a16z's latest research: Reputation-based dual-token economic model

Authors | Jad Esber, Scott Kominers
Compiled by | Gu Yu

Reputation systems provide platforms with an opportunity to identify and incentivize high-quality contributions from participants, including content creation, review, community building, and gameplay. This is crucial for the growth and sustainability of any web3 project. However, designing a reputation system requires complex considerations around reputation supply, distribution, credibility, and more.

As a result, while many are exploring this space—from DAOs like FWB to play-to-earn games like Axie Infinity and new social platforms like BitClout—builders have yet to reach a consensus on the best way to design these reputation systems.

Leveraging our knowledge in economic theory and game design, we argue that a dual-token reputation system design can serve as a tangible representation of meaningful contributions, with one token representing reputation and the other providing liquidity.

The Evolution of Reputation Systems

The fundamental premise behind reputation systems is not new. Since the dawn of civilization, we have invented reputation markers, such as badges awarded for merit or service. In the business world, employees are assigned titles or levels to indicate their position in the hierarchy—these "markers" often determine a person's salary and other benefits.

For a long time, games have also been pioneers of digital reputation systems. The "points" accumulated by players during gameplay can be converted into "currency" within the game, used to purchase new skins, weapons, characters, and more.

Overall, reputation tokens in the crypto world take the form of social tokens. These tokens can be issued by various entities (including individuals and communities, as well as games and applications) and can be used to acquire social capital, access services, and/or convert into rewards (financial or otherwise).

Importantly, in the crypto environment, social tokens can often represent ownership and uniqueness (in the case of NFTs), and due to the existence of blockchain, social tokens are decentralized and portable, allowing them to be used in the global internet economy rather than being restricted to a single platform or decision-maker.

The Paradox of Reputation Tokens: Are They Earned or Bought?

Reputation tokens on digital platforms typically serve two purposes:

• To identify and reward users who contribute value to the platform—a form of signaling that these users can leverage to enhance public reputation.

• To provide a form of compensation that allows contributors to liquidate some of the value they create into redeemable currency.

However, these two roles are in opposition. Tokens need to be exchangeable to have liquidity. But the higher the liquidity of the tokens, the worse their effectiveness as a pure reputation signal.

To illustrate this, suppose reputation tokens can be freely transferred. If the tokens are a credible signal of reputation quality, their trading value will be high, making holders willing to trade them. However, once tokens start trading, ownership loses its signaling value, undermining the reputation capital that the tokens nominally convey.

For example, a charity might start minting NFTs to reward those who complete over 500 hours of community service. However, if these recipients can sell their NFTs to anyone they want, then whenever you encounter a holder, you must wonder: "Did that person earn their NFT or buy it?"

Even if no one publicly trades their community service NFTs, the possibility of private sales diminishes the signaling value. In a fully liquid market for such tokens, the signaling value completely disappears.

This creates a paradox: if a token can be easily transferred, then those without reputation can simply buy it, reducing the token's ability to serve as a reputation signal.

Remember how easy it was to buy followers on Instagram? This made followers a much weaker metric for measuring reputation, to the point where brands began looking for engagement metrics that were harder to "buy."

That said, high prices can also reduce transferability. For instance, buying a CryptoPunk is now very expensive, and almost all holders must have purchased them in advance. This seems to eliminate the paradox—restoring the reputation value of the tokens—but because some people are willing to pay high prices for CryptoPunks to gain the appearance of reputation, the signaling value may still diminish. Currently, the number of people entering the space also has a natural boundary. As more people adopt crypto technology, imagine the scale of this problem.

Moreover, the decline in reputation capital associated with tokens can feedback into the market value of the tokens. If the tokens lose their ability to indicate reputation, then interest in trading them will decrease. In fact, as people began to buy Instagram followers in bulk, those followers lost their signaling value, which diminished interest in purchasing them for influencers.

Making reputation tokens transferable not only reduces their ability to serve as a reputation signal but also undermines their capacity as valuable compensation. Therefore, building reputation capital requires completely (or at least largely) non-transferable tokens. The question then becomes how to convert reputation into liquidity.

Social Capital Should Not Be Bought—But That Doesn’t Mean It Can’t Generate Liquidity

Tokens have signaling value when granted by credible sources (such as brands, universities, or governments). But in crypto and blockchain environments, there is not necessarily a centralized third-party source granting that authority.

Thus, if someone possesses certain potential characteristics, the second avenue for conveying value is to make tokens easier to obtain. For example, if you are truly skilled or very diligent, then achieving a high score in a video game is much easier; thus, a high score signifies some mix of skill and effort. High view counts on platforms like YouTube are similarly indicative.

The aforementioned paradox—easier transferability, less signaling capacity—arises because making reputation tokens transferable decouples them from the foundational institutions and/or efforts that serve as sources of signal value.

So, what if we separate transferability from the tokens themselves? This is why many games have separated scores or points from tokens. It’s easy to imagine how spending your "points" could backfire.

We propose a dual-token reputation system where one token, which we call points, serves as a non-transferable reputation signal. The second token is a transferable asset, regularly distributed to points holders. In practice, points are split into dividends that can be used as a tradable currency. Furthermore, since the tokens are accumulated by points holders, they are also related to potential reputation.

At a higher level, this design fosters a feedback loop where users can earn points from high-quality contributions on the platform (such as contributing content, reviewing, or winning gameplay). Then, when users with points receive tokens, they can be traded as currency. The demand for tokens drives the demand for earning points, which in turn incentivizes high-quality contributions.

Note that while we use points as a non-fungible language and tokens as a fungible language, the precise implementation may vary by application. The key is that contributors receive a non-tradable token, which derives a tradable token.

Points Should Reward Contributions

To maintain the signaling value of points and incentivize high-quality participation, they must be linked in some way to users' contributions. In the context of games, scores may simply be awarded algorithmically as a function of performance. On creator platforms like YouTube or TikTok, points may be a direct function of how people interact with specific creator content. In other cases, such as publishing platforms like Mirror, there may be a set of users empowered to award points or a governance/voting process that determines point distribution.

Importantly, points must reliably connect their holders to sources (or drivers) of reputation. Additionally, there needs to be a good understanding of the reward point rate as a contribution function, so users can understand how much effort they need to put in to reach a given point level. In other words: participants need to know the rules of the game before they start playing.

In many cases, points do not need to be scarce. For example, there is no limit to the number of badges Discord can award to moderators. However, scarcity can enhance or reinforce the reputation value of the system. For instance, the "Moderator" role on a Discord server is only granted to a select few, so those with a moderator badge are seen as having higher reputation on the server. If the server simultaneously starts making too many people moderators, the perceived significance and value of the role will diminish.

Scale Matters: The Importance of Dividends, Supply, and Distribution

To create a liquid value associated with reputation, tokens should accumulate dividends for points holders, with each points holder receiving tokens based on the number of points they hold.

When designing such a system, there are three key questions:

1) Scale: How large should the dividends be?

The total scale of each dividend—that is, how many tokens are allocated each time dividends are issued—depends on the macroeconomic goals of the system.

Unlike points, it is important for tokens to be relatively scarce to give them value as currency. Many cryptocurrencies (like Bitcoin) benefit from long-term limits on token supply—only so many can be minted. In this case, the average total dividend must decrease over time unless there is some mechanism to absorb tokens back into the system (for example, through in-platform payments).

Even so—contrary to some standard intuitions—token supply does not necessarily have to be limited. For instance, if tokens can be redeemed for a portion of the platform's treasury, then the total token supply can expand as the treasury grows. In these cases, the total scale of dividends may remain the same or even grow over time, as long as the intervals between dividends are sufficiently spaced out so as not to exceed the growth of the treasury.

2) Supply: How often should dividends be distributed?

For platforms with participation functions like employment, such as performance or creator platforms, the best strategy is to regularly distribute tokens to points holders: for example, monthly or even daily. This means that users making valuable contributions, thereby maintaining a certain level of points, receive a steady income.

Infrequently or irregularly scheduled dividends are better suited for platforms where contributions are less regular, such as in certain DAOs. For example, Forefront issues its FF tokens when members contribute articles or participate in coding projects. Another option is to issue tokens only when platform engagement, productivity, or funding exceeds a threshold. This is similar to dividends for publicly traded companies, as seen in the WRITE token airdrop case on Mirror.

3) Distribution: How should the size of dividends relate to points?

Perhaps the simplest way to link tokens to points is through linear dividends, where each point entitles the user to an equal share of each dividend. But this is not the only option.

Under a convex dividend rate, each dividend's share is disproportionately allocated to users with more points. That is, the increase in dividends from 1 point to 2 points is greater than the increase from 0 to 1. Convex rates reward users who participate in the system over the long term, providing more motivation to maintain status.

The opposite approach is a concave dividend rate, where moving from 0 points to 1 point is more valuable than moving from 1 to 2 points. This approach is particularly effective in attracting new users, as it rewards initial contributions more than later contributions, regardless of whether you are the second, third, or 27th. However, if users are anonymous, such a concave system is harder to maintain, as in this case, users can create multiple accounts and earn early points through each account.

Financial Rewards Should Align with Contributions

The three dividend design features—size, supply, and distribution—determine the number of tokens points holders receive in each time period. This relationship directly impacts the motivation for earning points.

Ideally, the distribution of tokens should be calibrated so that the marginal return for users earning points corresponds to the marginal benefit generated by the platform's point-generating activities. How you view this marginal benefit depends on how contributions create value for the platform. In other words, you should earn points based on the value you create for the platform.

For example, on platforms like YouTube, points would reflect video view counts—the platform can accurately estimate the contribution of a given view count to platform engagement and the bottom line. The platform should then issue tokens based on the marginal value of each creator's views. Here, convex dividends may make sense, as top YouTubers drive more long-term engagement than less popular ones.

YouTube has undergone several iterations of this—initially linking payouts to view counts, then to watch time, and most recently, to an engagement metric they named "YouTube Time," which considers more than just the time spent watching content on the platform.

YouTube continually adjusts its payment amounts. This creates income uncertainty for YouTube creators who rely on ad revenue to pay their bills. Ideally, the platform would make the income contributions of participants more transparent so that contributors can assess the value of earning points (and avoid frustration and burnout).

Moreover, the principle of marginal benefit = marginal return means that point distribution rules may need to be adjusted. This allows the platform to provide disproportionate shares to early contributors, commensurate with their importance to the system, even if their contributions are fewer. In other words, the platform may not have a clear understanding of the value of contributions, especially in the early stages, so there needs to be a mechanism for experimenting and recalibrating around token distribution.

Reputation Incentives Must Be Ongoing

The connection between points and tokens provides a natural mechanism to reward users who make high-quality contributions. However, because points are non-transferable, there is a lag: those who accumulate points earlier may ultimately receive disproportionate dividends, especially in a scenario where the token supply is asymptotically fixed.

In many crypto projects, the largest holders end up being simply those who were the earliest to learn about the project. But those early adopters are not necessarily the most valuable for the future of the ecosystem. Therefore, it is crucial to maintain incentives for ongoing contributions and participation.

One natural way to achieve this within a dual-token framework is to allow points to degrade or depreciate over time. This can be accomplished by reducing dividends as a function of the user's points age. But a simpler implementation is simply to let users' total points decline, either mechanically over time or as a function of their participation relative to others.

This is again similar to what happens in games: in absolute leaderboards, points are zero-sum—if players do not maintain their contributions, they will ultimately lose their standing as other players surpass them. The same is true on creator platforms, where competition for consumer attention incentivizes ongoing participation.

But even if the points pool itself is growing, rather than being zero-sum, mechanically degrading points just to create participation incentives may also be valuable. Clearly specifying the rate and reasons for point degradation may also be valuable, as it can help users optimize their contribution levels.

Point degradation means that token dividends, like stock dividends, automatically dilute over time, just as traditional publicly traded companies dilute their equity by issuing new shares when they need new investment to increase value. In this sense, point degradation reflects the need for users to continue investing in the platform.

Of course, like any currency system, the actual value of tokens depends on the community's perception of that value. This means that the value of points largely depends on the perceived value of the tokens. Therefore, the community's perception of token value needs to be at least high enough to support the total distributed tokens.

This means the platform may need to adopt monetary policies—potentially adjusting the total token supply through buybacks or temporary trading restrictions. For example, in the initial weeks after launch, BitClout did not allow BitClout to be exchanged for other currencies.

The dual-token system we describe provides a straightforward alternative: adjusting the rate of point accumulation or the flow of dividends. For instance, the governing committee of a game could make the game slightly more difficult as a way to slow point accumulation, or they could reduce the total scale of dividends as a way to decrease long-term token supply.

However, builders must be cautious with these types of interventions, as any change to the overall incentive structure for contributors will affect their behavior. In particular, any unexpected actions that reduce the value of points could undermine user trust.

Do Not Over-Financialize

We have outlined some core principles of social token design, but it is equally important to complement these designs with product-market fit to inspire users' intrinsic motivation. For example, a game focused on making money for users but failing to capture the "play" element correctly misses the point: games should primarily be played for fun.

Products that bootstrap around reputation systems but lack true product-market fit risk creating a community of speculators rather than actual users.

However, once product-market fit is achieved, incentive mechanisms take over. If a platform cannot appropriately reward users, it will struggle to scale and achieve widespread adoption. To make incentives work, reputation systems should separate social capital from financial capital, especially if the former provides a clear pathway to the latter.

There are many other considerations, such as the relationship between governance and reputation; enabling the reputation system to respond to the evolving contributor community; and allowing all types of contributors to build reputation. But we believe that if builders adopt a dual-token system design at a high level, they will be able to reward contributors with genuine reputation signals while maintaining their value even as liquidity is generated. This is precisely what these projects need to drive growth.