Trust has been an important word throughout my life. Many that know me well would say trust plays a bigger role in my worldview than for anyone else they have met. At the core of web3 is a belief that trust is no longer conferred by a centralized intermediary. Using the same framework that I referred to in my first post about web3, I think Clayton Christensen’s theory of disruptive innovation is relevant as we consider decentralization. I’ll explain.
At a base level, effectively designed decentralized systems don’t require the same type of trust as traditional centralized models and are more resistant to collusion (as Vitalik points out). As Albert Wenger highlights here, web3 allows for permissionless data. Trustless data. The databases don’t need to be controlled by a single entity (Apple, Meta, etc). As a consumer, you may not have an issue trusting Apple, but if you run a company that distributed through its App Store, you would. The trust we put into social networks, and arguably ad-based business models in general, has had scary consequences. A truly decentralized web3 doesn’t require putting trust in centralized companies. Great. Does it do more than that, and how much does that matter?
The power of decentralization took me a while to fully appreciate. I don’t believe that complete decentralization is the optimal answer in all situations.1 A decentralized product is censorship-resistant and (practically) can’t be shut down. While China can ban Bitcoin mining, even a governmental attack at that level doesn’t shut down a decentralized system (and the miners move to other countries). However, decentralization can be clunky and effectively forgoes scalability and usability. I’m not convinced a decentralized organization can move a product vision forward as efficiently or effectively as a centralized org can (see below). Pause on those thoughts. Under some of the key evaluation criteria for web2 products, web3 products are, at least today, often significantly worse (eg. the UI/UX on most web3 products is really tough). But under a different dimension of quality, i.e. removing trust from the equation, web3 wins hands down. Thus the disruptive innovation framework, or at least the start of it.
When I say the “start” of the disruptive innovation framework, I mean that even if we agree there is a new dimension of competition (or quality, as I typically say), that isn’t enough. Web3 will also need to move upmarket over time in order to disrupt companies and products.
Both centralized and decentralized users can become (effectively) equity holders via token issuance in web3. Suddenly stakeholders have a financial incentive to help the product grow (counter-argument: this can be confusing to builders!). Is this possible in web2 without token issuance? Perhaps. But realistically the legal paperwork creates too much friction, to say nothing of the fact that the equity isn’t liquid in the same way. However, I’m not sold on many of the oft-cited use cases for decentralization. For example, I question whether an Uber-type marketplace could be decentralized or whether the customer service, fundraising, and strategy work could effectively be performed by a decentralized entity. But perhaps a token would help with driver acquisition and retention, particularly as the world (drivers and riders) becomes increasingly crypto native. So maybe the answer is a centralized Uber (for governance and shared functions) and some sort of token that rewards users and changes the dynamics of the S-curve that platforms naturally experience. Wow, did I just waiver enough in this paragraph?
Decentralization also facilitates the building of apps that use pseudonymous, wallet-based identities that can span across web3 and don’t require the trust of a central identity (like your Google sign-in does). Not just smoothing out the onboarding but more importantly being able to take your identity and track record with you… or not. This facilitates permissionless commerce that allows users to browse, interact and transact across web3 apps and services without friction. It also allows users to bring their audience and data with them across platforms, thus having a single identity across the web (best discussion of data portability is here from David Phelps). People can own their own data and not trust a third party to keep it safe. This will matter for some users. And yet I also recognize that today the UI for many of the decentralized applications (dapps) is clunky because we’re still so early in the development cycle — which is by some measures equivalent to the internet in 1998 (you don’t remember how bad AOL was then, do you?).
For builders in the space, particularly those building dapps, the task of decentralizing a company appears (at least to many web2 CEOs) a bit more daunting than their previous roles. Step 1: Build a product people want. Step 2: Build a community. Step 3: Give ownership of the product to the community (see A16Z Crypto School; Jesse Walden & Robert Leshner2). Step 1 seems familiar. Step 2 seems fuzzy, not something that many CEOs have experience of actively cultivating. Step 3 seems… antithetical to everything most web2 CEOs know. Going back to the earlier Uber example, perhaps Uber in a crypto native world starts centralized, builds a community of drivers and riders, and over time provides tokens that have economic and perhaps even governance power. I can imagine additional tokens for recruiting other drivers/riders, loyalty, or other things that provide value back to the network.
While some parts of decentralization appeal to me, including those already mentioned, other parts worry me. Can a decentralized team continually innovate and grow a product? Decentralized products tend to stay in stasis. While the benefits may be real, it is hard to create a vision statement, or a product roadmap, based on “wisdom of the crowd”.3
I remember that a year into being CEO at my last fintech startup, we tried to write our mission statement. We asked for input. The input came from brilliant teammates that had relevant experience and wonderful values. I couldn’t have pictured a better team for the task. And it was a disaster. There are some things I don’t understand about how to run by committee — and mission and vision are in that bucket.4 Perhaps that means the most successful decentralized products will (continue) to be built by centralized teams. Or maybe the centralized teams decentralize over time if that provides value to the users. Perhaps subDAOs are an answer to help modularize the tasks of a typical leader, or perhaps the answer will unveil itself down the road. I tend to agree with Sarah Tavel that this isn’t a race to decentralization but more a race to create consumer value. Do users care enough about the benefits of decentralization (no required trust) to give up the benefits of centralization (usability and scalability)? I think it will depend on the product and the problem to be solved.
The relatively recent argument around the level of decentralization (just how decentralized is it really?) frankly bores me. Maybe I’m still too naive to understand the implications, but I don’t view the benefits as binary. Start with the problem to be solved for the users, not whether the product/team is decentralized. For some applications, decentralization should help, for others probably not. What matters is whether the value is created for the user.
- Does anyone think that?
- No really, you should watch them all, they are terrific.
- Interestingly this dynamic can be seen very strongly in some of the original “DeFi Summer” projects on Ethereum where some founders got rich off their ETH holdings and subsequently failed to innovate. H/T to Erik Vandekieft for this point.
- If Satoshi Nakamoto is actually a group of people, then that would be a fascinating counter example to my point. Btw, read the white paper. No really, do it.