Has anyone built a product around preserving content that platforms keep taking down or burying?
by•
Been thinking a lot about how fast original content disappears. It seems news footage, eyewitness video, documentary clips and other original video gets buried under noise within a week. Something goes viral on Monday and by Friday the original is gone, replaced by reposts, edited versions, and debates on what actually happened until it's just... not findable anymore.
Curious if anyone has tackled this problem, what approach you took, and what you think about the tension between platform distribution (reach) and platform dependency (your content disappearing under noise). Specifically interested in eyewitness footage, journalism, and documentary work — but any category where permanence matters.
What approach works? What doesn't?
15 views
Replies
this is really a provenance problem wearing a permanence costume. the reason the original gets lost isn't only that platforms bury it, it's that once footage is reposted and re-encoded a hundred times, there's no way to tell the original from an edit. nothing in the file proves what it is.
c2pa content credentials are the current attempt at this: tamper-evident origin data embedded into the file itself, so a clip can carry "shot on this device, this date, unedited" independent of whatever platform is hosting it. it's early and adoption is partial, but for eyewitness and documentary work it's the most promising direction, because it attaches the proof to the content instead of to the platform.
the honest gap: a marker can still be stripped, and verifying the signer against a trusted source is its own hard problem. so it helps prove an original is genuine, it doesn't stop the reposts from spreading. permanence and provenance are the same fight though: proving what a piece of content actually is.
@matthias_stralman This is exactly the framing I've been working from! Provenance and permanence are the same fight. You're right that C2PA is the most promising direction for at-capture signing, and the gap you mentioned is right: a marker can be stripped, and the signer verification problem is hard.
The approach we're doing at SeenPool is slightly different; Rather than embedding credentials in the file itself, we create a server-side chain of custody at upload: SHA-256 hash of the original file, HMAC-signed NTP timestamp, device signature, and a provenance log of every verification event. The file stays untouched; the record lives separately and permanently in the archive. The maker can then refer back to it to prove legal ownership when other copies come up. It's kind of like a notary for video files. You can prove you were the originator and what the original showed.
The limitation is the same as C2PA's from the other direction: we can prove what arrived at our server and when, and we can prove authenticity. but we can't prove what happened between capture and upload. The ideal is both C2PA at capture, and SeenPool at archive. They're complementary, not competing.
Would love to stay in touch, this is exactly the conversation we need to be having.
@elbunts seenpool is a genuinely clever inversion, and you've put your finger on why they're complementary: c2pa signs at the moment of capture but is fragile downstream, you sign at the moment of arrival but can't see upstream. together they close each other's blind spot. capture-time provenance plus an immutable arrival record is honestly the strongest chain i've heard described for this.
the "we can't prove what happened between capture and upload" gap is the real one, and i don't think it's fully solvable without hardware attestation at the sensor, which is years out for consumer devices. so the pragmatic answer probably is exactly what you said: layer both, and be honest about what each layer can and can't prove. the failure mode to avoid is either side overclaiming.
yes to staying in touch. i work the reading/detection side of this (checking what markers a file already carries, locally), so we're looking at the same problem from opposite ends. happy to compare notes properly.