P-Necromancer
No bio...
User ID: 3278
P-Necromancer
6d ago
Oh, apologies, I think I misunderstood your point. This method establishes a time lapse is real in that it demonstrates the various intermediate versions existed at the stated times. (They could all be made ahead of time, but each piece must have existed prior to its timestamp.) It doesn't prove those intermediate versions weren't made with AI. (It would have if the author in question could prove they had a substantially identical draft in 2019, but that's not terribly useful going forward.) In terms of 'proving' the work isn't AI, it really only makes it less convenient, not impossible. The same is true in principle of art time lapses (which are generally accepted as evidence), but it's probably easier to fake for written work, given your git commits are going to be a lot less granular. I suppose you could include your vim undofile (or equivalent) in the repo, which would be harder to fake convincingly.
P-Necromancer
6d ago
Sure, I guess? It just seems inferior to the blockchain method, which requires zero trust and has no risk of exposing your work. I'm also not sure how easy it would be to prove ownership over an archived website at the time of archival -- well, you could just put a digital signature on the page, I guess. I suppose the edge is that it's easier to explain how this proves the timeline to nontechnical people? And maybe cost, though I'm not sure about that given opentimestamp's batching strategy; both are free to the user, anyway.
P-Necromancer
6d ago
This is a solved problem. Just put the commit hashes on a blockchain as you make them.
Of course, there's (by design) no way to do it after the fact; you need to have anticipated you might be challenged on the timeline and set this up in advance. Doubt many people are actually doing this in tech, let alone writing. Certainly no one was worried about proving their novel wasn't written by AI in 2019, though this process does have other uses, like establishing authorship (or at least that you had access to the intermediate versions at the time).
P-Necromancer
21d ago
Likewise, there's not even an attempt to hide the corruption anymore. I can not even imagine how much more blatant he could get that this is attempted bribery. People can play the whataboutism game of former admins if they want, but at least those had the shame to hide it behind closed doors a bit.
This is a neat bit of rhetorical framing, but is that actually better? I think it's much worse to disguise your political bribes as prosocial programs. Medicaid is about a trillion, and not just once; every single year! Medicare is 1.2, social security 1.5; this bribe doesn't seem obviously worse to me. Likely better, as the lack of means-testing should mean a greater portion of it would end up getting saved instead of immediately consumed. That doesn't make it good, but I don't think it's whataboutism to discuss this political bribe in the context of other (ongoing! larger!) political bribes.
(The most direct comparison is of course Biden's student loan forgiveness program, which was also a ridiculous pre-midterm lie. But that was small potatoes compared to the bribes that actually did and continue to happen.)
Now, you might object, the old-age programs were founded by democrats but mostly benefit republican voters! (I'm not sure whether this actually true; Republicans are more common in this age range, but it's possible the democrat subset benefits more for one reason or another. But it might well be, and let's say it is for the sake of argument.) So how can I call that a political bribe? Well, the republicans have had numerous trifectas since, and not once have they repealed the programs. The bribe doesn't benefit democrats because republicans elected to match it. This is the worst possible case for everyone (who isn't getting bribed): enormous expense and no political edge (not that a political edge would be worth that kind of money).
This program, though, I believe is so blatant (and so very Trump) that I don't think the democrats will get suckered into matching it. Oh, maybe if it worked they would, but it won't; Trump has no credibility on this point, as you say. So on the whole, I think this is better for the nation than if Trump promised double school budgets (also around a trillion) or whatever.
Claude would get you there; that's how I found the one I linked, opentimestamps. They have a client that wraps git('s gpg function), generates the timestamps automatically, and submits them (for free) to a service that defrays blockchain overhead by bundling together many user's timestamps into a single transaction (as the hash of the list of hashes is sufficient to demonstrate they all existed at the time they were pushed to the blockchain).
I haven't personally verified each piece works the way they say, but it's opensource and this isn't really a complicated or expensive operation; I'd be surprised if it didn't work. (And you could have AI check it over if you didn't trust the community to do so.)
Still, yeah, there's an accessibility barrier. AI could probably walk a non-technical person through the setup, but 99.9% of them wouldn't even know to ask. An archive site is more discoverable.
More options
Context Copy link