Proof of Utility

Proof of work (PoW) and proof of stake (PoS) systems dominate crypto. But before blindly forking either for DoubleZero, we wanted to understand from first principles: which system delivers a dynamic and sustainable network?

Surprisingly, neither does — although in different ways. With their fixations on capital only, proof of stake systems have difficulties in unlocking value across both contribution type and time. Proof of work systems are better in this regard, but still imperfect. They suitably reward value across time, but they still struggle with respect to contribution type.

DoubleZero thus opts for a bespoke system, that is much closer to proof of work but still has a few key modifications. We call it Proof of Utility, as it best unlocks utility across both contribution type and time, and thus ensures the network remains dynamic long after its launch.

POSsification

Ossification — that is, the inability of systems to stay dynamic over time — is a powerfully corrosive force in crypto, despite occupying minimal mindshare compared to topics like centralization or censorship. It is rife in PoS systems, and stems from the solution to the cold start problem.

Every decentralized system needs to somehow compensate early enthusiasts in order to overcome the cold start problem. The answer is obvious in PoW and PoS systems alike: reward early participants generously. For instance, Bitcoin famously achieves this with halving events that exponentially reduce block rewards over time. PoS systems are similarly generous to early participants, although the decline of rewards over time is empirically less steep (e.g. Solana’s starting inflation rate of 8% is only roughly double today’s rate of 4.5%). This channel is massively amplified by the price appreciation of the underlying asset, such as BTC’s fifteen-year rise from a fraction of a penny to $100,000 or SOL’s five-year trajectory from a few cents to over $150, which makes early participation exceptionally lucrative conditional on selecting the right project. The combination mitigates the cold start problem.

But fast-forward a few cycles: the PoS system that solved the cold start problem now has an ossification problem. A cohort of early enthusiasts dominate the system, and latecomers struggle to break in. Through a combination of generous rewards and token appreciation, the initial capital contributions of the first participants represent the majority of stake. Professionals cannot get a meaningful toehold in such a system without mustering huge resources. In rough terms, $1MM invested in Solana’s seed round (between price appreciation and accumulated staking rewards) would represent 8% of the stake today. Today, you would need several billion dollars to reach the same allocation.

This is the ossification problem: a difficulty for PoS systems to handle maturation with respect to cohorts of contributors. These systems, where capital is both the input and output, get dominated by incumbents.

Economic Security is a Meme

To ardent PoS defenders, this criticism is irrelevant: a dollar is a dollar, regardless of whether it is provided by an early enthusiast or a latecomer, and it generates security for the system. Through their abilities to slash capital and impair future earnings, PoS systems impose immediate and transparent costs on attackers; and the larger the pool of aggregate capital underpinning the network, the more unviable these attacks become.

But this takes economic security too literally, and discounts all other forms of contribution type. For example, consider two validators in a PoS system powered by inflationary rewards: validator A has 10% of network stake and is running a mid-grade validator, while validator B has 0.1% of network stake and is running a top-of-the-line validator. It would be hard to argue that validator A is contributing 100x the value to the network as validator B. First, even if you believe economic security is not a meme — which these authors do — it is hard to argue that the value is linear. (For instance, is Ethereum half as secure today as it was in late 2021?) Second, their respective performances as validator operators are largely ignored by this system, as long as they can meet basic consensus obligations.

This illustrates the second general concern around a PoS system’s ability to deliver dynamic outcomes: the general inability to reward any contribution type beyond simple stake. PoS systems require a range of inputs, well beyond just capital, in order to be successful; but yet they cannot reward anything but capital within the protocol.

Such an ethos is most apparent at network genesis. The majority of a network’s tokens are pre-minted, and often sold in pre-launch rounds or subsequent private sales. This allows for strong participation from investors, while leaving the fate of developers and the community more uncertain. Both sets may receive reasonable token grants — but these are manual and discretionary channels outside of the formal economic mechanisms, and as such they do not scale well to independent developers or community members who contribute to the network’s success. Therefore, it is extremely difficult to join a PoS system at genesis without purchasing a substantial amount of tokens or being connected to the founding team. Networks that begin with high levels of inflation used to be able to minimize this, but since most networks now support the staking of investor tokens, these benefits are mitigated.

Admittedly, the situation is not as dire in practice as described here. Many PoS systems do offer mechanisms — namely, block-building rewards and delegated staking setups — to allow for some dynamism. High-quality validators do earn more than low-quality validators from constructing efficient blocks (although by relatively slim margins on most networks), allowing for some rewards to flow heterogeneously across contribution type. Moreover, delegated staking allows early enthusiasts to reallocate capital from the latter to the former, and thus allows latecomers to participate. But contracting and behavioral frictions are limiting factors here. A PoS system does not start with the premise of dynamism across contribution type or across time, and so these adjustments can only do so much to correct it.

PoW, Fresh but Narrow

PoW systems, best exemplified by Bitcoin, are in many ways more dynamic than PoS systems. Since capital is not both the input and output of the system, they inherently have a greater ability to welcome newcomers without necessarily falling prey to the cold start problem. But they still struggle to reward the diversity of contributions needed to be successful.

Of course, Bitcoin and other PoW systems still face the same cold start issue, and still solve it through a combination of token price appreciation and a declining rewards issuance curve. But critically, those financial rewards do not immediately translate into ongoing dominance. There are plenty of accidental Bitcoin millionaires who provided hashpower on their home machines in the early days; but they cannot meaningfully compete with professional miners today, so they simply enjoy the fruits of their early labors. The network continuously attracts and rewards professional operators without taking away value from early enthusiasts, and this open access prevents it from ossifying in the same way. Incumbency in a PoW network means very little.

But PoW systems still struggle with rewarding a range of contribution types. Using Bitcoin as the reference example again, hashpower is all that matters; and in expectation improvements in hashpower have the same linear relationship with rewards. Anything else that improves Bitcoin, e.g. improvements to its utility as a payments platform, are ignored by the protocol and only monetized in second-order and limited ways.

Moreover, the work that Bitcoin and other PoW systems require is costly and mostly wasteful, and indeed deliberately so. In a system that lacks staking and slashing, the costly wastefulness of the work is precisely what generates economic security. If the work was useful (e.g. modeling out protein folding), then the system would lose much of its underlying security as it could be used for multiple purposes.

But taken together, these collectively represent a missed opportunity. Contributors perform substantial work, but only a fraction of a single type is directly utilized by the protocol. This limits the system’s efficiency.

PoU: DoubleZero

DoubleZero is inspired by both PoS and PoW systems, in both positive and negative ways. But with an eye on utility above all else — both with respect to contribution types and contributions over time — how should it be designed?

The answer starts with first principles. On the DoubleZero network, the ultimate quotient of contribution is useful connectivity, i.e. connectivity that aligns with demand and that is meaningfully more performant than the public internet. This metric is directly plugged into the rewards model. The model then issues rewards for useful connectivity and — critically — in proportion to the incremental value that the connectivity generates, above and beyond the contributions of their competitors and the baseline of the public internet. This approach, known formally as a Shapley value methodology, is discussed and operationalized in a Github repo.

This mechanism rewards contributions both in breadth and over time. Above all, rather than contributors all fighting to perform the same task, novelty is implicitly rewarded. For instance, contributors who connect important but neglected centers earn especially handsome rewards, as do contributors who invest in new routes that shave latency off existing endpoints. Contributors implicitly space out from one another, aiming to add to the system rather than duplicate each other. Moreover, without any notion of preexisting stake, incumbency is irrelevant. Network contributors, regardless of tenure or prior rewards, must compete fresh in every epoch, judged only by the quality of their links and no other metric. This allows for continuous entry by new high-quality network contributors and an effective "exit" for existing contributors that don't continuously invest to stay at the cutting edge.

Indeed, this visually looks like a PoW mechanism. Those who do the most useful work for DoubleZero earn the largest rewards, irrespective of how many 2Z tokens their wallets hold. Holistically, this also keeps DoubleZero as competitive as possible with traditional permissioned systems. Such systems would ordinarily have a short-term advantage over permissionless ones like DoubleZero, because they can leverage identity, reputation, and legal liability rather than capital to establish trust. But in dispensing with traditional stake, DoubleZero stays in the short-term race, while also ideally being more robust, resilient, and extensive in the long run.

There is only one gap to patch. Unlike standard PoW systems, contributions in DoubleZero are actively useful, deliberately varied, and not wasteful, which leaves an attack vector with respect to economic security. That is, attackers can monetize their contributions outside of DoubleZero even in the event that an attack is unsuccessful, whereas contributions cannot be monetized outside of the network in traditional PoW systems. To solve this, rewards are deferred for some short number of epochs, which functions as implied stake — but without imposing direct capital constraints on network contributors, who might otherwise be liquidity constrained. In concrete terms, an operator’s rewards vest linearly over (say) ten epochs, and the protocol can seize back these unvested rewards if the operator’s behavior is shown to be malicious. Delayed payouts thus mimics the stake that validators post in PoS systems, but with the key exception that the collateral is earned rather than pre-posted — and so new entrants can join without needing to front large amounts of capital.

By breaking free of both the shackles of PoS, i.e. not taking economic security too seriously, and the shackles of PoW, i.e. work must necessarily be wasteful, the design space of decentralized systems suddenly opens up. DoubleZero’s PoU is evidence of this: it builds a bespoke system that modifies a PoW framework extensively. DoubleZero hopes to demonstrate innate dynamism across space and time, and so hopes to inspire other infrastructure projects to prioritize their own forms of utility too.