The pros and cons of decentralized platforms
Blockchain technologies and smart contracts have created the possibility for decentralized autonomous organizations (DAOs). For regular firms, this means replacing the traditional principal-agent problems between investors and managers, and between managers and employees, by flat, collectively owned and collaborative structures, where everyone owns a share (in the form of tokens) and has a vote in what the organization does. For platform businesses, this means it is now possible to create decentralized platforms, where users can discuss governance and design proposals on Discord and can vote on them (typically via Snapshot, a popular DAO voting portal), being rewarded with tokens for their active participation.
In this post, together with economist and blockchain expert Hanna Halaburda, we evaluate the advantages and drawbacks of decentralized platforms, relative to traditional, centralized platforms.
It is useful to start with three examples of decentralized platforms: a marketplace for products (Rarible), a marketplace for services (Braintrust), and a social network (Minds).
As we’ve discussed in a previous post, Rarible is an NFT marketplace that has opted for decentralization to compete with OpenSea’s centralized approach. Since its founding in early 2020, Rarible has steadily moved towards becoming a full-fledged DAO. It incentivizes people to build on top of their open protocol, rewarding the early builders with their token RARI which was released in July 2020.
The primary function of the RARI token is to give holders the power to submit and vote on proposals that impact the marketplace (the token determines voting power, but need not be spent in order to actually vote), to moderate creators, and to curate featured artwork. Currently, every week, 41,250 of RARI tokens are distributed among applications building on Rarible Protocol (including Rarible.com), proportional to sales volumes in a program called Marketplace Liquidity Mining (MLM). Some of these are distributed as rewards to buyers and sellers for NFT trading, with the exact formula outlined here. Buyers and sellers get an equal split of the tokens; verified users get 15% more tokens; for trades of NFTs created outside of Rarible, buyers and sellers get 50% less. Other tokens have been reserved for airdrops and for investors and Rarible team members.
An interesting challenge noted in a current proposal being voted on is that most community members convert their tokens into Fiat currency (as they say, “people need to eat!”). As a result, these types of participants end up with little or no voting power. The proposal is to grant a group of five community members 20,000 tokens and to grant the two co-founders of Rarible another 20,000 tokens for the purpose of voting on governance matters. This is viewed as an interim measure until key contributors have built up sizable holdings of tokens themselves. And this illustrates a broad issue with decentralized platforms: how can they sustain incentives for participants to hold tokens for voting purposes if they don’t expect the tokens to appreciate in value, i.e. if there is no speculative reason to hold tokens?
Founded in 2018, Braintrust is a two-sided marketplace matching top technology talent (developers mainly) with clients (enterprises) who want to hire them. Its blockchain-based, decentralized approach stands in direct contrast with its centralized counterparts such as Fiverr and Upwork. Braintrust users can earn its BTRST token by building features of the platform, inviting and vetting talent, and referring clients. After discussing ideas in the Braintrust Discord group, proposals for governance changes can be voted on-chain or off-chain.
On-chain voting can only concern the modification of a limited number of parameters (this is restricted by the BTRST smart contract). Right now, these are: a) admin permissions, b) the marketplace fee charged to employers/clients (there is no fee charged to talent), and c) the categories featured on the platform (currently, they are engineer, designer, product and other). Proposals that pass an on-chain vote are then executed automatically on the smart contract.
Proposals regarding any other governance aspects (not covered by on-chain voting) are handled via off-chain voting in a dedicated discussion tool. Users need to submit proposals, allow several days for discussion, distill into specific recommendations, then vote. Outcomes are decided using rules detailed here.
As of right now, there is no clear commitment regarding the implementation of proposals that pass an off-chain vote, but Braintrust plans to add more and more decisions to on-chain voting.
Launched in 2015, Minds’ stated goal is to provide a decentralized alternative to traditional centralized social networks like Facebook. In a 2018 Wired article, Minds was actually described as the “Anti-Facebook”.
Like other decentralized platforms, Minds has its own utility token MINDS (issued on the Ethereum blockchain). Similarly to Braintrust, users can earn the token by posting content (more popular content earns more), being active contributors, finding and fixing software bugs, referring new users, etc. In turn, users can spend tokens to boost viewership of their content on Minds.com beyond their current connections/subscribers (akin to buying Facebook ads to have one’s content reach a wider audience), or to pay other users for access to their content (in the form of tips or subscriptions).
The governance of Minds is an interesting blend of centralized and decentralized. The terms of service that define what content is allowed and what content is objectionable are entirely defined by Minds.com. It is worth noting that Minds keeps its content guidelines quite short and aims to be committed to free speech. In particular, it does not try to define hate speech or mis-information, as traditional social media networks do, ending up with long and unwieldy content rules.
The enforcement of content moderation on Minds is mostly decentralized and relies on a jury system. Specifically, when a user reports a piece of content or a channel for inappropriate content, the following sequence is put in motion:
Minds staff can accept or reject the report.
If the report is accepted, the user under investigation is notified and given a chance to appeal.
If appealed, a jury of Minds users votes on appeal - the decision of the jury is final.
For each report, the jury consists of 12 randomly selected Minds active users who are not subscribers to the channel of the user under review. If 75% or more of the jury members vote to accept the appeal, the administrative action is overturned. Otherwise, the administrative action is upheld.
With these examples in mind, let’s start with the advantages of decentralized platforms.
Credible commitment to providing value in the future
Perhaps the most obvious and important benefit of decentralization (especially when implemented via smart contracts) is that it can act as a credible commitment not to hold up participants in the way some centralized platforms do. The most important forms of hold-up by traditional centralized platforms are increasing fees over time and/or introducing new types of fees (e.g. DoorDash with restaurants, PropertyGuru with listing agents, and Amazon with sellers as detailed in our earlier post), and unilaterally changing the governance rules and their enforcement in unpredictable ways (e.g. Apple with rules for third-party apps in its App Store, Facebook/Twitter with hate speech and misinformation).
By contrast, well-designed decentralized platforms can commit (e.g. via smart contracts) to clear and transparent design and governance rules, which can only be changed in restricted ways via majority voting. It is therefore unsurprising that the transaction fee charged to clients is one of the first three parameters subject to on-chain voting on Braintrust as discussed above - indeed, transaction fees are presumably a major concern freelancers have with centralized platforms. Similarly, a key way in which Minds differentiates itself from traditional social networks is through its decentralized content moderation: what users hate the most about Facebook and Twitter is the inconsistent and ad hoc enforcement of their content moderation rules.
Equity is often issued in centralized platforms to align incentives, for example between founders and senior employees. Issuing tokens can be seen as playing a similar role. However, they may align incentives better than equity. This is because tokens can be issued to a much wider group of stakeholders, including users and developers (not just employees). The ability to hold tokens and participate in governance voting can enable the creation of vibrant communities, encouraging stakeholders to invest time and resources to improve and promote the long-term interests of the platform. For example, this can be one way to address the usual chicken-and-egg problem of starting a marketplace, as we discussed here.
There are at least a couple of reasons why tokens can be issued to a wider group of stakeholders:
The primary reason is regulatory arbitrage. There are numerous securities laws that would make it prohibitive to issue small amounts of equity for users and developers. Because tokens are often framed as utilities and not securities, they circumvent these existing securities laws (at least for now).
A secondary reason is that blockchains and smart contracts may make raising funds via tokens more efficient. This is especially true when compared to private companies raising equity. For example, the process of closing a private equity round, where all investors settle at roughly the same time contingent on others also settling, may be more efficiently handled by a smart contract on the blockchain. And by having everything transparently on the blockchain, it will be more obvious when a new set of investors tries to introduce opportunistic terms that align with the founders to exploit existing outside investors, thereby reducing the likelihood of such behavior. In some sense, issuing tokens is like going public, and the blockchain has made that option possible even for early-stage companies without revenue.
We now turn to the drawbacks of decentralized platforms relative to centralized platforms:
Ruling by committee
The most obvious flaw of decentralized platforms is that they rely on decision-making by (very large) committees. This creates several problems.
First, participants may have diverging or heterogeneous objectives that are hard to reconcile in a unified strategy. This can lead to disputes that pits the different interest groups against each other, possibly resulting in lobbying and politicking, or worse, the forming of competing factions that do not work in the common interest.
Second, participants may engage in opportunistic behavior. For instance, speculators could seek to gain control over governance and vote for actions that drive up the token price in the short term but are not in the long-term interests of stakeholders. This is much harder to do with traditional private companies that issue equity, since this would typically require obtaining control of the board and replacing the existing CEO, something that is hard to implement when the founders decide which investors can participate and whether they get board seats.
Third, decision-making by committee is inherently slow. By contrast, a centralized platform operator making decisions unilaterally is much more likely to come up with a clear objective and execute on it faster, something that can be crucial in fast-changing competitive environments with uncertainty.
Fourth, voting on decisions suffers the usual free-riding problem of voting. If it is costly to evaluate proposals and to vote, and no single vote is likely to determine the outcome, participants may just rely on others to do so, free riding on their efforts. This can lead to low participation in the voting process.
Another problem with decentralized platforms is that it requires participants to have a certain level of sophistication and knowledge in order to productively participate in governance decisions. That works well for, say, Braintrust, given their workers are tech developers who know software and are likely to have a reasonable understanding of how DAOs work. It would likely work a lot less well with, say, a decentralized version of Uber, where the average driver may not be in a good position to make good and informed decisions on governance issues.
The flip side of the regulatory arbitrage that can be exploited by decentralized platforms is the regulatory and legal uncertainty they face. Founders and investors may not know if they are acting within the law and whether they face potential liabilities for their actions. They also face the risk of regulatory changes that could undermine their whole structure. Many questions remain unanswered. When do tokens become securities? What taxes are payable? Can DAO’s own property or enter into standard contracts? Can they be sued?
Based on the tradeoffs above, it is apparent that the decentralized model may be more suitable for certain types of businesses and centralized for others. At a high level, there are two main conditions that create a particularly favorable environment for decentralized platforms to succeed:
A market where network effects push towards winner-takes-most competition among platforms and there is a severe risk that users will be exploited by the winning platform (through higher fees and/or adverse changes in governance rules). This makes decentralized platforms particularly appealing to potential users, especially if they are also good at addressing the inherent cold-start problem.
Platform participants have fairly homogeneous preferences and interests. This is probably true for Braintrust provided it remains mainly focused on top-tier software developers on the supply side of its marketplace. It is likely less true for Rarible given the huge variety of NFTs and diversity of buyers, sellers and developers building on it.
Ultimately, we think the decentralized platforms we’ve discussed above (Rarible, Braintrust, Minds) are unlikely to cause their centralized counterparts to go out of business anytime soon. Yet, they can probably co-exist and attract a certain set of participants, particularly those that feel exploited by the traditional centralized platforms, or that have an intrinsic preference for decentralization. Indeed, there are many reasons why people may seek decentralization: strong aversion to any form of censorship, strong preference for transparency and transaction security. Blockchains and smart contracts enable such decentralization to an unprecedented extent, while also allowing for monetization, which was a challenge for historical decentralized platforms.
Finally, it is worth pointing out that the centralized vs decentralized choice is not a 0-1 decision. Blockchains and smart contracts have opened up a whole continuum of degrees of decentralization that platforms can consider. Unless one is trying to attract decentralization maximalists, the optimal choice will usually be to decentralize only the design and governance decisions for which transparency and commitment matter a great deal (e.g. the level of fees charged to participants, the rules for de-platforming participants). Other decisions (e.g. competitive strategy, hiring and firing, resource allocation, etc.) are best kept centralized.
Interestingly, the Ethereum white paper (I just read it) talks about DAOs and the author means by DAO “ a virtual entity that has a certain set of members or shareholders which, perhaps with a 67% majority, have the right to spend the entity's funds and modify its code.” https://ethereum.org/en/whitepaper/#decentralized-autonomous-organizations He allows for a pretty general definition of DAO where DAO can be Decentralized Autonomous Corporation with shareholders as well as Decentralized Autonomous Community with 1 vote per person. In both cases Ethereum just provides an infrastructure to safely make collective decisions (voting) and distribute benefits. But I feel that currently, the notion DAO gained the anti-establishment connotation and usually means something like a community of enthusiasts.
"... collectively owned and collaborative structures, where everyone owns a share (in the form of tokens) and has a vote in what the organization does. " -- It seems that DAO is a reincarnation of a standard corporation with shares but in the land of crypto. I, as an employer of a tech company, receive part of compensation in the company shares and can technically vote in shareholder meetings. Do I understand correctly that the benefit of doing things with tokens is the overhead is smaller (eg, verification and enforcement)?