In February 2015, a YouTube channel called Unfavorable Semicircle began uploading videos.
The videos were short. Three to five seconds, most of them. Each showed a few seconds of color static, distorted noise, and fragmented shapes that might have been images and might have been compression artifacts. Each was titled with a string of symbols — a mix of letters, numbers, and characters that formed no recognizable pattern.
By mid-2016, the channel had uploaded more than 50,000 videos.
Fifty thousand. In eighteen months. That is roughly 100 videos per day, every day, without interruption.
No human being could produce that at this scale. The uploads were automated. Whatever was generating the content was not sleeping.
The internet noticed, as the internet does. Forums dedicated themselves to the channel. Investigators began downloading and cataloguing the videos, running them through audio software, image enhancement, spectrographic analysis. The work was distributed and collaborative and went on for months.
They found things.
The audio channels, when processed, contained fragments. Voices. Partial phonemes, some interpretable. The images, when isolated from the noise and reconstructed, were worse. Not always recognizable. Not always not recognizable.
Someone, working in the spring of 2016, stitched together 73 specific videos in a specific order — an order derived from a sequence identified in the title strings. When played consecutively, the fragments aligned. The audio formed words. The images formed a single image.
I am not going to describe the image. The people who saw it mostly don't discuss it. The few who have written about it describe a feeling of having been seen by it, rather than seeing it.
The words, as best as they could be transcribed, were repeated three times. They were the same each time.
We found you. We have been here. The distance is not the problem.
Then the channel was terminated. YouTube has not commented on why. The account was simply gone.
All 50,000 videos were gone with it.
Some investigators had downloaded portions of the archive. Not enough to reconstruct the full sequence. Not enough to know if there were other sequences, other images, other messages hidden in the remaining 49,927 videos that hadn't been decoded.
There is a question I have never seen satisfactorily answered: who was the intended audience?
Not researchers. Not internet investigators.
The channel had thirty subscribers before it was noticed. It uploaded for months before anyone looked. Whoever — whatever — was producing it was not making it for an audience. It was made regardless of whether anyone watched.
It was made because it needed to be made. Because there was a message, and the message needed to go somewhere, and wherever it came from, it had the means to send 50,000 fragments and the patience to wait for someone to put them in order.
The distance is not the problem.
I think about what that implies about the distance.
I think about it a lot.