Algorithmic PoS rewards may be a bad idea

Daniil Gorbatenko
Door to Crypto
Published in
7 min readJul 30, 2020

Many people in the crypto space believe that decentralized finance, or DeFi is the industry’s best candidate for the killer app at the moment. Indeed, the amount of cryptocurrency locked in DeFi has quickly skyrocketed to almost $3.5 billion from just $1 billion a year ago. DeFi-generated transaction activity has even succeeded in bringing Ethereum back to the congested state in which it last found itself two years ago.

However, when one looks beneath the surface of the DeFi boom, one will notice that what is actually driving the bonanza is the particular form of staking first dreamt up by the team behind the Compound protocol. In this arrangement, both lenders and borrowers are rewarded in the protocol’s governance tokens for participating (especially, locking crypto within the system). Its result has been a peculiar situation in which the bulk of the DeFi transaction activity is aimed exclusively at generating more governance tokens, often from several protocols simultaneously. The sophisticated protocol reward exploitation activity received a colorful name “yield farming.”

While this phenomenon has attracted a lot of attention to DeFi lately, it is unclear whether it is sustainable or beneficial for the long-term development of the sector. Codefi has published a report suggesting that yield farming mostly just involves those who were already active in DeFi before, so no significant additional adoption is taking place thanks to it. The amounts and complex leverage arrangements involved create risks of massive failures that could taint the reputation of DeFi for years to come and attract the sort of unwelcome regulatory attention that the ICO boom drew.

A larger potential problem

However, what is much more interesting is that the DeFi’s yield farming phenomenon is just an instance of a larger potential problem of using algorithmic staking rewards in crypto projects.

Consider that a recent study found that 76% of the whole transaction activity on the Tezos platform revolves around the participation in its consensus mechanism (a delegated version of PoS not to be confused with EOS’s block producer election system called DPoS). According to the data from, PoS platforms like Tezos, Cosmos, Algorand and Harmony have 60–80% of all the eligible native tokens staked. Zilliqa, which does not even rely on staking in its consensus mechanism (a hybrid of PoW and PBFT), has managed to create a lot of buzz around itself by introducing staking for seed node operators. The Staking Rewards website does not yet have a locked token percentage for Zilliqa, yet, but it is almost certainly quite high.

Even though in the short run, algorithmically rewarded staking is beneficial for the current tokenholders as they can receive a form of quasi-fixed income with their crypto, is it really the main point of the native tokens of PoS blockchains to be locked in the artificial loop of staking to protect the integrity of staking?

In other words, algorithmic staking rewards seem to inevitably create high volumes of rather artificial activity whether they are used in DeFi, consensus mechanisms or other contexts. What is the fundamental reason for this?

Algorithmic staking rewards are economically arbitrary

Some people could just say that staking fundamentally resembles a loan. Stakers, in a sense, lend their crypto to the whole community of the platform users, and the community remunerates them with interest for the service they provide.

The problem with this is that in traditional lending, when a loan is made whether it is going to be repaid with interest depends on the capacity of the borrower to independently create value that is determined in the market. In the case of algorithmic PoS rewards, the interest paid is unrelated to the creation of value. Even if a blockchain validates transactions containing what is essentially worthless spam, the staking rewards will be the same as with any other transactions that could be validated, at least in terms of the blockchain’s native token.

Algorithmic staking rewards may be arbitrary, so what?

One could still object here that it is unclear what the major problem with staking is. Yes, algorithmic staking rewards are sort of unrelated to the value provided by staking-based platforms but so what? For instance, yes, most of the activity on Tezos relates to staking but what is the harm? Put differently, is there any evidence that too much focus on staking hampers the wider adoption of the staking-based platforms?

Although answering this question conclusively is almost impossible because it involves unobserved counterfactuals, there are at least four major reasons to believe that excess focus on staking is damaging.

First, the presence of algorithmic staking rewards will tend to heighten the opportunity cost of genuine adoption of the stakeable cryptocurrencies. For instance, someone who is considering whether to adopt, say, XTZ for making payments will need to wonder not just whether enough of her counterparties will do the same, whether the fees and other peculiarities of using crypto are worth it, whether the XTZ price is not going to be too volatile in the future, etc. She will also have to wonder whether just staking XTZ is a better idea.

Secondly, accommodating widely used staking arrangements requires effort, attention and resources that could be allocated to something more useful, including for the development and adoption of the relevant platforms. Staking-as-a-service platforms need to be set up, for instance, or in DeFi protocols, efforts need to be made to prevent abuse and to protect the sophisticated leverage schemes from implosion.

Crypto projects also leave or die by their user communities. Hence, it probably matters a great deal what kind of users they attract, especially early on. In sectors like DeFi, too much focus on staking may reorient certain project communities from helping create a much needed complement to the rigid existing financial system to dubious and unsustainable fads like yield farming. Blockchain communities may disproportionately attract people interested in fixed-income alternatives to the existing ones that are impaired by artificially low interest rates. The kind users crypto platforms will attract will also probably play an important role in shaping the perception of what the point of the relevant platforms is in the eyes of the broader public, investors, innovators, opinion leaders, etc.

Last but not least, there is also the issue of inflation. Algorithmic staking rewards by definition continuously increase the quantity of the network’s native currency in circulation, thus, reducing its value compared to what it otherwise would be. There is probably a reason why Bitcoin’s native currency BTC still retains overwhelming dominance in terms of market capitalization: its inflation rate is set to fall with time, eventually reaching almost zero.

Now, one could claim that inflation in crypto is different from its fiat-world counterpart. As future staking rewards are preprogrammed, it becomes much more predictable than the inflation generated by central banks. However, it still erodes the value of the stakeable tokens, undermining their attractiveness.

Why the situation may be different with Proof of Work

One could certainly object that there does not seem to be that much of a difference between algorithmic rewards in PoS and PoW. In both cases, the number of native currency units the miner publishing a new block earns is determined by fiat rather than by the value her mining services creates for those who benefit from them.

This objection is fair but there is an important feature of PoW systems which is absent in the PoS ones. In the former, it is very difficult for the communities to defend themselves should the network be taken over by an attacker. They can of course hard-fork the network and change the mining algorithm but chances are that the attacker will quickly manage to fine-tune the mining rigs to be able to mine on the new chain, too. This is where it may be useful to have mining rewards that are much higher than the average transaction fees in order to incentivize miners to behave themselves.

In PoS, however, a long-term takeover is much less likely because the stake of the attacker may be rendered useless by merely deleting it on the new chain. And this is not merely a theoretical argument. Recently, the old Steem community has demonstrated the power of this approach by managing to successfully fend off Justin Sun’s takeover of the old Steem through creating Hive. Although Hive has not yet conclusively overtaken the Justin Sun-controlled Steem in terms of the market cap, the community has even managed to port the most-used Steem-protocol DApp Splinterlands over to Hive, preserving the user count in the process.

Can consensus work without algorithmic rewards?

The key argument against eliminating algorithmic staking rewards on public blockchains has to do with the uncertainty about whether relying on transaction fees alone will suffice to ensure the desirable properties of blockchain consensus. In other words, one could say that years of experience demonstrate that algorithmically set block rewards allow mostly smooth functioning. In terms of both safety and liveness. Whereas there has not yet been a fully transaction-fees-based public blockchain.

This objection is plausible, indeed, but I believe that there are factors mitigating it. First, as I mentioned in the preceding section, PoS-based platforms have a powerful weapon against attackers — they may render their tokens (next to) useless through hard forks. Secondly, it is quite possible to test whether this objection is valid in a relatively controlled manner. Project teams behind PoS blockchain platforms can start by running the majority of the validators and look at whether they can bootstrap a sufficient number of validators if the block rewards only consist of the transaction fees. In any case, an argument can be made that if the transaction activity on the relevant blockchain is insufficient to maintain high validator interest, the platform in question will probably not attract a lot of attackers.



Daniil Gorbatenko
Door to Crypto

PhD, economics (2018) from Aix-Marseille University, independent blockchain adoption consultant based in Aix-en-Provence, France, Email: