Jesse Walden, a former a16z investment partner and Mediachain cofounder, and Robert Leshner, founder and CEO of Compound, do a “Deep Dive on Decentralization.” Walden starts with a playbook for progressive decentralization — the process by which crypto project creators build a useful product, create a community around that product, and then gradually hand over control of the maturing network to the community. This process is in keeping with the cooperative model of crypto networks, which drives rapid, compounding innovation through better alignment of incentives and open participation. Leshner follows with a case study of his experiences at Compound, an automated money market for crypto assets in which lenders and borrowers can come together to transact without the involvement of third parties. Compound, one of the first crypto projects to move through the full progressive decentralization model, built a thriving community of third-party application developers, who have set up shop on top of Compound’s smart-contract protocol. The Compound team has gradually brought this community further into the protocol’s inner workings; in the final stages before handoff to the community, the founding team made changes transparently, with greater reliance on the community’s input, and created a sandbox for experimentation to test governance mechanisms. Decentralizing “allows the protocol to live forever,” Leshner says, which fosters innovation because developers can trust the protocol with their businesses and livelihood.        

Andreessen Horowitz’s Crypto Startup School brought together 45 participants from around the U.S. and overseas in a seven-week course to learn how to build crypto companies. Andreessen Horowitz is partnering with TechCrunch to release the online version of the course over the next few weeks.

Find more Crypto Startup School videos plus additional reading and info: https://bit.ly/2ThsHen

I'm the program manager for Crypto Startup School and prior I was on the investment team at Andreessen Horowitz.

Prior to that, I was also a founder of a project in the space called Media Chain.

And today I'm going to talk about progressive decentralization and give a playbook for building crypto applications.

Now before we dive in to this playbook, why are we even talking about decentralization and why is it so important for crypto apps?

Decentralization is important because it's a way to build trust.

The internet is based on open decentralized protocols like IMAP, POP, and IP, and these form the basis of the internet services that we use today.

On top of these neutral protocols, entrepreneurs have built some of the most valuable companies in the world.

These companies are valuable because they extend the functionality of the bare internet protocols and give us rich experiences like search, email, and social media.

But today, we know that these types of internet companies follow a predictable pattern.

When internet platforms start out, they do everything they can to recruit users, developers, and businesses to join the platform, because each of these community members forms the basis of a multi-sided network effect that makes the platform more valuable and defensible.

Yet, as we know, as platforms grow, they move up this adoption curve and they follow a predictable pattern where they slowly start to change their behavior.

Instead of cooperating with their communities, they start to extract value from them.

For example, many web 2 platforms used to have rich developer ecosystems.

I think at 1 point, Twitter had more traffic going through third-party clients than through their own official app.

But in order to capture more of the value, Twitter changed rules and handicapped the API and centralized the client.

Spotify and countless other platforms have done the same thing.

So for users and developers, this is a bait and switch.

These rich ecosystems that started out cooperative ended up getting killed off and the innovation emerging from them was stifled.

In contrast, crypto networks offer a new model for building internet platforms.

Unlike corporate-owned networks, crypto networks aspire towards being decentralized.

And what I mean by that is that they strive to be community-owned and operated.

Well, here it's important to understand that crypto tokens are an exciting new innovation in that they enable value to flow at the cost of information.

And that means it's now possible to financially reward independent users all over the world for their contributions to a network.

And the result is that crypto networks can have a more cooperative economic model.

Instead of having to extract value from users in order to return that value to shareholders, well-designed crypto network routes value directly to those who contribute it.

And that in turn means that crypto networks can be much like earlier internet protocols in that they can grow to much larger scales of adoption while enabling sort of unabated and innovative ecosystems on top.

So that's why decentralization is so important.

And now I'm going to talk about sort of how building in crypto is actually different from traditional platforms.

So in order to explore that, we should probably start with things that are the same, because there is a lot that is the same.

In traditional startups and crypto startups, you need product leadership, you need rapid iteration, and a managed go to market.

But this is where the differences start to come to the fore because in crypto, you also need to consider how the product can successfully run in a community-owned and operated manner.

So, that's tricky because community ownership is often at odds with these other 3 things.

Product leadership, rapid iteration, and managed to go to market require a core team.

So, how do crypto founders resolve this tension?

So to help founders think through this I'm gonna run through this three-step process that hopefully serves as a guide for how to run the process of progressive decentralization And taking a step-by-step approach allows teams to focus on the right thing at the right time, and also posits a path towards regulatory compliance.

And so before I jump in, I want to note that, you know, crypto is still operating in a pretty uncertain regulatory environment.

And so the final stages of this playbook are really only pointing towards potential solutions rather than identifying proven winning strategies.

But you can think of this as a possible path forward.

All right, so let's start with an overview.

So I believe that a successful crypto project will need to execute 3 objectives in this sequential order.

The first should be obvious, which is building a product that people want.

Without a product that people want, you don't have much.

So objective 2 is building a community around that product.

And then finally, the thing that's new and exciting about crypto is giving ownership of the product to the community.

The earliest stage of building a crypto application is all about finding product market fit.

And as I mentioned earlier, this requires a lot of the same ingredients as a normal startup.

You need a great team, you need lean development, you need tight execution and quick learning.

And again, all of this is in service of product market fit.

And here it's important to avoid design by a community or the community.

Great product needs opinionated leadership to do all these things, to hire a great team, to execute quickly.

And so in practice, this can mean having admin privileges on an application that uses smart contracts.

For example, that would allow the team to effect upgrades or shut down the application quickly as needed, and in turn, iterate rapidly.

So at this stage, there should be no pretense of decentralization, since a core team is driving all product decisions by necessity.

And accordingly, launching a token at this stage probably isn't advisable because the product is dependent on the core team's efforts and dependence on the efforts of others is 1 way you might fail the Howey test, which is an SEC regulatory framework that you're going to hear a lot about in the coming weeks.

And you don't want to fail that Howey test because thinking about securities regulation is a huge distraction from product-market fit.

So the idea of a core team controlling all the aspects of a project may sound weird because we're talking about decentralization and it's likely to ruffle the feathers of some of the people in the early crypto community who hold decentralization you know to in high regard But if users are complaining about the control you've got, that's actually a good problem to have because it means that someone cares about what you've built.

So that said, it's pretty important to communicate about control where it exists, because if you're faking decentralization, that's a quick way to undermine trust with your users, whereas transparency about control is a better way to build it.

So At early signs of product traction, which could include a growing user base, developer ecosystem, or some network effects starting to take shape, it's time to start devoting more cycles from thinking about product to thinking about how to engender harmony between the core product team and the community.

And that's what objective 2 is all about.

So this is going to look different for different types of communities, but a lot of the products building in the crypto space are developer facing.

So for developer facing products, community building might involve investing more heavily in the best practices for running open source communities, which can include good documentation, building a developer evangelism program, or issuing bounties and grants and other incentives for third party development.

And going even further, it could make sense to start thinking about how to eliminate dependency on the core team in order to build more trust with the community.

And so limiting control over the product is 1 way that you can start to invite community participation.

For example, Compound, who we're going to hear from later, made a change to their smart contracts that delayed upgrades for 48 hours, which allowed time for the community to sort of review the upgrades and respond.

So relinquishing some control comes with this opportunity to invite community participation.

But as you go about engaging community members, it's important to revisit their incentives and ask the question why will community members contribute to the product in the first place.

So last week you heard from Ali about business models for crypto projects and hopefully you recall from that presentation that the business model for crypto is actually somewhat similar to the business model of web 2 platforms.

Crypto applications are multi-sided marketplaces and like web 2 platforms their defensibility comes from network effects.

Those network effects create switching costs which allow for the possibility of charging a fee so long as the fee That's charge is less than the cost of switching to a new platform This model is sustainable and so the concept of charging a fee for a service of course is nothing new But as we touched on earlier what's new in crypto is who benefits from that fee stream.

So because crypto tokens reduce the distribution cost of moving value to the cost of moving bits, it's now possible to distribute economic value directly to community members to incentivize their contributions.

That being said, it may not make sense to start charging a fee right at the outset of the project.

And it almost certainly doesn't make sense to just to start distributing fees right at the outset either because that has the potential to trip regulatory wires.

So what is important is to think about how to design an effective incentive and token distribution plan.

And part of that needs to mean conceiving of a distribution of incentives that's both fair and effective.

Fair in that it rewards past participation and effective in that it will sustainably reward ongoing contribution from community members in the future.

In 2017, we saw a wave of ICOs, which is 1 way to distribute value, selling tokens to a community.

But I think since then, we've learned that the ICO model is somewhat suboptimal in that ICOs invited sort of unengaged communities of speculators and a ton of regulatory scrutiny that distracts from finding product market fit.

So crypto applications that focus on product market fit have an opportunity to do better than ICOs did because they can distribute tokens to an already active user base.

And to start, teams might do this by testing an initial distribution of tokens with a managed group of community members.

A number of layer 1 blockchains have done this by running testnet programs where independent community members can register to run Node.

Those testnetwork participants are promised tokens from a future distribution, but only after they've demonstrated their ability to run the network.

So a permission group can be an effective way to demonstrate that a network can operate in a truly community-owned and operated fashion.

Once you've got a sustainable network operating a service, it's time to start thinking about how to scale this distribution of incentives to a wider group of active participants that might contribute to the network in the future and the wider community to boot that might join even later on.

So an effective distribution should of course consider the core team that built the product as well as the users that make it useful.

And some common questions to think through here are what percentage of tokens should be allocated to the initial team versus how will you, you know, reward different types of contributors to the product or service in the future?

And 1 important 1 is how will technology leadership be rewarded going forward?

And getting this right is hard, and I don't know that any single project has figured it out perfectly yet.

But starting to model this out is really important at the stage where you're thinking about building a community, because communicating about incentives and the roadmap for community ownership in the future is a powerful way to invite community participation.

And once you've got a model in mind it's it's time to move to the last and final objective which is all about actually distributing ownership to the community.

Now I keep coming back to this idea that crypto tokens are a new instrument for distributing value, including a fee stream to a community of stakeholders.

And what I mean by this is I mean actually giving that community tokens.

So in practice how do you actually do this?

I like to call this this stage exiting to the community because I think 1 way you might do it is by actually air dropping tokens to users based on the plan that you mapped out in the previous step.

So imagine there's a smart contract that mints and distribute tokens per the schedule, the distribution schedule defined prior.

At the moment that that smart contract is triggered and those tokens are minted and distributed, a number of exciting things will have happened.

Number 1, the core team will have ceded control of the application from the company to the community.

So the product is now governed and owned by its users.

And that of course mitigates the risk of the platform changing against the rules, against the wishes of the community.

The community can now sort of make changes to the product that they agree on.

And hopefully another thing that will happen is the company will found a sustainable business model by having retained enough tokens to benefit from the stream of this product and its growth in the future.

And similarly community members who are now part owners of the product will also start to be rewarded economically for their contributions and hopefully they will see increasing returns to scale as the cooperative economics of the service allow for continued growth.

And finally Because the product is no longer dependent on the core team, it's possible that the token may not be considered a security, which eliminates some of the regulatory overhead that comes with SEC securities regulation.

So the spirit of this objective is really all about marking a specific moment when you call the smart contract, where a crypto product completes its journey from a traditional product company to a sustainable community owned and operated network.

Now, a reminder, if you're a crypto founder and you're ready to exit to the community that means you've done objectives 1, 2 and are on number 3.

So you've built a product that people want, built a robust community around that product, and figured out a way to incentivize community participation on an ongoing basis such that you can turn over ownership to the community.

And I think right now we haven't seen a crypto app application execute on all 3 of these yet, but there are a few that are well on their way.

And 1 of those is Compound, which is why I want to invite Robert Leshner, the founder and CEO, to help us think through how they've been approaching progressive decentralization in real time.

And so for those that don't already know, Compound is a money market for borrowing and lending crypto assets, and it runs a smart contract on Ethereum.

The company was founded in 2018, and they've been executing a progressive decentralization plan ever since.

So with that, I'm gonna invite Robert to join me here on the Zoom and we're gonna sort of have a little chat about what their journey has been like.

And afterwards, we're gonna open it up to questions from you guys.

From, looks like from the Compound office, which is great.

First, I briefly introduced Compound, but maybe we can just start with your quick summary.

So Compound is a market for interest rates on crypto assets.

It lives as a smart contract on Ethereum, and for a number of supported aetherium assets our protocol has created interest rates that an application or a user or any aetherium address can earn if they have the asset and supply to the protocol or that an application or user can pay to borrow an asset from the protocol.

That's a much better explanation than the 1 I gave.

So also, I'm curious, I'm sure a lot of folks are curious.

So before this, I went down the traditional Silicon Valley founder and product manager track.

So I founded 2 software companies prior to this, built teams, built products, brought them to market.

And I also worked in a large tech company as a product lead, going through like a large software development cycle, building hardware and software.

So I imagine a lot of that experience was useful and applicable to Compound.

It's actually surprising to a lot of people, but I think there's a lot of similarities between developing a product for crypto and Developing a product that's traditional software.

It's more similar than it is dissimilar in a lot of ways.

Okay, cool I mean, yeah, I agree as as you probably just heard in my presentation I think you know finding product market fit is still the very obvious number 1 objective for all founders in crypto just like traditional startups.

So I'm curious, what did Compound look like at the early days as you guys were striving to find product market fit?

So most people don't know this but there's actually been 3 versions of Compound that we've built over time.

And the goal has always been to find product market fit, to iterate, to incorporate feedback from users and the community, and into a cycle of building a better product.

The cycle for us really has been 1 of 3 phases.

In the first phase, we had version 0, which was an unreleased product that was really for our team, our friends, and a very small number of people to privately test and sandbox.

The core idea behind Compound, that we could create interest rates for cryptocurrencies.

It was clunky, it was difficult, it was ugly, but it helped us get the principles into the hands of a very small user base.

We then used that to build and launch the first main net version of Compound, which we launched in August of 2018.

And from there, there's nothing that gives you data and education and product feedback, like being live in the market.

And then we used the opportunity to be live starting in

to learn as much as we could before building and releasing what we consider to be the final version of Compound that incorporates all of the lessons learned from having the first product live in the market.

We made a number of incredibly large changes based on user and developer feedback, including tokenizing balances, which is now C tokens, including having risk parameters that are set for each asset and streamlining the code base to make it simpler for developers to use.

And so for us, our vision has always been the same from day 1.

We're going to create interest rates for crypto assets, and we followed a very traditional product cycle to find and refine the product until we've achieved a semblance of product market fit.

So it sounds like when you started out, you guys kind of built full stack.

You built the original protocol, you were the first consumers of that protocol, building your own user interface, and then you started to open it up to you know what you learn from you know building full stack and then you start to focus on the developer community.

Can you talk a little bit about you know this this sort of second phase of developer community building and what that looks like.

So most people won't remember this, but we've always envisioned Compound as a protocol that developers could build.

You could launch applications with interest rates built on top of Compound.

And This has always been the idea that we've been pursuing, but the very first version of Compound didn't see developers building on top of it.

What we did is we actually built the first interface on top of Compound version 1 in order to prove that we could, to dog food the process of building on top of the protocol, and to create a very basic interface to help bootstrap liquidity.

Because the basic idea is until there's usage, other developers don't want to plug in.

The same thing happened as we launched the second version of Compound, but this time we had the advantage of knowing what the underlying system should be to make it easier for developers to build on.

And so when we launched Compound version 2 we had a depth of experience so that we could launch with a number of partners.

From The moment that we put it on testnet, we actually worked side by side with a number of different teams preparing to go live for when we finally launched the production protocol.

And so when we launched, we actually did so with a couple of core developers that had been building with us the entire time.

So very much the case that you guys sort of you found product market fit by dogfooding and building a product yourselves, bootstrap some liquidity in your marketplace, and then shifted gears to, you know, focus on the developer community.

Can you talk a little bit about today what that looks like?

Like how do you work with the developer community and what's the relationship between the company and the community?

So this goes to phase 2 of decentralization, building a community.

We view developers as the users of Compact, as the community of Compact.

And at this point, 100% of our effort is focused on enabling developers to build on top of Compound.

For us, that means thinking about the product as being holistic and everything that developers need.

So we spend a lot of time doing things that most people find boring like working on documentation working on you know guides working on content working on things that really make the entire product experience when the user is another developer And so it starts off with thinking about what is the product that you're offering.

For us, that's a protocol and the ability to build on it.

We also spend a lot of time building a community of developers.

You'll notice that when we talk about, you know, how to interact with compound, we always say, join us in Discord.

It's a chat server that we really try to focus on developers and creating an environment where everyone can ask questions, can compare notes, can share code, can collaborate, can trade ideas, and to build collectively.

We probably have 1 of the most active discords in crypto and it's been a way to bring everyone into 1 place to enhance what everyone's already doing.

And lastly, we try to celebrate and market what our users built.

Our homepage of Compound is just a list of the products built on Compound.

We think that the most important thing we can do is enable other developers to succeed.

For themselves to go through the cycle of finding product market fit to building a community to decentralize on their own.

And that means doing everything we can to empower them.

It means helping them think about product development.

It means helping them think about fundraising.

And it really means leveraging our expertise and giving that as much as we can to other teams.

So earlier in the talk, I was, you know, at the very beginning, I talked about why community ownership is so important for compound, or for crypto at large, but also for compound.

So I guess I wanna hear from your perspective, you know, you've been very vocal about the goal of turning, you know, ownership over to the community members.

So for us, we believe that having a decentralized protocol does 2 things very powerfully.

1 is that it allows the protocol to live forever.

Right now, you know, our team is necessary to maintain the protocol, but if we can distribute the responsibility and the ownership of the protocol to a wide audience, there's the expectation that it can run forever.

I like to say that, you know, if Bitcoin was run by the Bitcoin Corporation, nobody would trust it or use it.

It's the fact that it's not run by any 1 company that you expect that it's going to be there next year, next decade, next century.

And we wanna put Compound into the same position.

You know, That's the biggest advantage, especially if you're building a developer-centric product and something that you want developers to build on.

They want to know that it's going to be there.

They want to know that the rules aren't going to change because, you know, I wake up 1 morning and feel like it.

And they want to know that it has a sense of permanence.

So that's the first major advantage of decentralization.

The second, and I think this gets talked about a lot less frequently, is that it enables a wider audience to upgrade and to contribute to the protocol.

Right now, you know, we're a small team based in California.

There's only so much that we can do to upgrade the protocol.

But if you can open up to the entire community of open source developers to enable them to contribute changes, upgrades, and improvements, then the speed at which compound advances can accelerate on its own.

So we see that as 1 of like the biggest things that's gonna change is that the pace of evolution is going to accelerate once we decentralize.

Yeah, so I guess another thing I wanna touch on is, right now you're in the middle of this process, right?

You're you're sort of you've got the community going you're trying to figure out how to turn ownership over to them I'm wondering like has there been attention there?

How do you keep everyone sort of rowing in the same direction?

So since day 1, we've always communicated that the goal eventually was to decentralize the protocol.

And when you're day 0, just launching, nobody really takes you at your word because you haven't proven anything yet.

Everyone says, oh, of course, you're planning to decentralize.

But it's been like us pounding the drums about this for 2 years consistently.

And we've taken a number of steps along the way, such as restricting our ability to make sudden changes to the protocol, creating a significant number of pages around transparency of governance, starting to show people how the protocol operates.

We're now going through what I consider to be the final 3 changes as we start to hand it off.

And this really is step 3 that Jesse was talking about before, which is the decentralization process.

And right now we're starting to talk about it heavily.

See all a16z crypto transcripts on Youtube

Robert Leshner and Jesse Walden: Deep Dive on Decentralization