There’s “it’s easy for someone to publish a thing and prove it was published before $time”. Honestly that’s pretty easy already … marginally more trustless (blockchain) would be marginally good, and I think cheap and easy.
More trustless is the main point. The marginal value could grow over time, or depending on your situation/jurisdiction/trust, be larger already for some people than others. Perhaps there are already certain countries where journalists aren’t supposed to trust institutions in other certain countries?
But I think what you’re talking about here is doing this for all public content, whether the author knows or not. … the big problem I see here is that a lot of things get edited after publishing … everything on imgur can no longer be verified, at least not to before the recompression … There’s a related question of how you distinguish content from metadata
If you change it, then the new thing is different from the old thing, and the new thing did not exist before it was created. If you change compression method, blog theme, spelling or some other irrelevant aspect, you can start from the archived old version, prove that it was created before a certain point in time, then try to convince me that the difference between old and new is irrelevant. If I agree, you will have proven to me that this new thing was effectively created before a certain time. If not, at least you have proven that the old version was.
I do not propose trying to centrally solve the question of what changes are relevant, or distinguishing content from metadata, because that is up to interpretation.
This problem goes away if you’re also hosting copies of everything, but that’s no longer cheap. At that point I think you’re back to “addition to the internet archive” discussed in other comments
Yes, I do admit that an archive is necessary for this to be valuable. I would prefer to execute this by cooperating very, very closely with the internet archive or another big archive for a first project.
you’re only really defending against the internet archive going rogue (though this still seems valuale), and there’s a lot that they don’t capture.
Yeah, more or less. I think more. We are also defending against other kinds of current and future compromise including hacking of a well-meaning organization, government overreach, and people’s/organizations’/journalists’ unfounded distrust in the internet archive. Organizations change, and the majority of the value proposal is a long-term one.
I think having to rely on an archive makes this a lot less valuable. If I find something in 2033 and want to prove it existed in 2023, I think that’s going to be much harder if I have to rely on the thing itself being archived in 2023, in an archive that still exists in 2033; compared to just relying on the thing being timestamped in 2023.
I also think if you’re relying on the Internet Archive, the argument that this is urgent becomes weaker. (And honestly I didn’t find it compelling to begin with, though not for legible reasons I could point at.) Consider three possibilities, for something that will be created in May 2023:
We can prove it was created no later than May 2023.
We can prove it was created no later than June 2023.
We can prove it was created no later than June 2023; and that in June 2023, the Internet Archive claimed it was created no later than May 2023.
A one-month delay brings you from (1) to (2) if the IA isn’t involved. But if they are, it brings you from (1) to (3). As long as you set it up before IA goes rogue, the cost of delay is lower.
If I find something in 2033 and want to prove it existed in 2023, I think that’s going to be much harder if I have to rely on the thing itself being archived in 2023, in an archive that still exists in 2033; compared to just relying on the thing being timestamped in 2023.
Yeah, I think this is an unfortunate technical necessity, but only in the case where the thing you find in 2033 has been changed (in an irrelevant way). If you find something in 2033 that was actually timestamped in 2023, you don’t need access to an archived version, since it’s identical to what you already have.
I also think if you’re relying on the Internet Archive, the argument that this is urgent becomes weaker … As long as you set it up before IA goes rogue, the cost of delay is lower.
This is fair criticism. IA does in fact timestamp content and has for a good while, just not trustlessly (at least not intentionally AFAIK). So, to the extent (in jurisdiction and time) that people in general can trust IA, including their intentions, competence, technology and government, perhaps the value really is marginal at the present time.
Perhaps I will slightly decrease my belief about the urgency, although I remain convinced this is worth doing. I see the value as mostly long-term, and IA’s claims for what content was created when, is itself a treasure of arguably reliable recordkeeping worth protecting by timestamping.
More trustless is the main point. The marginal value could grow over time, or depending on your situation/jurisdiction/trust, be larger already for some people than others. Perhaps there are already certain countries where journalists aren’t supposed to trust institutions in other certain countries?
If you change it, then the new thing is different from the old thing, and the new thing did not exist before it was created. If you change compression method, blog theme, spelling or some other irrelevant aspect, you can start from the archived old version, prove that it was created before a certain point in time, then try to convince me that the difference between old and new is irrelevant. If I agree, you will have proven to me that this new thing was effectively created before a certain time. If not, at least you have proven that the old version was.
I do not propose trying to centrally solve the question of what changes are relevant, or distinguishing content from metadata, because that is up to interpretation.
Yes, I do admit that an archive is necessary for this to be valuable. I would prefer to execute this by cooperating very, very closely with the internet archive or another big archive for a first project.
Yeah, more or less. I think more. We are also defending against other kinds of current and future compromise including hacking of a well-meaning organization, government overreach, and people’s/organizations’/journalists’ unfounded distrust in the internet archive. Organizations change, and the majority of the value proposal is a long-term one.
I think having to rely on an archive makes this a lot less valuable. If I find something in 2033 and want to prove it existed in 2023, I think that’s going to be much harder if I have to rely on the thing itself being archived in 2023, in an archive that still exists in 2033; compared to just relying on the thing being timestamped in 2023.
I also think if you’re relying on the Internet Archive, the argument that this is urgent becomes weaker. (And honestly I didn’t find it compelling to begin with, though not for legible reasons I could point at.) Consider three possibilities, for something that will be created in May 2023:
We can prove it was created no later than May 2023.
We can prove it was created no later than June 2023.
We can prove it was created no later than June 2023; and that in June 2023, the Internet Archive claimed it was created no later than May 2023.
A one-month delay brings you from (1) to (2) if the IA isn’t involved. But if they are, it brings you from (1) to (3). As long as you set it up before IA goes rogue, the cost of delay is lower.
Yeah, I think this is an unfortunate technical necessity, but only in the case where the thing you find in 2033 has been changed (in an irrelevant way). If you find something in 2033 that was actually timestamped in 2023, you don’t need access to an archived version, since it’s identical to what you already have.
This is fair criticism. IA does in fact timestamp content and has for a good while, just not trustlessly (at least not intentionally AFAIK). So, to the extent (in jurisdiction and time) that people in general can trust IA, including their intentions, competence, technology and government, perhaps the value really is marginal at the present time.
Perhaps I will slightly decrease my belief about the urgency, although I remain convinced this is worth doing. I see the value as mostly long-term, and IA’s claims for what content was created when, is itself a treasure of arguably reliable recordkeeping worth protecting by timestamping.