I believe it is the full network in terms of Fleming to begin with, but correct me if I’m wrong.
3 Likes
We are putting everything in place. So there is an option for payment first etc. but the technical issues are behind us so we can choose the best launch options here. I would say the technical hinderance is not an issue now.
So let’s see how the dust settles this week and probably part of next week, but we will see. It seems to be flying along nicely.
22 Likes
Is there possibly some old write-up about group consensus? I tried to search the forum, but I’m only on phone now, and didn’t find anything clear cut topic or some such.
Also, now that we are at this point, maybe it’s soon time for some marketing 
4 Likes
Fantastic update. Perhaps a dedicated thread for discussions on the pros/cons etc of a network vs payments only initial launch?
4 Likes
You need to search for “close group consensus”. He’s an explanation from the SN website:
4 Likes
Two more questions:
-
Does CRDT:s still play a role?
-
In this new (old) setup, do you @dirvine expect the concept of “edge case” still be valid/ needed? When talking about membership, concensus etc. I always grimaced, when I heard talk about “edge cases”. Just feels so wrong to have a division between cases and edge cases. There should be only cases.
1 Like
This is exciting to hear.
One question I have is; how can you have 100% certainty before seeing a successful and stable test-net in the wild? Does the new/old approach and libp2p certainly solve all issues seen in test-nets of the past?
Very much looking forward to seeing the progress in the coming weeks. It’s fantastic to hear the team is so fired up about this shift!
One thing is for sure… the world really needs Safe right now, so its realisation would/will be phenomenal.
8 Likes
Yes and for sure client side. So work offline etc. then update when you reconnect. Safe will handle any data types we want, but present them as mutable (append only with conflict resolution) or immutable (chunks). Safe may have registers to allow git like work, but that may also be pushed client side.
Not quite sure what you mean here?
We don’t care about membership any more, they are all just wee ants, i.e. individual nodes. It’s a probabilistic network, so there always exists the probability for anything to happen, it’s just we make the probability of a bad thing to be negligible (like bitcoin where it’s possible to create satoshi wallet, but not feasible).
The security comes from many places
- Num nodes makes targeting difficult
- Using a vrf type mechanism makes offline key generation attacks unusable
- Having decisions come from a close group to the data is powerful (insta sharding, but done properly and automatically)
- Extending close groups using the
Hash(data_name)and then `Hash(Hash(data_name)) means a close group could be attacked a yet rendered useless (and detectable). i.e. in a close group there is a range of data to mess with and opportunity for some kind of attack where DBCs are created and spent in that group (kinda useless, but an attack never the less). Using this trick it mean rather than just a larger close group we distribute the data again across the address range. i.e. the close group now has no dominion over the data in it’s range and the data is in fact in many different close groups. So when you look at the chain of where the data is, then it’s unfeasible that more than 1 piece of data is shared between the 3 close groups in this case.
We all have knobs to turn here
- Size of close group
- number of recursive groups (the hash(hash trick))
- number of replica copies of data
- number of replicas to ask for a decision
I hope that helps
13 Likes
Ha ha, no wonder, it was very unclear. 
I just meant that something in the old ordered approach always made me feel that the problem of changing the authority while moving was somehow half-done and half-thought. And to define something as an “edge case” was a sign that the problem was not seen from far enough nor close enough. It was kind of convoluted so that yes, you can untangle it to a point, but there’s this small looking part(s), that are called “edge cases”, in a way to diminish their significance, even though the system needs to be water tight. I don’t mean that this was done intentionally, but… somehow.
I don’t know if this makes any sense to anyone else. I understand that there will always be situations next to failure, and they are rightfully called edge cases. But I mean something else here.
Kinda like if we talk about a bucket, then the brim is a rightful edge case, but a hole in the bottom is not.
It’s getting late here… I better get some sleep.
2 Likes
While we’re (almost) on the subject - it’s “close group” with an ‘s’ sound as in “nearby group”, and not with a ‘z’ sound as in “not-open group”, right?
4 Likes
Yes sir. 
And by “close” it is meant as close to each other in XOR space, or as others are saying, close in the address range. Not close in the physical sense.
Does the number 7 or 9 still hold significance or relevance in a close group like it did for Elders of a section?
That is neat because the data isn’t just redundant but redundantly overseen by a totally separate close group. Would be basically impossible to target multiple random close groups, not to mention churn.
I should probably stop asking so many questions and just let you guys whip the network out and with the reduced complexity, we’ll probably be graced with an updated Primer in no time. 
11 Likes
Not so sure, but not 100% on it either. However kad groups are generally better in power of 2 size. i.e. 4 8 16 32 etc.
We did a bunch of work on that a while back, we can dig this up to check the size we want and test it in the field now (thank God)
10 Likes
So 7 of 9 it is? 
sorry coulden’t resist…
8 Likes
Haven’t heard or seen much from @jlpell lately. How are you doing man and how are you feeling about these updates?
10 Likes
Curious, after these design changes, will the whole token supply be generated at genesis and distributed immediately or whole supply generated then unlocked by close groups as resources are provided? Or will only the crowdfund/investor/etc tokens be generated at genesis and remaining supply be minted as resources are provided?
2 Likes
I’d expect it will follow the already submitted and approved RFC for the token. To change that now would be another long winded effort to do it. And @JimCollinson might not look forward to redoing it over again.
6 Likes
Correctamundo
7 Likes
So effectively 30% genesis supply to meet investor obligations / foundation funding and 70% remaining supply to be released by the network for Resource Supply Rewards.
1 Like
So effectively 30% genesis supply to meet investor obligations / MAID > SNT swap / foundation funding and 70% remaining supply to be released by the network via Resource Supply Rewards.
Diagram for the eye candy:
12 Likes
Super diagram Jim 
“Upload triggers an emission”
This is a crucial element. It accounts for 70% of supply over time and remains undefined AFAIK.
Ideally it would be created using a deterministic pseudo random approach, something like the original coin address space, but now this must be done using DBCs which are fundamentally different: stand alone data which can hold any quantity of SNT.
Seems like a hard problem?!
4 Likes