By Jan Horlings, based on discussions with the DeepFunding Focus Group
Dear community,
We are sure that you are looking forward to the outcomes of DFR3!
While analyzing the results, we identified that there was activity from bots upvoting and downvoting content, and users applying generated content that did not contribute any intrinsic value. It seems clear that some of this activity was aimed at the sole benefit of the creators of the content, not to add any value to the ecosystem or support projects that deserve it.
While this is not too surprising in the current state of AI, and the youth of Deep Funding, it does open the question of what to do with this, if anything. Last Thursday the Focus Group (FG) organized a ‘listening’ meeting where we gave you, the community the opportunity to give your feedback. The second part of the meeting was dedicated to the specific problem of bots and generated content. As promised, we have contemplated all feedback, including the notes left on the Miro Board.
There was no simple decision in view, but it is clear that many people would like to see the bots filtered out somehow, and also most reactions proposed some kind of change to the rules of DFR3. We realize, however, that this may also be attributed to the questions asked and the eagerness of the community to help.
After another in-depth discussion in the FG, we have arrived at the following advice for the Deep Funding team.
1. We advise to go ahead with the payout of the defined amount of engagement rewards – as planned.
The reason for this is that we don’t think we are able to identify with 100% certainty who was engaged in automated behavior. Moreover, the use of generated content is not necessarily a bad thing; it depends on the motivation of the user (just gaining rewards; improving one’s text; helping research; etc). Therefore we accept that some people who were gaming the system will reap the benefit of their gaming activities – at least for this round, until we come up with a better defense.
2. We advise to use the added weights in voting – as planned
The main reason for this is that a first analysis (done by SNET) indicates that the most ‘suspicious’ wallets have not participated in voting (or not with any AGIX). We have been told that there have been votes cast by 182 ‘collection ID’s’. In total 37 wallets could be identified during the voting event as ‘contributors’, and 24 of them had a voting weight larger than 1. This is a substantial amount and we believe there is not enough reason to keep this impact from the results.
3. We would like to give the best contributors an extra bonus reward
Although the ‘bots’ are compensated (which is not desired), a worse outcome is that their revenue is taken from authentic and engaged community members. We deeply regret this, and would like to avoid that these enthusiastic individuals will turn their back on the program due to the misbehavior of a few. Therefore we have discussed with the DF team to have the most outstanding individual commenters compensated with a small bonus on top of the regular rewards. We don’t intend to filter out all humans, or make complex calculations, but simply and manually identify the top ‘x-number-of’ contributors and give them this bonus. Note that this does not say anything about all other contributors, except that in the eyes of the assessors they did not stand out as much as the ones that got rewarded.
The Focus Group will take a bit of time to determine who we think is best positioned to make this assessment.
4. Top contributors to be involved in peer review in future funding rounds
Finally, we and the DF team see a role for this group of top contributors to strengthen the current ‘peer review team’ to help them in the recommendation process in the next rounds. But this is still open for discussion and further ideation, to be done in the coming weeks and months.
No inside information
We would like to emphasize that we, as a Focus Group, have not seen any results and have not based our advice on any preferred outcomes. We simply have no idea which projects will or will not benefit from this advice (compared to alternative scenarios).
Seeking your consent
If you are strongly opposed to this advice, the Focus Group would love to hear from you. Based on your insights we may or may not change our views and advice, but in any case, we will answer your concerns.
We do however require 2 things from you:
- You will give clear and compelling arguments on why the advice is not sound
- You will explain what would need to change in order for our advice to secure your consent.
We hope to learn from you, and are not afraid to change our views should new and important arguments or insights surface. If, however, our final assessment is that the advice should remain unchanged, we expect the same respect from you.
Please use this Google form to give or withhold your consent BEFORE Thursday 19 Oct at 20:30 UTC (the end of the DeepFunding Town Hall), so in case there are no substantial complaints we can still have the voting results published on Friday the 20th as promised.
We hope the final outcome is the best it can be under the current circumstances.
But more importantly, we hope that together we will learn from this, and use this opportunity to improve our voting and selection processes.
Finally, we sincerely hope that our handling of the situation as a Focus Group is deemed useful and appropriate and this ‘consent’ approach will help our community to navigate such issues in the future.
Best regards,
Jan Horlings, on behalf of the Focus Group.
(Christian, Okan, Tommy, Ubio, Vani)
Post Views: 832