Musings on Crypto Infra

Recently, Delphi Digital posted a mega post about how they selected a new ecosystem where they’ll be hanging out. The post was so well written that I thought I’d summarise the key parts and offer my perspective on it as an outside founder to the Cosmos ecosystem.

fwiw, I’m not associated with Delphi/Cosmos in any way apart from being friends. I hold no bags in Cosmos or Delphi. They just happened to publish the best research on this topic and forward it to me.

Delphi Labs spent the last few months/years in the Terra ecosystem building. However, without carefully understanding the risks of the ecosystem they were building on, they got rugged. Given that they needed a new ecosystem to build in, they’ve spent a considerable amount of time doing research on all the various chains with their various drawbacks and advantages. However, what I really liked about the post was how they zoomed out to form a critical thesis of how blockchain applications as a whole will work. It’s something that I spend time thinking about in the background but given the number of moving systems there aren’t enough hours to put all the pieces together (especially when you’re focused on Identity vs L1s/scaling).

The emerging matrix of blockchain applications is going to be on 3 spectrums imo:

• Build on a general purpose chain (Ethereum)

• Build your own specific chain (Cosmos SDK)

• Co-build on a new chain with a limited number of protocols (Polkadot)

The post is mainly geared towards why apps will host their own chains since general purpose chains trend towards the same issues in the long term. While I don’t fully agree on this premise, I do see its merits. Especially with dYdX and Axie ultimately rolling their own chains. The trade-offs are put pretty succinctly in this image:

However, I think there is a world in which this doesn’t hold true. For example, MakerDAO or Uniswap being on their own chain doesn’t make sense. Both rely on deep ecosystem integrations and composability, that if removed, would remove a lot of their moats. However, for something like dYdX, those things don’t matter. The goal is simple: offer the best experience to your users at the lowest price possible.

Delphi acknowledges this point in the post and draws some great parallels to how the future is collection of independent nodes and grouped clusters working together through bridges and cross-chain messaging protocols.

So for example, you might have cosmosdYdX on their own independent chain but then cosmosMakerDAO and cosmosUniswap on a single chain as the “city” while cosmosdYdX lives in the suburbs with a strong enough messaging layer to communicate between the two. This view sounds about right to me although will be critical for bridge and cross-chain infra to be up to scratch. Given the number of hacks and technical trade-offs being discussed at the moment we’re still a while away. That being said I thought the multi-chain future would be years away but turns out it was only just a…