Non-mandatory flip

[hlolve]

Please review and discuss about this

https://github.com/hlolve/non-mandatoryflips/blob/main/README.md

[sekisanchi]

Please don't seriously get this comment, but "Whit influx" is not universal, and better to avoid using specific cultural assumptions. Please try using plain and easy description for fundamental mechanism. Looks similar to my own wish idea but not sure for the referenced previous IIPs. This is a first comment for appreciation, and will be back for the detail.

[hlolve]

Proposed changes in rewards affect IIP-7 and is optional to introduce a change in IIP-6. Anyway, the main change is validate in 2 groups, and these changes are only for modify rewards to incentive participants in group#1

[sekisanchi]

Splitting human makes people imagine racial/religious conflict, and the expertise on FLIPs does not rely on them.
So I prefer having expertise/profession like attribute upon current human category, and that's a natural consequence of my experience on this project, and also from protocol design perspectives.

However, defining characteristic upon the two groups, with or without that attribute is crucial on both protocol design validation and adaptation for the universal public.

[0JustinMiles0]

This proposal significantly reduces sybil resistance properties of Idena. An attacker who would be interested in farming validated identities for a third-party service such as Gitcoin would be able create more validated identities with less humans by just validating profiles in the group 2. If implemented, Idena wouldn't be a reliable sybil-resistant solution for third party apps anymore.

Moreover, farmers that are interested in farming identities to earn mining and validation rewards could use the group 2 to raise sybil identities to the status of Human in order to acquire full stake protection before moving those to group 1.

Removing IIP-7 seems also a risky choice as it would redirect a significant share of the rewards towards farming pools.

I would add that the motivations of the proposal are unclear. It seems that this proposal assumes that participants from third-party services create less qualitative flips which is unproven and that providing them the option to opt-out from flip making would increase the quality of the flips. Flips quality is highly correlated to the share of farm pools participation hence, by the mechanics mentioned above, I think this proposal would result in a decrease in flips quality.

My first question to author would be clarify what this proposal is attempting to fix/solve.

[hlolve]

Is not for new gitcoin hunters, just review Idena history, why network don´t growth ? New gitcoin integration just put this in more evidence. Idena rules are not for an average "not interested" person. Only few "nerds" will save Idena

[hlolve]

We need that "nerds" happy in the group#1 leave the mass in the group#2

[0JustinMiles0]

I'm against trading-off sybil resistance for more users. The virtual identity staking proposal seems to present better trade-off to grow our user base without sacrificing sybil resistance of the network.

[hlolve]

Is not to bring more users, is to adapt to reality, something Idena never did, and maybe that helps to users not lo leave the network. You will not have any sybil resistance to an AI attack with bad quality flips and reports.

[0JustinMiles0]

I already addressed your comment on flips quality in previous comments so unless new arguments are brought up, I'll stop here.

[simontakouridis]

Firstly, I love the concept of having an isolated group 2 of identities that only participate in solving flips. They piggy back on the consensus reached in group 1 and therefore do not influence the sybil-resistant properties of the network. HOWEVER, this proposal in its current form does not adequately separate and isolate group 2 and therefore reduces the network security. The following changes are recommended:

1. Identities should not be able to switch between group 1 and group 2, and visa versa. When joining Idena, a user must select whether they are in group 1 or group 2, and this will not change for the life of that identity.
2. Group 2 identities should only have the ability to invite group 2 identities. However, Group 1 identities can invite in either group 1 or Group 2.
3. The size of Group 2 in terms of numbers should be limited by proportionality to Group 1, e.g. If group 1 has 1,000 identities, then group 2 must not exceed 10,000 identities (by disallowing new group 2 members from joining the network). This is not about sybil resistance of the Idena Network, but instead maintaining integrity in the identity solution of Idena for 3rd parties e.g. Gitcoin Grants, where a designated "Idena human" is very likely to be an actual human.

If this proposal is altered as suggested, we could see mass adoption all whilst maintaining security. I look forward to see how this evolves.

[hlolve]

Group#2 can´t in any way affect network security, their mining capability is disabled. IDs in group#2 are ruled by IDs in group#1. Invites will not affect, but I considered increasing the possibility to group#1 receive majority of invites. Any ID invited (candidate) will go to group#2 and only if that ID wants to stake the minimum required will be able to select submitting flips during next epoch. Anyway I don´t see any problem to limit at all the capability for group#2 to invite.
I don´t see the problem of IDs switching groups for next epoch. An ID in group#2 can learn an decide to be part of group#1
Is important the proportion of IDs between groups, but is more important to reach a minimum number of IDs in group#1, and that number must be determined.

[simontakouridis]

My concern is that Group 2 has less sybil resistance than Group 1 - why? Because in aggregate, the same number of flips will be challenged to both groups, yet Group 2 will have many more identities and therefore a greater sharing of any given individual flip. As such, we cannot allow an identity to reach a given level of trust in the network through Group 2 and then switch over to Group 1.

Also, having a limit on the size of Group 2 in relation to Group 1 is also important so that sybil resistance, within Group 2 only, does not break down and 3rd party apps e.g. Gitcoin Grants, consider an Idena Human to be meaningless.

[hlolve]

If an ID from group#2 then goes to group#1, it will have the same job as any ID from group#1 (same as currently).
Group#2 might be less resistant to sybil only if we don't have the minimum number of flips in group#1 or a percentage ratio between 2 groups.
With validation sharding it is said that a group of 5000 validators is as secure as any other group of other 5000 IDs, we need to set that secure point between the minimum participants in group#1 and the top number in group#2 and considering the fragmentation model.

[hlolve]

Of course if we limit the capacity of group#2 we need to have a way for that, could be possible to suspend IDs of low score or low age to participate at lottery time. In that way is similar to current mode, and similar to group#1 if the ID forget to submit flips is not allowed to participate in ceremony.

[simontakouridis]

I don't think we need a hard limit on the capacity of group#2. Instead we could have a soft limit whereby if the capacity is reached (i.e. the proportional size of group#2 to group#1), no further identities are allowed to join group#2. This is much better for the user experience. Of course, the soft limit could be exceeded if the number of identities in group#1 shrinks. However, if we are conservative with the soft limit that is set, this shouldn't be an issue.

Or we could have both a soft-limit and a hard-limit for resiliency.

[ubiubi18]

Please review and discuss about this

https://github.com/hlolve/non-mandatoryflips/blob/main/README.md

I like the proposal and would love to see it happen. In the last 18 month i had my eyes less on the identity count but more on the mining node count. The later one is more or less stable, going a few dozens up and down. Rather down. I think those identities that run mining nodes should be more destinct from IDs that dont care about mining. But the community made a good point regarding the switch from group 2 to group 1. Maybe that should be reconsidered. What about a completely distinct identity status for mining humans, that requires identities to go to the whole process of newbie-> verified -> human again, if nonmandative flip identities want to switch to the group eligable for mining?

[hlolve]

But apparently farm pools receive enough rewards to keep them in business and flip rewards is a large share of their rewards. They will definitely stake the minimum required to keep earning flip rewards (which by the way will give them more voting power). So I don't think we can assume that this change would reduce farm pools participating in flip making.

Yes, because they submit flips that are not reported. I am not against pools if they submit good flips

[0JustinMiles0]

The main problem is that new candidates are forced to learn a lot of rules. For they is easy to pass the flips test but not reporting or worse submit flip that later will be reported. IDs that only wants to prove humanness don´t need to be forced for all of this

In this case, maybe a simpler solution is to simplify the rules. I've proposed a simplified framework but I'm sure there are other that could be implemented:

• Approve flips that are highly or moderately interesingly unique
• Abstain for flips that are slightly interestingly unique
• Report flips that are not interestingly unique (even if they contain the two required keywords)
• Report flips that don't contain the two required keywords regardless of how interestingly unique they are
• Report flips with offensive/NSFW content regardless of how interestingly unique they are

[0JustinMiles0]

But apparently farm pools receive enough rewards to keep them in business and flip rewards is a large share of their rewards. They will definitely stake the minimum required to keep earning flip rewards (which by the way will give them more voting power). So I don't think we can assume that this change would reduce farm pools participating in flip making.

Yes, because they submit flips that are not reported. I am not against pools if they submit good flips

But again, I don't see how this system will disincentivize farm pools from submitting flips. All, they need to do is to stake to the minimum required amount.

[hlolve]

I am perfect with current rules, buy I am a "nerd" and maybe there are other 300 like me that are happy with these rules, but if we don't see the reality of people leaving the network or the improvement in AI, we will not be happy in a near future

[0JustinMiles0]

It's not that I'm not seeing the current problems but that I'm not convinced that this proposal will fix them without introducing other similar problems.

[ubiubi18]

@midenaio is it fruitful if the community still discusses IIPs, or do you have no capacity currently and we should wait for better times?

[sekisanchi]

@ubiubi18 copy to @midenaio Hi, These are a bit behind from a "design proposal", which leads into a "design change" We do have both user and designer roles here, but not clear where in the protocol have to be changed within the current protocol model consistency.
On the other hand, in theory, any person or party can "fork" to start new idena protocol as seen in bitcoin and ethereum.
Considering the scale of changes from small work around fix to hard fork, impact of this requirement is sitting at midium or midium-high,
and thus it requires a bit specific as an IIP which includes both protocol change detail and the impact to the current model.
I have not seen that within above chronologies yet.

[morienzo]

Thank you, Hlove, for starting the discussion. Here are a few elements I would like to propose:

• The number of flips that one can create is determined by age or stake:

Age below 25 or stake below 25K IDNA: 5 Flips
Age between 25 and 50 or stake between 25K and 50K IDNA: 7 Flips
Age between 50 and 75 or stake between 50K and 75K IDNA: 9 Flips
...

• 1 reported flip results in rewards being halved. 2 reported flips result in 0 rewards.

• To achieve human rank, one must be a flip creator (and not have 2 reported flips).

• All addresses with 2 reported flips or that do not create flips will have a "Verified" rank.

• Only "Verified" or "Human" addresses can create flips.

• Only Humans can earn mining rewards during the next epoch.

• A minimum of 3 flips is required for all creators.

• All flips beyond 3 receive quadratic rewards.

• Keywords are distributed dynamically: 9 pairs of keywords initially, and then a new pair is received for each flip created (if a flip is deleted, no new pair is received).

[hlolve]

The top 60k threshold idea is to avoid the risk of abuse by high stakers and allow more incentives for lower stakers.
With packs the risk to submit extra flips will be split in each pack.
I like the idea to allow packs by stake and not by type of identity, but I will allow to any validated ID to submit flips.
The restriction will be by stake, and newbie and verified don´t have stake protected.

[morienzo]

True, the lack of stake protection is a hindrance in itself, so the minimum rank is not necessary (everyone is free to take the risk). Regarding the 60K limit, I find the ambition to share rewards fairly is a good thing, but the method bothers me. It is logical for a quadratic reward that the person with a high stake is better rewarded. These quadratic rewards only concern a portion of the flips that will be made and should not be limited. However, I would support opening a discussion on the distribution of rewards in case of network modification, but I think that should be done later, through another IIP. There are several solutions to consider for a fairer distribution:

• Modify the number of extra flips possible per pack
• Modify the parameter K of the stake power
• Modify the percentages attributed to each type of flip

Passing this IIP without modifying the distribution would also allow us to gather statistics on the rewards. Today, all of this is hypothetical.

[hlolve]

The upper stake limit can be removed. We can see first how "flip packs" works and after modify rewards incentives if required.

[sekisanchi]

Sorry to cut into the middle. I agree with the direction of number of flips with staking incentives, but have a large concern on heavily biased flips dominance. With current flip slashing rules and frequent late submission errors under the noise induced FLIPs, it will increase first onboarding hurdle, and thus dis-incetivize independent or non organized brand new FLIP creators to participate.
I want some plan to improve heavily biased, complex and ugly FLIPS as a pre-req.
Attached below is the best FLIP I've ever seen, but this type of FLIPs or creator will may be defeated by strange biassed mass production. In another word, I want to prioritize the fundamental health surgery than incentives.
Snake/Cane https://scan.idena.io/flip/bafkreifxwsimn5rels5thqmmht6qubplzvbbgx5gv77ldsqk2tfpmr56be

[hlolve]

Late submission error is a problem from users that don´t know how to participate in the ceremony.
I and my family/friends never had a late submission, using desktop app or web app in more than 80 ceremonies.
If we allow people that only want to prove is "human" and not interested in Idena, we are exposed to bad flips.
Flips production and reporting is based in "rewards", if someone participating don't care of that rewards you need to isolate that people, but is possible to give them their "human" status only solving flips.
That is the purpose if this proposal.