On Monday, October 15, Civil's token sale came to an end without having reached its soft cap of $8 million. Undeterred, the company is moving forward and plans to hold another, more accessible token sale.
The Civil platform is chock-full of big names and impressive resumes at every level. It's backed by ConsenSys; it boasts collaboration with AP and Forbes; its newsrooms are run by journalists and editors from well-respected outlets and institutions; and its Civil Council is stacked with industry veterans. There's no question the platform has a lot of backing from people who matter.
Perhaps this time around the sale will be more successful, but the company will need more than a successful token sale and a qualified team of journalists to accomplish its mission to create a new economy for ethical journalism. Once people buy the tokens, there has to be a solid incentive structure to promote the desired outcomes. In other words, Civil needs a solid cryptoeconomic model.
The Cryptoeconomics of Civil's TCR
Cryptoeconomics – a term used to describe the study of economic incentive models in blockchain systems – is a relatively new discipline in a state of rapid flux due to ongoing technological evolution, experimental learning through trial and error, and theoretical political/philosophical development. One blockchain-specific economic incentive model that falls under this area of study is a token-curated registry (TCR).
The Civil platform hosts newsrooms using a TCR called the Civil Registry. Despite the nascence of the discipline, there's been a good deal of commentary on TCRs, and I think the viability of Civil's TCR is worth considering, even if the frameworks for judgment are relatively untested.
According to Yin Wu, founder of the DIRT protocol, there have only been two TCRs launched on the Ethereum blockchain so far: adChain and FOAM. That is to say, Civil doesn't have much data to build its model around – only theory and gut instinct.
The Civil Registry
If a newsroom wants to join Civil, it submits an application and deposits a set amount of CVL tokens, right now set at the equivalent of $1,000. There's then a review period for token holders to consider whether the newsroom is in alignment with the Civil constitution. If there are no concerns, the newsroom is added to the registry.
If at any point a CVL token holder sees that the newsroom is not abiding by the constitution, they can stake a deposit equal to the newsroom's application deposit. The deposit of the newsroom and the deposit of the challenger are then pooled together and eventually act as the "prize" for the winning side.
Once a challenger and a newsroom are staked against each other, a vote opens to CVL holders. Token holders vote whether the newsroom has violated the constitution. One token stands in for one vote, and CVL holders can vote however many times they want and can afford to. The tokens used to vote, however, are not spent – they are only staked until the end of the voting period, at which point the tokens will be returned to the voter. At the end of the voting period, all votes are counted.
The deposited stake of the losing side (the challenger or newsroom's deposit, not the voters' tokens) is halved: One half is distributed among winning voters in accordance to the number of votes they cast on the winning side, while the other half is given to either the challenger or the newsroom (depending on who won). All the tokens used for voting are returned to the voters, regardless of whether they won.
The idea is that this should incentivize voter participation and newsroom honesty. However, it remains unclear if the equivalent of $1,000 is an appropriate deposit – if it's low enough that a challenger would risk losing the deposit if voters disagreed with them, and if it's a high enough reward that token holders would care to spend the gas, time, and mental energy to vote. After all, $500 distributed across a whole voting base couldn't add up to much, unless voting participation were terribly low, which is also obviously not an ideal situation. It's also unclear if the deposit is a sufficiently high amount to dissuade a newsroom from dishonesty. However, if they are proven to be operating outside of the constitution, they also lose their spot on the Civil platform and their reputation takes a hit, which are probably more important factors.
Another issue with TCRs is that the "one token, one vote" model privileges those with more tokens. Of course, the tokens used to bet are not spent (only lent), so it's not necessarily about what one can afford to lose in a bet, just how many tokens a person or entity possesses. This scheme means that it isn't necessarily who cares more or who is correct in a vote; it's more about who has the most tokens to participate. This model seems flawed.
TCRs fall under the "still working on it" category of cryptoeconomics. A lot of projects seem to be utilizing them, and a lot of smart people with more expertise than me have expressed enthusiasm about their potential. But others have expressed skepticism.
In an April Medium post, software engineer and CoinFund co-founder Aleksandr Bulkin asserts that TCRs "don't work." His assertion is rooted in an argument about the conditions under which oracles best function. Oracles are more or less the same thing as TCRs, he argues. It's an incentivized game to take information from "the real world" and verify the truthfulness of that information on-chain through token holder voting. The model he uses as an example of an oracle is only slightly different than Civil, namely in that he assumes that token holders actually spend money to vote. Under Bulkin's model, each vote incentivizes token holders with a dissenting opinion to contribute: The pot is not fixed but ever-increasing based on (and incentivizing) voter participation. Civil chose its model out of concern that fear of losing would dissuade potential voters, which seems to be an equally valid concern.
Bulkin argues that the best subjects for oracles are things that two honest people are unlikely to disagree about: location, weather, and dates. The first flaw with TCRs is that they profess to find "truth" about things that two reasonable and honest people could disagree about.
And this is one of the major potential issues with Civil's TCR model. Even in "perfect" conditions, assuming the deposit price is appropriate and that token holders vote exactly how much they care – and ignoring the problem of wealth inequality – who wins is still not necessarily the "correct" party because this incentive structure incentivizes token holders to vote for who they think others will pick, not necessarily who they think should win.
Slava Balasanov of Relevant, a token-curated reputation platform, wrote a post on Medium back in July about this flaw common among TCRs. He argued that TCRs are not great at answering subjective questions. His example was a list of "the best" restaurants. He said, "The question boils down to who are the people curating information are and why are they doing it." Depending on what a token holder thinks about the other token holders, they will choose to curate the registry differently. For example, Balasanov points out that if the token holder believes that most other token holders are food connoisseurs, they might curate the registry differently than if they thought that the token holders were people who didn't really care about food and just wanted to earn some tokens.
Assuming that a CVL holder believes their values and judgments align with the majority of token holders seems a bit of a stretch. However, Balasanov points to a potential solution. He argues that for such subjective lists to be effective, there must be a "general alignment of subjective social values." He says this in the context of discussing the potential benefit of combining TCRs with social reputation systems, but the point applies to Civil as well. Civil does not yet have a social reputation system (though it intends to at some point), but it does have a guiding constitution that is intended to stand as a unified statement of shared subjective social values. Ideally, a CVL holder would have a relatively high level of confidence about the metrics others were using to vote. That might be enough. To be fair, though, it's not unheard of for people to disagree about the interpretation of a constitution.
Even if CVL holders vote for who they think should win – and they generally agree on shared values – it's nice to think that the majority will get something right (in an ethical sense). However, there are plenty of historical and contemporaneous examples that show this assumption is also something of a leap. In short, it's hard to see how a TCR incentivizes ethical journalism so much as it incentivizes newsrooms who tell stories that align with the community's values. It's a potential cesspool for confirmation bias.