what do they do? Can you point to some successful use-cases today (other than some kittens) where this might apply?
> These "smart contracts" might still be executable centuries later.
chances are that there is no such thing as the Internet a few centuries later in the same way the Telegraph is pretty much dead today!
Nitpicking but ain't collectible card games with proof-of-ownership a use case though? In the Magic The Gathering world (for real physical cards) ensuring that thought-after cards are the real thing is an issue. There are brick and mortars grading companies grading the cards. Then you have to trust the seller you're not getting a swapped-in chinese fake.
But another use case I've seen used: fully decentralized messaging, either "pseudonymous" (say tied to a BTC address: there's been a working demo showing on to do this on BTC's lightning network) or fully anonymous messages, using ZK cryptography.
In this case you don't know who send a message to whom. You don't even know if it's a message or an amount transferred.
And if the cryptographic primitives behind that are broken, we would pretty soon know (price would crash).
To me decentralized messaging, fully anonymous using zero-knowledge cryptography primitives, is quite a use case compared to the "we-need-your-phone-number-but-trust-us-our-protocol-and-app-ain't-backdoored-and-we're-not-collecting-any-data" messaging solutions that are out at the moment.
But I agree it's a world obviously full of scam. Still: a blockchain is basically a computer: a very slow and bloated one. But it's also an incredibly correct and very hard to hack computer.
I think there are valid use cases for a fully decentralized computer running advanced cryptographic primitives like zero-knowledge cryptography. To me fully anonymous decentralized messaging is one (and a badly needed one at that). There are probably others.
They are creating a p2p network for live video broadcasting infrastructure, essentially a competitor to "AWS Elastic Transcoder" or the GCP equivalent. The thought is that through an automated decentralized auction for transcoding capacity, customers will always get a better price than centralized cloud computing providers. It would be interesting to see this concept applied to general compute capacity as well. I think I read about something like that elsewhere.
But I find the concept of dapps fascinating. I could even make a prediction that early dapps might eventually become the longest continuously "running" programs in the world. The VM is simple enough to maintain and port, and there is an incentive to do so. These "smart contracts" might still be executable centuries later.