So if some people collude they can create new fake identities. If a lot of them do it this fake entities can create more fake entities. Finally they could take over the who group. I know there are some restrictions like a maximum number of connections per person.
However - have you done some analysis about the percentage of people that need to collude of a group of 1000 people who are decently connected all together?
P.S. In general I really like what you are doing!!
We often have this kind of disclaimer from newcomers of the forum. Do we look so easily offended? But thanks. 
For your question, let’s assume we have the following parameters:
Then we can imagine following graph:
Here, C stands for Corrupted and S for Sybil. As you can see, 2 corrupted members are able to create 4 Sybils. With the same idea, you understand that each Sybil can help to create up to 4 Sybils:
Of course, you can have a third level of Sybils (S13 … S28). But you can’t have another level.
Indeed, if C1 and C2 are members, they are certified by other members in a first place:
See: these 2 corrupted members could create 28 sybils in the worst case. Indeed, we assumed here that C1 and C2 were at a distance of 1 from the whole community, which implies every single member of the community has recognized them. That makes it very unlikely, if not impossible without having a WoT completely corrupted, which is a non-sense (indeed, having corrupted nodes is a comparison to non-corrupted other nodes. If everyone is corrupted, then no one is).
So in reality, the above case is only possible in the very early days of a currency for distance to be minimal (1) for the corrupted nodes.
In practice, with a greater number of members (I don’t have any number to give … but let’s imagine 50 members), we are more likely to be at a distance 2 from any other member (at best). So these 2 corrupted nodes could create only 14 Sybils.
And the bigger the community, the harder the Sybil attack.
I hope I have answered a bit to your question. 
From the discussion here I indeed got the impression that this might happen 
Anyway - thanks for you detailed answer. The max distance might create a “dead lock” where also real members can not be added without getting a signature from the sybills? But this is only a guess, I need to think about it in more detail.
Well, the distance rule does apply from every member of the WoT.
Indeed, imagine someone who just joined by being signed by 2 individuals. He doesn’t need to sign anyone, he just needs signatures. This way, this individual does not have any possible path from him to someone else, since he did not make any signature!
If the rule was applyed from him, the community would be locked. So a rule of thumb would be to apply the distance rule from members with at least 2 signatures (the same number that is required to join).
But isn’t that already a problematic result?
Let’s assume this is a vital group of 50 members with a money with a decent value: all it takes is 2 evil actors and they can increase their share of the groups money creation significantly?
First, 2 signatures was for the simplicity of the example, it is very low. We rather think of 8 signatures for our first try (~1 million members max).
You also have to see that there is a delay between each signature: during last FMM, we thought about 1 or 2 weeks between each. So the Sybils does not join instantly, we have time to see them coming. We could argue that this operation is time costly.
But more importantly, this type of attack can be lowered by using the distance rule. Typically, you just have to sign individuals you deeply know (family, neighboors, etc) and do not try to reach your maximum stock of signatures. This way, you limit the possible paths from you and maximize the distance between you and Sybils.
Finally, we have to understand that cheat have to be tolerated, by the simple fact we know we cannot avoid it. Even centralized systems cannot avoid it. So let’s assume this will exist. The only thing we want is to limit it so it won’t cause a currency collapse. By the way, it is exactly what our body does with bacterias 
And ultimately, this is a human organization. Other constraints may be applied on top of uCoin by them. Simply think about Laws.
In the “circles” proposal this problem is solved by an individual currency per person. This way you are only effected by sybils if you directly sign/trust them. The disadvantage is that it might be more complicated with way more currencies and thus the acceptance will be more complicated.
This is why there is also the concept of “groups”. Groups can have there own rules how to join and uCoin like web of trust rules might be very useful but with ethereum I can think about way more possible rules. The advantage is, that if a group should be taken over by sybils the damage is limited because the system would just fall back on underlying groups (groups can be hierarchical) or eventually on the p2p level.
What I clearly perfer about uCoin is the “no mining needed” blockchain.
Yeah, the circle proposal is smart. I really liked your approach from a technical point of view.
Thought, it cannot be applied to ucoin. In ucoin, every time new money is created, it depends of :
Because of this, we need to know who exactly are the members of the community, and what is the current monetary mass. We cannot have money issued locally without knowing the global state of the community, saved in the blockchain.
From what I understand from the RMT the result will be pretty similar to the result of the circles system (as soon as a stable ratio between money supply and newly generated money is reached). However, I might be wrong - I am really looking forward to find more of it translated into english.
Unfortunately, that’s not enough. Circle currencies allow to solve spatial asymmetry (everyone can create money units, in the same quantity), but not the time one.
Indeed, creating 1000 units in an economy with 10k units is not the same as creating 1000 units in an economy of 100G units. In the latter, you creation is peanuts, not the same proportion at all.
It’s one of two things. Either:
Case 1) will lead to have very strong money of young generations, while old people will tend to have a money that worth nothing. Huge power in the hands of young people.
Case 2) = Bitcoin. Early adopters create money. Not as quickly, I admit. But long term leads to it.
In the 2 cases, you create a set of individual currencies. I have to warn you about this idea I reject, even if in the early days of uCoin I had nothing against it.
The problem is: money is about common tool of measure and medium of exchange. By using individual currencies, we break the common aspect of the medium. This way, some mediums will worth more than others. Rule of 1.000 units/week for everyone then becomes an illusion of symmetry.
If you want real symmetry of money creation between a set of humans, you need a freedom money.
I fully agree to this.
However, you have overlooked that the basic income increases by 2% per year in Circles. (2% is still up to debate, maybe 4-5% would be better). Therefore the ratio between existing money and new created money will stabilize in the long run. It is mathematically equal to a 2% demurrage fee on all money - however, from a psychological point of view I preferred the constant increase of the UBI.
Ok, I misunderstood the rules then. Your solution is an approach to freedom money, but I am still sceptical about some points:
Ok, that’s enough for now 
Ethereum has a 12sec block time and every block has a timestamp. The timestamp of the latest block will the the accepted time.
There is no universal answer to this. Everyone can create as much accounts as he likes and create money. However, the incentives are in such a way that everyone should limit himself to one account and get this account/money recognized by as much as possible other people.
Well, it does not disappear if someone “leaves” but it might happen that no one will accept it any longer. This is where groups come into play to provide more stability.
Not excatly sure what you mean/ have to think more about an answer
I have the fear that a number higher than 4-5% will make the money not a good store of value. People need to hold a decent percentage of their savings in this money (and not spend it in gold/stocks/…) My thought on this number:
3:
The growth rate g = 2%.
The growth rate g determines the ratio between the total coin supply and the income per month. This number is subject to discussion. It should be chosen to maximize the value of the income.
Note that a high growth rate (inflation rate) destroys the capability of the currency as a store of value. Consequently, the potential dollar market cap is a function of the growth rate: market cap = f(g). (The market cap of the currency is a function of g that most likely will decrease sharply for bigger g like 5% or 10%). g should be chosen to maximize gf(g).*
The growth rate can be seen as a capital tax that finances the basic income. People with exactly the average wealth would pay the same amount of money in fees as they would get from their UBI.
Sure, with etheruem this is very simple.
Ok, hadn’t noticed you planned to build on top of Ethereum.
Ok, by reading a bit more the website, I understand your point. I still think many instability is brought by this “trust” connection system, since even “group money” is backed by an entity - the group, composed of trusted individuals.
So one day the “money” is considered money, and a year after it is not. I am really sceptical about how much I could trust in money stability.
There is no fear to have, only comprehension of time symmetry:
Because using 10% leads to the fact that, after 40 years, you’ve created (proportionnaly) as much money as any other member in the system.
More than 10% will make you rich this symmetry earlier, giving more weight to young generation, giving less will make you reach it later, or maybe never (because death will come up before). You then favor old generations.
Also, it is not clear how groups decide wether to accept or ban a member.
Where do you see an advantage of uCoin? Why is one uCoin currency different/more stable than a group on cirlces? I would argue that it is pretty much the same?
Well - with ethereum you can every arbitrary rule sets. One could clearly be a WoT one like uCoin. But every thing else is possible: centralized m out of n, majority of group members, …
Groups can also be hierarchical, so there can be groups that consists of just other groups. The group Germany, could consist of Frankfurt, Berlin, …
Because in uCoin (but also Bicoin & other altcoins) a coin has only 1 nature which does not change over the time: a coin is just a coin among others, and recognized as such on an equal basis. It is a common medium of exchange.
Once the system has acknowledged that this entity is a coin of the metabrouzouf currency (with a few confirmations, let’s say by a day) you known the value of this particular coin won’t change differently as any other coin of the currency.
In Circles system, if I understood well, this is not true. 1 Circle(x) ≠ 1 Circle(y). I guess this is the price of wanting not common currency, but a set of individual currencies (and group currency does not change this fact, since there is potentially more than 1 group).
Does that mean “members” decide of the rules when creating the group? Also, can the rules change thereafter? (without creating a new group)
I agree to what you are saying. However, as far as I have understood there should be very much different uCoin coins (groups). And one person can be member of many (hundreds?) All those (group) coins will have a different value? I would imagine that they are all traded on an exchange?
The value of such a uCoin group could change because some will flourish and gain in value and others will be forgotten? So this is the very same for the circles groups.
The only real difference is: In both systems a person can be member of lots of groups. In uCoin the income of a person is the sum(IncomeFromGroup1, IncomeFromGroup2, …, IncomeFromGroupN)
In circles you can be also member of lots of groups but you receive only one basic income. You can no convert this income to “group money” of the groups you are member of. Thus you will choose the most valuable. So instead of sum() it is: max(IncomeFromGroup1, IncomeFromGroup2, …, IncomeFromGroupN)
Yes, when a group is created you have to specify a ethereum contract that contains the rules of the group. This rules can also contain rules how to change the rules. However, I expect the first groups rules to be more simple, like the ones mentioned. I will publish some possible group rules in one of my next blogposts.
Some questions :
The rules of the group. See my last post.
Yes, in circles every person creates its own money. This money can be converted into the money of the groups the person is in. However, if more people join the system more money is generated.
You find the full ruleset here: http://www.ourbasicincome.wordpress.com (its only 12 simple rules)
P.S. or was this question about uCoin?
Yes “group currency” is equivalent to a particular uCoin currency. The main difference here is that in uCoin you will have different money names to describe different currencies, whereas in Circles you only have “Circle” currency to describe a set of different currencies (maybe I am wrong about this point if “subCircles”, the currency of a group, can have a name).
Also, I do not see why you could not cumulate several BI in Circle. Why could not you create 3 accounts, have each recognized in a particular group, have cumulate money creation?
I have absolutely no problem with this idea by the way. As an individual, I have several lives for each I could have basic income. Of course each BI would not have the same value, for example I could have:
Why would I have to choose on BI or another?
Anyway, you should definitely continue with your idea. I cannot say wether people will prefer your approach or mine (probably this won’t be all black or white ;)). If you search a bit in this forum, you will see that non-core members prefered your approach in a first time, and most of them still prefer it.
But what I am sure of is that you try to implement a UBI that is near from what RMT says, and that’s good