The Contentious Genesis Parameter, delta

With the launch of Alchemy Earth, users will now start using the Genesis .2 protocol. Here are the final parameters of the Genesis DAO, as selected by the DAOstack team:

As may be known, DAOstack’s BizDev team has been working together with other organizations to define their parameters for their DAOs. A common insight in this BizDev that has come up repeatedly is contention regarding the delta parameter, that is:

Early voters reputation loss percentage (%)

Now, it’s important to remember that there is nothing inherently good or bad in the Genesis parameters. They only exist as defining the design space we have for DAOs within the holographic consensus framework. However, there are different arguments for why the parameter itself is useful or not.

Pro-parameter argument:

Last month, the DAOstack team, specifically @Ezra published a post commenting on this parameter:

Voters on non-boosted proposals risk losing some of their Reputation if their vote goes against the eventual majority decision. This parameter, targeted at votes made early in a proposal’s life cycle, drives a DAO to be more coherent. A more coherent DAO produces more consistent and predictable decisions, which can be valuable for productivity and building organizational trust. Incentivizing coherence can also discourage the voicing of minority opinions, however, so DAOs will want to consider this parameter carefully.

Similarly, voters on non-boosted proposals receive a Reputation reward if their vote goes with the eventual majority. This reward is a redistribution of the Reputation lost by minority voters on non-boosted proposals, and it is the mirror image of the previous mechanism, existing to promote coherence and consensus within DAOs.

To me, this coherence is best represented by an analogy to lubricant: coherence is the grease that keeps the DAO generally aligned, mitigating the risk of forking, but at the cost of potentially discouraging the voicing of minority opinions.

Anti-parameter argument:

The crux of the anti-parameter argument, that is, setting the parameter to 0, is more or less thus:

If people get rewarded for voting in alignment with the outcome, why won’t everyone just vote the popular opinion rather than their own opinion?

At first glance, this seems like an optimal strategy for voters, but there’s more nuance here. Ultimately what this boils down to is a very specific question: is the expected cost of incoherence greater than the risk that the disincentivization of voters (-4% of their individual Rep) will cause them to not vote their heart?

While there’s no clear answer here, we can articulate what these costs may look like. When incoherence becomes dominant, the DAO risks forking, which results in:

  • Diminished treasury – I imagine that there will be schemes in the future that allocate a percentage of a DAO’s treasury to its fork, in order to protect minorities. In the case of the DAO having its own token, the treasury is diminished through an immediate doubling of the token: it and its forked version (ETH vs. ETC, BTC vs. BCH, etc).

  • Diminished reputation – An organization that splits is generally considered less trustworthy.

  • Increased likelihood of forking again – Forks tend to beget forks, as it becomes ideologically ok to do so. This is orthogonal to the principle of the network effect, which becomes more defensible the larger it gets.

  • Reduced supply of attention – Contentious forks often result in spirited debates, which reallocates attention away from the DAO’s mission and towards the partition itself. One item to note is that even without forking, when there are contentious proposals, the DAO’s attention is being spent on these proposals; increased incoherence results in higher attention spend.

In my opinion, after evaluating these costs, it seems to me that forking is an extreme high cost activity that generally should be avoided. 4% is a very small amount of a grease to help avoid incoherence. It’s possible, even, that delta should be set higher.


Good points Pat.

There’s still a big doubt that I have in regards to what you said in the end:

4% may be a small amount but we can’t really predict how it will affect the balance when there is a very large number of reputation holders. It smells like a winner take all effect could emerge by virtue of the 4% compounding.

If I’m not wrong, even if you voted incorrectly 5 times in a row, you’d still have 81.4% of your original reputation. I don’t think the compounding is that much of an issue, as it becomes a relatively smaller deduction each round.

The issue is that one man’s reputation % loss is another man’s reputation % gain.

When the “set of incorrect voters” loses reputation, all other reputation holders are essentially rewarded with reputation (even if they hadn’t voted at all)