Some suggestion for finetuning the Certification parameters

Archives Discussion
1

Hi all,
I think to choose the right certification parameters is one of the most tricky things in UCoin, so here are some ideas to that:
A too low parameter of allowed cetifications per time would hinder growth, especially in the beginning
A too high parameter would encourage creation of double accounts.

To start with two assumptions:

  • growth will happen most likely in bubbles, as awareness comes and gos normally
  • if ucoin succeed average growth of ucoin is exponential

that means sometime we need more certifications if there is a bubble and new users come in sometimes we need less.

  • lets say we allow only one certification at time frame X (to get some idea lets say one week)

that means with 3 required certifications per member we could grow round about double if 3/4 of the members take part in certifying new people.

normally communities in the longer run tend to have at round 10% active members if there is no burst time.
so we could grow round about 10/3 = 3% per month

the question is now how we could allow to have more certifications in case of bubbles and at the same time do not much encourage creation of double accounts

A possible solution would be to give some economic incentive for that:

Lets say one certification cost the Universal Dividend for one week
Lets say 1 certification per week per member is free
Lets say max 1 certification per day is allowed

that means in case of need members could start to certify more new members
but still we would have en economic incentive to be honest, because if the certified member was not able to join the web of trust, through for example not fulfilling the distance rule, the used coins would be gone.
Because all transactions are on the blockchain is would also be transparent from where the coins for the certification came.

Second question:
What do we do with the fees?
One solution would be to just destroy them.
Another solution would be to give the fees to the miners.

I think to give the fees to the miners would be the best solution, because under the condition that the mining is fair for all members, the coins would be redistributed to all members who take part in securing ucoin.

Secondly we should think about allowing the undoing / revoking of certifications.
The same as with certifications, they should be limited per time, so that certifications are not canceled easily

so lets say we choose the same parameters:
Lets say one revoking of a certification cost the Universal Dividend for one week
Lets say one revoking of a certification per week per member is free

This would allow to cancel to certifications to known fraudsters, before they could do more harm to the web of trust.

Further we should consider to change the distance rule for new members.

One rule could look like that:
**- new members are defined as members who have up to now never renewed their member ship **
- new members are excluded for the consideration of the distance rule check of members who renew their membership

What are your thoughts?

UPDATE:

I think thats a clear point we should think about it.

Here some thoughts to that:
Even if the certifications are 100% free, as long as they are limited, they have some value, so it could be very likely happen, that in the longer run some market is created around certifying people. I think we should not consider this as per se bad, as long as it helps to certify true people in an transparent decentralized fair way.

Nevertheless here an suggestion without fees, that would give more flexibility then the one week proposal:
one can make only 4 certifications per month
one can undo only 4 certifications per month
the upper limit of certifications is limited by the limited certifications per month and limited validity of one certification

now we have different models to compare.
4 certifications per month would give more flexibility then the one week proposal, but still would be very inflexible in case of bigger needs, see the numbers above. and still certifications would be limited, that means also here in the longer run an certification market could happen to evolve.
the question is if we want now think about or later, and yes we should also consider the feeling we give to new people.

2

Rules are becoming to much complicated.

Fees on block generation is not on accord with money creation symmetry from RTM.

3

? how can an economic transactions, that do not increase or decrease the quantity of the monetary mass (M) have an influence on the relative money generated per member? Neither they influence the monetary mass (M) nor the number of members N and therefore not the relation M/N

yes if the fees would be destroyed, then it would break the symmetry, because it reduces M

Distribution of fees through mining has as much influence as someone getting some payment for working something, in both cases M and N is not influenced. neither M/N. If mining would create new coins, then yes mining would break the symmetry.

So the alternative is to just allow 1 verification per member per week?

4

Where the money should come from to go to identities generating blocks without money creation?

Rules are becoming to much complicated cause you are involving money and certifications.

5

from transaction fees?

What I tried in the suggestion above is to give an economic incentive to stay honest and at the same time handle times when many new members want to join.

6

I also like the idea of rewarding securing the network. I believe, most humans don’t work without a personal benefit (even if there’s a community benefit, that work might just do anyone else). But this reward could also be implemented as a special “reward account” and users could decide to transfer some money (maybe each time he gets his UD / maybe in regular intervals) to this account. Anytime a block is created, all content of the reward account would go to the block creator. There are propably also other ways to implement the fundamental idea of a block reward.

2 Likes
7

yea true, but this topic was more meant to be about how to scale the verification process.
mining incentives would be another I think needed topic :smile:

8

I agree with the growth problem you identify. However, I am not sure the solution is in fees.

I would suggest another idea: what if the initial parameters could be adjusted over time? Perhaps algorithmically by adjusting them automatically to the network parameters.

For instance, 3 certifications required might be more appropriate for a medium group. If the currency network has very few individuals for the first year, certification can be annoying for them… But if they suddenly start to grow, the number might become limited.

This would be a way of adjusting the parameters to the network in an algorithmic way rather than introducing these fees that seem to be dangerous.

9

why dangerous?
in the suggestion above fees would only apply if you happen to use your free certification for this week already.
the thing with an algorithmic approach is similar to try to calculate algorithmically the needed high of transaction fee in year x in Bitcoin.

10

Perhaps dangerous because it attaches a monetary value to something that I feel shouldn’t have one?

The purpose is to allow for growth, so I don’t see why creating some sort of market out of certifications would be the only solution for that. It will be not unlike carbon credit in a sense… You have “the right to certify” one person per week. You need to pay to certify more. Perhaps people will start exchanging their “right to certify”: I sell you this item and you certify my friend in exchange. Or let’s accumulate credit and use it all to provide extra certificates. Or let’s build a business out of certifications. I give people money so they can certificate others. Etc…

What I mean is that it creates a whole virtual market out of certifications (which are just a way to allow people in), which could lead to inequalities.

4 Likes
11

yea i think its good to outline different solutions how to solve that, or at least which parameters we should choose, because certifying is one of the main key features.

I just dont want to have this particular suggestion dismissed already at this point, because using economic incentives is considered per se as something ¨dangerous¨.
even if the certifications are 100% free, as long as they are limited, they have some value, so it could be very likely happen, that some market is created around certifying people.
I think we should not consider this as per se bad, as long as it helps to certify true people in an transparent decentralized way for an fair price.

take your example about the carbon credits.
if the income out of the carbon credits would be distributed fairly and transparent to all people on this planet and if this certificates help to save our planet i would be most happy to use them.

currently the environment is mostly destroyed, because there is not much incentive to support the environment.

1 Like
12

I Updated the main post and added this:

I think thats a clear point we should think about it.

Here some thoughts to that:
Even if the certifications are 100% free, as long as they are limited, they have some value, so it could be very likely happen, that in the longer run some market is created around certifying people. I think we should not consider this as per se bad, as long as it helps to certify true people in an transparent decentralized fair way.

Here an suggestion without fees, that would give more flexibility then the one week proposal:
one can make only 4 certifications per month
one can undo only 4 certifications per month
the upper limit of certifications is limited by the limited certifications per month and limited validity of one certification

now we have different models to compare.
4 certifications per month would give more flexibility then the one week proposal, but still would be very inflexible in case of bigger needs, see the numbers above. and still certifications would be limited, that means also here in the longer run an certification market could happen to evolve.
the question is if we want now think about or later, and yes we should also consider the feeling we give to new people.

13

I like Sciths thought about a dynamic certification limit a lot. I think it’s right that there’s a need for additional certifications when a bulk of new users want to join. These waves could for example be triggered by media reports. So maybe the certification limit of a certain time period could be based on the growth of certified members in the last period.

As a formula:
CertLimit(x+1) = max{f(m(x)-m(x-1)), MinCerts}

where

  • CertLimit(x) is the max. Number of certificates a single member can issue in period x,
  • MinCerts is the minimum number of certifications a member should have per period,
  • m(x) is the number of certified members at the end of period x and
  • f(n) is a suitable function to issue enough certifications to handle the increased demand.

I’m not sure about what f(n) should ultimately be. Just that it must be continously ascending, i.e. for all i>j: f(i)>f(j) .
In this way there would be an increase in certifications each time there is an increase of new members.

UPDATE:

  • changed the formula: “min” to “max”
14

About your propositions @Arcurus, I am opposed to all of them this time :slight_smile: you will find details on why just below.

So a 10% growth every trimester (roughly the same as 3.33% / month) means we make x10 every 8 trimesters, so 2 years. Let’s say we reach this 10% of active members once we reached 1.000 members (because obviously, the very first members of the community will actively try to make people join). So we reach 100.000 members in just 4 years. That’s what Bitcoin did, we are all good! Two years later, we reach the million, even Bitcoin can’t do this! :wink:

So to me, your first question is off-topic. We do not need it.

About fees, we already taked about it in a different context though: Transaction fees: bug or feature?

I already answered why I am against in this topic: Certification best practices - #23 by Moul

Anti-certifications, or certifications revocation which is the same, are offensive weapons. We only need defensive ones.

edit: also wanted to comment this:

Making certification costly does not lead to honesty. It leads to power in the hands of people in function of their money supply. So the power to the richest people.

15

If I understood you right, you are against anti-certifications, because then a single group could judge to “anti-certify” a single person and in such way would prevent certification by others. I agree that this might be potentially unfair and could be abused.

On the other hand, I understand “certification revocation” as taking back a single certification by the individuum who made that certification. So it’s not the same. The result would be as if I never certified the account in question, or as if my certification would run out more early. This has propably a much smaller potential of abuse. However, it might be of use in some cases, e.g. if I get notice of someone cheating that I believed to be trustworthy (which isn’t so uncommon in our world :slight_smile: ).

1 Like
16

Yes it is not the same, but it is sill a voluntary act of violence against an identity. My opinion would be to say “OK, I made a mistake, now it’s done, I cannot take this action back.”

Just like when you said something hurting to someone, you cannot take your words back from him. Still, you can let time go. After some time, the hurted individual may have forgotten about it, or just forgiven. But this won’t be made instantly. You may have to wait months or years for that.

Certifications expire. So your mistake will be forgotten. We should just let it go and think before we act.

I know it is kind of philosophical position, but what is not? Libre Currency is not war, it is an acceptation (of both relativity & causality principles). This is pure peace to me, so I won’t include war in it.

I hope this will help you guys better understand my positions :slight_smile:

2 Likes
17

Ahh… comprehensible :smiley: You are right that this is propably a philosophical question. But if it’s an explicit goal of UCoin to keep “war” or competition out of the protocol (which I can comply with), then perhaps it’s best to leave it as it is.

1 Like
18

lol, sadly i dint include the need of renewing the certifications:
with one certification per week and currently one month validity of certifications, our growth would be ¨little" bit more limited :wink:

but maybe we could allow to renew without counting for this limit, then we could grow :smile:

currently my suggestion would be to use at least something more flexible,
like for example 4 certifications in the last 30 days instead of 1 per week

this would allow more flexibility.

19

Always keep in mind, whenever designing any certification methods. I prefer that at no point in time growth is restricted in any way. I like to recall as a reference point what Facebook did, which has after the bootstrapping phase of about the same 4 years a user base of 100 million, and then continuously a yearly growth of roughly 200 million people. Those stats probably might even double, since we are well more connected 10 years later. So whenever thinking about any parameters about certification, I suggest redefining and stacking up the point of reference.

I know you currently prefer to design it for a maximum user base of about 1million people, but I still see the need for a global spanning currency as well.

2 Likes
20

+1
but maybe not one community must solve that scaling problem. the value of different communities with similar parameters could be exchanged 1 to 1 to a meta currency.

the web of trust between communities in this meta currency could be handled the same way like the current member based web of trust.
if one community would not fulfill the wot anymore, only their new generated coins could not be exchanged to the meta currency.