(Originally posted here)
As a Cambrian explosion of blockchain solutions floods the press and marketplace, there have been several calls for creating standards and interoperability between blockchains. The common historical episode cited to invoke collaboration is the “videotape format war” between VHS and Betamax in the 1970s and 80s. However, interoperability between blockchain-enabled ledgers is a much more interesting and difficult problem that may not be solved easily.
In order to illustrate my point, it is important to draw a distinction between information networks and consensus networks. Information networks generally act as conduits for getting data from point A to point B. In this context, interoperability can be achieved with the use of wrappers. If an actor formats their data properly at point A they can then reformat it at point B. For an example of this type of interoperability, look no further than email: today, most people access and send email via HTTP through web interfaces though the underlying email protocol remains SMTP. Though the two protocols are largely different, those differences did not hamper the adoption of webmail. A similar example of this in a blockchain network might be converting positions between Bitcoin’s UTXO representation and Ethereum’s more standard notion of account balances. In this case, converting between one representation or another to understand a position is straightforward, requiring few lines of code. We no longer live in the 70s when choosing between Betamax and VHS meant committing to a specific and expensive piece of hardware; today this type of tension ceases to exist between information networks as a result of innovations in software.
Beyond examples of data conversion, the analogy between interoperability among information networks and consensus networks breaks down. In consensus networks, the goal of interoperability is mutual knowledge, which means robust validation of any conjecture from one network to another. In other words, any interoperability between consensus networks requires creating a proof of consensus which can be deemed valid by another network. One way to sidestep this issue is to build various “colored coins” on top of a single consensus layer. However, there are technological limitations that generally come up with such approaches, such as poor transaction semantics.
It is possible to achieve some interoperability between different consensus networks but the requirements can be tricky. In general, the networks will need to explicitly support such operations and proofs of consensus must be ported from one network to the next. This is considerably more technical and involved than what is required for the type of interoperability in information networks. The interoperability that matters will be that of consensus — not representations of positions or information, which is trivial to accomplish.
