RFD: the alien abduction prophecy protocol

"It's tough to make predictions, especially about the future."
- variously attributed to Yogi Berra and Niels Bohr

Right. So let's say you are visited by transdimensional space aliens from outer space. There's some old-fashioned probing, but eventually, they get to the point. They outline a series of apocalyptic prophecies, beginning with the surprise 2032 election of Dwayne Elizondo Mountain Dew Herbert Camacho as the President of the United States, followed by a limited-scale nuclear exchange with the Grand Duchy of Ruritania in 2036, and culminating with the extinction of all life due to a series of cascading Y2K38 failures ^[1] that start at an Ohio pretzel reprocessing plan. Long story short, if you want to save mankind, you have to warn others of what's to come.

But there's a snag: when you wake up in a roadside ditch in Alabama, you realize that nobody is going to believe your story! If you come forward, your professional and social reputation will be instantly destroyed. If you're lucky, the vindication of your claims will come fifteen years later; if not, it might turn out that you were pranked by some space alien frat boys who just wanted to have some cheap space laughs. The bottom line is, you need to be certain before you make your move. You figure this means staying mum until the Election Day of 2032.

But wait, this plan is also not very good! After all, how could your future self convince others that you knew about President Camacho all along? Well... if you work in information security, you are probably familiar with a neat solution: write down your account of events in a text file, calculate a cryptographic hash ^[2] of this file, and publish the resulting value somewhere permanent. Fifteen years later, reveal the contents of your file and point people to your old announcement. Explain that you must have been in the possession of this very file back in 2017; otherwise, you would not have known its hash. Voila - a commitment scheme ^[3]!

Although elegant, this approach can be risky: historically, the usable life of cryptographic hash functions seemed to hover at somewhere around 15 years - so even if you pick a very modern algorithm, there is a real risk that future advances in cryptanalysis could severely undermine the strength of your proof. No biggie, though! For extra safety, you could combine several independent hashing functions, or increase the computational complexity of the hash by running it in a loop. There are also some less-known hash functions, such as SPHINCS ^[4], that are designed with different trade-offs in mind and may offer longer-term security guarantees.

Of course, the computation of the hash is not enough; it needs to become an immutable part of the public record and remain easy to look up for years to come. There is no guarantee that any particular online publishing outlet is going to stay afloat that long and continue to operate in its current form. The survivability of more specialized and experimental platforms, such as blockchain-based notaries, seems even less clear. Thankfully, you can resort to another kludge: if you publish the hash through a large number of independent online venues, there is a good chance that at least one of them will be around in 2032.

(Offline notarization - whether of the pen-and-paper or the PKI-based variety - offers an interesting alternative. That said, in the absence of an immutable, public ledger, accusations of forgery or collusion would be very easy to make - especially if the fate of the entire planet is at stake.)

Even with this out of the way, there is yet another profound problem with the plan: a current-day scam artist could conceivably generate hundreds or thousands of political predictions, publish the hashes, and then simply discard or delete the ones that do not come true by 2032 - thus creating an illusion of prescience. To convince skeptics that you are not doing just that, you could incorporate a cryptographic proof of work ^[5] into your approach, attaching a particular CPU time "price tag" to every hash. The future you could then claim that it would have been prohibitively expensive for the former you to attempt the "prediction spam" attack. But this argument seems iffy: a $1,000 proof may already be too costly for a lower middle class abductee, while a determined tech billionaire could easily spend $100,000 to pull off an elaborate prank on the entire world. Not to mention, massive CPU resources can be commandeered with little or no effort by the operators of large botnets and many other actors of this sort.

In the end, my best idea is to rely on an inherently low-bandwidth publication medium, rather than a high-cost one. For example, although a determined hoaxer could place thousands of hash-bearing classifieds in some of the largest-circulation newspapers, such sleigh-of-hand would be trivial for future sleuths to spot (at least compared to combing through the entire Internet for an abandoned hash). Or, as per an anonymous suggestion relayed by Thomas Ptacek: just tattoo the signature on your body, then post some post some pics; there are only so many places for a tattoo to go.

Still, what was supposed to be a nice, scientific proof devolved into a bunch of hand-wavy arguments and poorly-quantified probabilities. For the sake of future abductees: is there a better way?