Child chains are a concept that was invented by an altcoin developer community (Ardor/Nxt). It could best be described as a method to integrate sidechains and the main chain in a very tight way.
The inventors claim that IBD (initial block download) and block storage requirements for regular full nodes are only about 3% from a traditional monolithic blockchain, potentially even less. So I wanted to explore it as a possibility for Bitcoin.
The method works as follows:
- All nodes - mining and not-mining full nodes - validate all transactions on the main chain.
- Regular payments are meant to ocurr on "child chains". These are not separate block chains like the sidechains we know, but each one has a separate payment token. Every couple of minutes a node called the "bundler" bundles the transactions of a certain "child chain" and writes a special transaction (the "child chain block transaction") into the main chain that contains, as an attachment, a list of these transactions.
- From these lists of transactions, only those ocurring in the last blocks (in the original concept, in the last 1440 blocks) are validated by all full nodes. Older "child chain block" transactions are pruned in a way that only their headers are validated and stored.
- Nodes can choose to be archival nodes of the child chains they choose and validate and store all child chain block transactions fully.
(I am not a developer so my description may contain errors. The original concept is described
in this overview.)
Now my question is: Would this method be possible to implement (in the long term, obviously) in Bitcoin?
It seems to be a decent compromise between on-chain and off-chain scaling, maybe comparable to extension blocks, and would give drastic advantages regarding to a one-chain solution, but a tighter integration with respect to sidechains. A 2-way-peg would be much easier (and more trustless) to realize than with the Drivechain/Elements concept.
