Nick Szabo  🔑

As a programmer, it intuitively seems to me that blockchains primary function is to be a source of trustless truth. Turing-complete systems can then anchor on data in blockchains for this, not having to be become blockchains in themselves.

Sometimes -- probably many many times, no reason to think this will be less common than when Turing completeness is needed in normal programming -- you need a Turing-complete program to be that trust-minimized truth, anything less can corrupt the computation in undesirable ways.

There is data and then there are programs or algorithms, that work on the data for new output. The turing-complete algorithms themselves can be verified ad-hoc, there is no need to put them in a blockchain.

The algorithms are typically business logic and iterations over dynamic sets in polynomial time, and their verification code is typically at least as complicated as the original algorithm. So no there is no general win from verifying instead of just running them.

But the data in blockchains are anything but dynamic, they are trustless truth, so no, it is not complicated at all. As to polynomial time, blockchains are already fixed by a new block time, so that is not a factor either.

They are all crucial factors, especially the dynamic nature of most useful, and likely also most useful trust-minimized. computations. Go out and try learning how Ethereum works and what you can do with it before mouthing off.

None of those things need turing completeness. Discrete Log Contracts, scriptless scripts, Simplicity

How do you propose to implement pools of money or cash flows with dynamic membership without Turing completeness?

The blockchain does not need to execute a turing complete smart contract. Post's Theorum for more info. Simplicity will provide witness validation necessary for arbitrary complexity without any turing complete scripting on the chain itself.

But "cash flows" and "dynamic membership" are opaque enough to not be something I can peg down with any certainty.

What's your favorite current (or upcoming) Turing-complete blockchain?

Still looking for one. It's very hard to do well, and technological scalability while maintaining trust minimization is a big problem, but I suspect their is big room for improvement.

Could it be possible to make one a side chain to Bitcoin?

I hope so, but so far success has seemed to elude sidechains.

Maybe a second-best solution could be a completely separate blockchain that's capable of swapping with BTC over the LN.

Can't those be done with non-turing complete tokenization of assets, in particular with upcoming upgrades to Bitcoin scripts like MAST?

No these have very different trust assumptions than replicated running of code.