hands go, what happens in the twenty seconds before the question, who else is in frame, how long the client waits. The second is that you cannot search for what you have not thought to ask. A search returns hits for the term you typed; it does not tell you the stop went nine minutes past the point the reason for it ended.
The third is that the obligation does not bend. RAND’s panelists said that to be effective, attorneys must watch and listen to all of the raw video to determine what might be useful or harmful, and that much of that time may seem wasted, because nothing is on the video, but the review is how you learn that [2]. One panelist made the point exactly: only by watching does an attorney catch a Miranda violation or a relevant side comment [2]. The ABA’s defense function standards put the duty behind that instinct: counsel has a duty to investigate in all cases, and the investigation should include evaluation of the prosecution’s evidence and consideration of inconsistencies and avenues of impeachment [5].
Sampling is what an agency does when it cannot watch everything. The New York City Comptroller’s review of the NYPD body-worn camera program describes precinct sergeants selecting five videos a month per assigned officer for compliance review, and reports that of the 7,797 videos on file, cameras were activated late or deactivated early in 1,436 instances, or 18 percent, and that 4,319 of 12,116 dispatch review determinations, or 36 percent, had no corresponding video on file at all [4]. A sample is a sample. Your client’s nine minutes are either in the part someone watched or they are not.
What “per issue” watching means
The practical answer is not to watch more. It is to stop watching for “anything interesting” and start watching for something specific, one pass at a time.
Before the first minute plays, write down the issues: the elements the state has to prove, the suppression theory, the impeachment you expect to need, the provenance questions about how the footage was made and produced. Then watch each pass for one of them, note the timestamp and the exhibit, and move on. Anything that bears on an issue gets an address, the way a document gets a Bates number.
The output of that work is not a summary. It is the raw material of a hearing memo: what the record shows, where it shows it, and what is missing.
That is also the part a machine can do first, as long as what it hands you carries a timestamp for every line and says out loud what it could not establish. It is the reason Apodicta exists in the form it does: the watching happens before the deadline, and the review is a spot check against the record rather than a second viewing of it.