At Blocky, we think about trust as infrastructure you can provision - like an EC2 instance.

There are many ways to provision trust, each with trade-offs. Those trade-offs determine which flavor of trust is best for different use cases.

Care to join me for a quick walk & talk on trust models? 🧵

Building the next layer/Posting hot takes from the grind.  Get in Builder, we're doing #truststuff

A key driver of Web2 was the ability to quickly provision compute resources for applications (storage & messaging too, but let’s simplify).

In Web3, a key driver is the ability to provision trust - just like compute.

There are several methods for provisioning trust, but they all stem from a few core categories:

*Proof of Work (PoW) - consensus
*Proof of Stake (PoS) - consensus
*Zero-Knowledge Proofs (ZKPs) - cryptography
*Restaking - consensus
*Trusted Execution Environments (TEEs) - isolation

Industry terms like PoW or ZKPs describe a mechanism for how trust is achieved - the root being either through consensus, cryptography, or isolation.

(Simplifying a bit again, but this should help illustrate how different flavors of trust exist.)

Each flavor of trust has unique properties that shape its strengths & weaknesses:

PoW/PoS - Best-in-class guarantees, poor performance, slow provisioning

ZKPs - Highly accessible, good performance, fast provisioning

Restaking - Good performance, decent provisioning, immature

TEEs - Extremely performant, best provisioning, physically centralized

Let’s break it down:

PoW & PoS - Unmatched security, but slow

*Bitcoin’s massive PoW network is incredibly resistant to bad actors. It’s slow & took years to provision, but its guarantees are unparalleled.

*Ethereum is similar in its strong guarantees (other than that one time… but we don’t talk about that one time now).

ZKPs - Fast, flexible, but not best-in-class

*ZKPs can be provisioned quickly & computed on any hardware.

*Want to do something locally & prove it to the world? ZKPs are great - they can be spun up & torn down quickly.

*They’re more cost-efficient than consensus networks but don’t offer the same level of trust.

Each root of trust is best suited for different things:

*Want to create a new form of money that governments can’t censor? A massive PoW network is the right choice.

*Want to build a Web3 app that’s trustless & scales to millions of users? TEEs are the way to go.

And when you combine TEEs with consensus networks, you start unlocking powerful new properties - like scalability and best-in-class guarantees. Maybe we’ll save that for the next thread tho.

Hopefully, this helps frame how to think about design decisions based on your root of trust.

Each of these methods has a place in the stack - understanding them helps you reason about them effectively.

Thanks for hanging with me anon, let’s do it again soon!

TEEs - Insanely fast, but centralized

*TEEs provision trust at the hardware level.

*With the right setup, you can turn a datacenter into a root of trust - bringing powerful compute, fast networking, and local storage into the equation.

*But since TEEs are physically centralized, they’re more susceptible to censorship.

Restaking - Interesting, but untested

*Restaking extends PoS guarantees while dramatically improving provisioning time.

*It can also improve throughput (not diving into that here).

*If you want to build a Web3 app quickly while leveraging PoS security, restaking is interesting - but still pretty new.