Automate Away the ATF


A simple cryptoeconomic game could distribute the ‘track proposals and enforce accountability’ function, thereby removing the need for an Accountability Task Force and related budget, comms channels, spreadsheets, etc.

Any Genesis member (rep holders) or observer (such as predictors) submits a proposal to simultaneously:

  1. slash the reputation of the specified offender
  2. reward herself

This creates an ongoing incentive for anyone to hunt for undelivered proposals. If the reporter is wrong, then that evidence should come to light during the voting period and the DAO will reject the proposal.

While this is already technically possible at the protocol level, it would require minor modifications to Alchemy Earth’s Contribution Reward Queue views (the U.I. level):

  • “New Proposal” pop-up should allow a user to specify multiple addresses with different reward types and amounts.
  • Proposal Card should display the reward types and amounts for each address separately.


This post is a follow-up to a comment by @Stratis during the break-out portion of today’s Genesis call about finding simple ways to distribute the ATF’s role across the DAO.


Caveats: I still think it would be useful to have…

  • an Accountability working group to get this sort of game going (encourage DAOstack to add this U.I. feature, submit the first few example proposals, …)
  • Paid Defense Officer(s) to monitor the DAO for attacks and iterate on defense protocols.

For additional context, here is a description of the ATF’s responsibilities.


I want to note that the above example is “proposal multi-targeting,” which is something I’ve discussed with @dragonfly at length. That is, being able to target more than one address to assign funds and Rep (contribution rewards).


I think a combination of Adjudication and Escrow contracts could do the trick nicely for ATF.

Note that Multi-Target has a lot of intersection with Escrow. They can easily be combined or chained.

There are more than one scenarios here, such as where certain rewards, such as reputation, are not distributed until the adjudication has completed favorably (Curacao). Or like Genesis, where adjudication happens after the rewards have been distributed.

I’ve been thinking about this in the context of Curacao…



Maybe there is an interim solution until we have this functionality available?

-maybe this can be achieved via two proposals
-maybe rep is either given to Undelivered Proposal proposer or taken from initial proposal proposer

It would be interesting to run this in parallel with the current ATF process and see if this is effective.


I`m totally aligned with this idea of involving community into accountability issues.

My suggestion on this regard differed only in part, namely, setting an escrow to reward active community members.

Why escrow? To put a group of ATF members in charge of the process, because this process won`t run autonomously:

(1) recent low voting activity gives all grounds to forecast low accountability activity;
(2) autonomous running of this process will not guarantee implication of standard rules and practices – new DAO members ‘arrive’ every day and they yet don`t share same experiences as existing members do.

Plus we got no written rules, no place to put them and no enforcement engine to make people follow them. The only law is – the current vote, which never guarantees the next day vote gonna be on the same base.

What written rules on this matter? For example: (a) what is the way (if any) of preliminary notification of a failed proposer and receiving his explanations, (b) what is the sum of reward for this activity - should it be fixed (in GEN and REP) or proportional to the sum of undelivered proposal and undeservedly acquired reputation, © etc…

To conclude I should say that this is a very promising idea. To let it grow up to an autonomous system, I suggest putting it under ATF temporary supervision.

1 Like

Great conversation, I’m really happy to see all these ideas about how we can automate this process more (which I don’t know how to do!).

I’ve just started a new thread with a related topic, about how we would like to flag bad behavior in the GenDAO, because this has come up recently in ATF conversations: How do we want to flag negative behavior in Genesis?

One key question there is whether the element of flagging can be automated as well (and whether that makes sense or might cause even more harm).

I would love to hear your thoughts over there on this topic!

1 Like

I’ve been living with someone this week (and next) from the regen network. They basically are supporting project validation, but exclusively for regenerative projects. But it is an interesting “pluggable” framework they have, so I plan to discuss with her how/whether it can be extended to any type of project, and be able to have an automated interface with our DAOs. A particular challenge is that it is not on ethereum (Cosmos), so interoperability is a question…