That sounds great ! I would myself have to understand that better by looking at the code, and any other help on that is appreciated!
1 Like
The docs will be back on line very soon (changed location and dns catching up) that will help for sure, there is a load more info there. It would help me to understand what you have seen that was wrong though. You know me, “dog with a bone” if there is a technical issue and especially deeper in the xor space 
6 Likes
My feeling is that you are sending mixed signals.
You say you want a calm and steady discussion.
Your words come across as passive aggressive which makes it very hard to achieve your stated goal.
My question is what definition did David use ? “close group” is no definition is it ?
And what working definition did you give ? “effective close group” is no definition either, is it ?
If any of these are definitions, they hardly explain anything. Not enough for any serious discussion about them.
1 Like
I do really have to go now, and I hope people will respect that and I do hope to find this thread open and maybe with additional contributions. I will want to do my part and take time to review the routing table code and the documentations; but please understand that this also takes me time.
4 Likes
I absolutely do, my only question was if there was something to investigate if you’d leave the discussion 
I fully acknowledge I was and am passive aggressive. I don’t expect people to cheer and rejoyce for my comments made. I do ask for a focus on the questions asked.
I am passive aggressive because I do not agree with some decisions made by David, but that is why I left. David does not have to agree with decisions I have made. I still feel I have a legitimate place in this community and I still want to help with limited time where I can.
4 Likes
and my tone is much more softer, because I do appreciate David being open to this discussion! (as I expected him to be)
5 Likes
Yes, this is on purpose I think. The closer something is to a node, the more the node knows. The more further away, the less a node knows. This adds to security. It could destabilize the network as well, because the more you know, the faster you know when a corrupt node (or group) tries to behave badly. But that’s where sentinel jumps in. You probably know this stuff better than me, but for the other readers:
Sentinel makes sure that the network is safe (no pun intended) even while nodes and close groups don’t know that much about the whole network. So can you explain which part of this you want to change? I’ve read the document but it get’s quite technical quite fast.
3 Likes
Exactly, the sentinel (although at least when I left, was not activated in the code; but the idea here is to start the basis of a PAXOS consensus algorithm at the concept of a group.) Sentinel was not yet at the level that it would stand up against a formal argument, but for sure that was the direction (we again I can only speak now in past tense as I ve been out of the game) saw it.
The concept of a effective close group, while subtle in implementation, was important for me to get a hold on how to reason about (specifically churn) in the actual local close group, so those are two different questions, I believe.
thanks @anon40790172 I really appreciate you jumping into the technical discussion!
4 Likes
Well then that is definitely another point where we disagreed on technical questions! And I think it is super-valuable to have such discussions :), pro’s and cons
maybe, just because now it is getting fun, let me be a bit more clear about that statement: where I saw sentinel go, as I learnt the original proposal from David and then tried to reason about it myself and helped with the implementation - I hope David can confirm that that implementation was a legitimate effort to be exact to the designs proposed - …
so, where I saw Sentinel grow into was a PAXOS consensus algoritm where a full group was the “node” of a traditional PAXOS algorithm. But as some have shouted at me, I didnt have the stamina to stick to it
and I will use my lack of stamina to bail out of this conversation, but I do hope we can keep such discussions going. Again I am eager to learn and still committed to help.
3 Likes
btw to give some food for thought: the reason I wanted/want to formalise the group in above linked draft, was exactly to use that definition to build a formal proof for the consensus algorithm. Whether PAXOS based or not.
1 Like
I had to reread my draft paper to remember, but the subtle change I made in the routing code relating to the close group was that I implemented it as the node itself to always be counted as part of its own close group. - I referred a couple of times to the subtle implementation detail changed in this thread. (corollary 4.2 in that draft)
On churn the implications were less simple to describe. David and others might remember me talking about second order effective close groups, but as I said I was more patching than proving there!
Hope that helps @dirvine
2 Likes
Out of all the possible fields of work / passion that are possible out there in the world, both of you chose ones that are unbelievably similar and complementary. You must have much in common when it comes to how you want the future to unfold, so whatever happened, try to focus on and remember how you are still on the same team, pushing for total freedom and Decentralization for all the worlds people
And not everyone will have the exact Exact same approach, but that is a beautiful thing
Biodiversity is a very very healthy force of nature
Just remember we’re all working towards the same goals! Don’t let small disputes make you forget that about the others here
7 Likes
The real business world is a cruel and unforgiving place. There is nothing agressive or unfriendly about the tone here, its all about business, I think some members here dont understand thats how business rolls. Its not personal.
4 Likes
In the meantime, all the docs are still accessible at maidsafe.github.io (e.g. routing and kademlia_routing_table)
Also, the RFC titled New Kademlia Routing Logic might be a useful resource I think.
7 Likes
My guess is someone at a bitcoin company caught wind of the disruptive technology possibly about to be implemented in the safe network code called data chains. This is just a fun conspiracy theory folks.
1 Like
Thanks Francis! I read - a first read - over the routing rfc, and it makes very interesting assertions, and starts with new constraints set on the routing table.
I will need to read this again, but it is super exciting ! Thanks again for pointing me there.
5 Likes
Agreed! Competitors (trolls) are indeed cruel and unforgiving.
However, within a strong business that wants to remain successful, only constructive criticism is encouraged from all participants. These type of organizations strive to keep things business professional throughout this constructive process.
On the other hand, disgruntle employees do typically name call while on their way out the door and for the most part are typically dismissed.
A ask that you please help us remove the name calling and reduce (to the best of our ability) the aggressive and unfriendly tone on this forum that you may not notice but others do.
Thank you.
2 Likes