In Malibu Media v. Doe, EDNY 15-3504, Judge Locke has denied the defendant's motion to quash, and lifted the stay on all of the EDNY Malibu Media cases, which had all been consolidated.
The Court accepted the representations of plaintiff's expert, a Michael Patzer of a company called Excipio, that Malibu relies on "direct detection" rather than "indirect detection", and that it is "not possible" for there to be misidentification.
Memorandum Decision of Magistrate Judge Locke denying motion to quash
2 comments:
The assertion that each 16 kB worth of data corresponds to 2 seconds of video is erroneous; 16 kB cannot be even one full frame of video. Modern video compression relies on Temporal Compression, by which data for any given frame only describes the changes from previous frames. It is impossible to determine whether this "differential" data conclusively comes from any given video file because the same "diff" can exist in many videos. Analogously, the "differential" of 5 and 4 is 1. Likewise, the differential of 9 and 8 is 1. Without knowing where you started you cannot apply the differential of 1 and have any meaningful result.
Modern video compression relies on "key frames" every so often in the video stream which do not use Temporal Compression, and it is from these fixed frames which every sequence of differential (i.e. "intermediate") frames is derived. In order to determine whether you have genuinely downloaded the video you believe, you must download at least 1 keyframe.
Think of it like this: I start at my house. I go one block then left, then 2 blocks and right. If you know where I started, then you know where I ended. If you don't have my fixed starting coordinate then I could have begun anywhere, and ended anywhere. That's like video compression -- the intermediate steps (frames) are meaningless unless you have the keyframes.
Of course, because the defense failed to call an expert witness no one called-out Patzer on his fallacious misrepresentations to the court. Fact is, 16 bytes of data is an incredibly small amount; not nearly enough sequential bits to statistically be high-confidence the files are the same. One cannot decompress video to ensure it's your own content without keyframes; any arbitrary video can be compressed to have matching differential bits.
Thank you very much for pointing that out.
I was shocked by Judge Locke's just taking this "expert" at his word.
Post a Comment