Update 06 April, 2023

Nah - I’ll stick with @Southside for now. Thanks all the same :slight_smile:

4 Likes

It would have suited your icon. :stuck_out_tongue_closed_eyes:

But good to have you back in the land of the normal.

3 Likes

The icon is Rank Bajin - the resident villain of Calton Creek. He is immortalised in bronze as part of the worlds only two-legged equestrian statue.

image

Many good things came from Troon…

9 Likes

Great update! I’ve been watching from a bit of a distance and it’s hard to make sense of it all.

It all sounds very promising and I hope the new design and network library brings the safe network to life. It has been quite the journey!

22 Likes

Bajin is the only main character in the strip who speaks with a standard middle-class English accent

1 Like

Very true, dear boy.

1 Like

Splendid! Chin chin old chap

1 Like

Shall I pour you a G&T to go with your jellied eel?

2 Likes

It would be great to have a summary of the pros/cons of the two approaches.

Are there different areas of concern/dangers with the new approach?

Obviously, if it means having a working network at all, that is a huge gain. It would be interesting to know what sacrifices may need to be made though.

11 Likes

Theoretical comparisons won’t shed much light. Following pioneers means twists, turns, backtracking and leaps into the unknown.

For myself I’ve learned to keep abreast of the journey, understand up to a point, and to accept it is a rollercoaster ride that just like in the fair, I have to trust will get me home ‘safe’.

And I help out when and where I can because I want this idea to succeed although lately there’s little I can contribute due to a lack of time and capacity. I keep hoping to free up time to do a bit more but :man_shrugging:t3:

Like everyone here I’m hopeful that David and the team are on the home run. It does seem likely so I’m looking forward to the next test network and what follows.

We should soon be able to see for ourselves.

19 Likes

The highs and lows of real adventure and discovery for sure.

16 Likes

Nice update. Been here since this gig started. Never sold a single coin. Can’t wait for a stable network to MAKE SHIT HAPPEN

21 Likes

It’s really down to understanding and accepting group consensus works, but strict binary type consensus does not (in p2p networks). It needs the decision makers to be stable and by stable, that means no change. The complexity is also bewildering. However many Engineers see this as the way forward and with incredible levels of intransigency.

MaidSafe (original), then algorand, avalanche and others showed group (or whole network) consensus works much better and scales.

MaidSafe has had many engineers who did not accept group consensus but instead focussed on classical consensus. However these other projects provide proof positive the original model is not only simple, but it’s extremely stable.

With that in mind then the tech stack.

Folks will remember our “rotting team”, basically that was the core of the team and the focus of management of the team for a few years. I found that team more and more difficult to work with, not personally, technically.

The function of the routing team was. stable p2p network with all the things like hole punch, ip based PKI and so on. Then have kademlia on top of it. That was a job we did not think we would need to do, but it was the most difficult part of the network.

So researching some network parts, it became clear libp2p had a network stack and kad impl , that opened up the opportunity we have now. The routing teams jobs done with that (we hope).

But in terms of classic consensus using abba etc.to determine voters it cannot work. We cannot order the world in that way.

You can see the maths. We have 7 voters, 2 can be byzantine. We need 5 votes for a decision.

So now a single honest node fails. We now only have 4 honest voters. So the byzantine nodes just kill the network as they can just stop all decisions being reached

Fundamentally that is what kills the parsec/abba type approaches. So mathematically they have guarantees of correctness and that is what attracts people, but that guarantee requires there is a guaranteed stable set of voters, and this eludes people.

So Safe, bitcoin, avalanche, algorand etc. all sacrifice correctness for liveness. To be honest most of the definitions around consensus are “head up ass” in my opinion, but the main thing is nothing in nature is perfect, there is always error, but the systems that can handle error and still work and present the correct answers are the systems we want. Not perfect internal correctness.

I detest that 100% correct thinking. If we made a ballbearing 100% exact it would not turn. Mathematicians will love its beauty, but the rest of us have a dead machine to work with.

22 Likes

YES YES YES!!!

This is evolution, this is anarchy, this is free-markets. This is decentralization … this is what historically is proven to work.

4 Likes

Thanks @dirvine ! It sounds like a pretty fundamental pivot in approach.

Will the safe network using the new approach still be able to do all the primary things we hoped for? Is the potential feature set unchanged or are there substantial limitations?

If there will just be a simpler way to achieve the same goal, that’s awesome!

7 Likes

The beauty is that this is the Safe network from way back. It has all we figured possible, plus much more. What we have done here is show group consensus was always good enough.

With libp2p handling the networking components in a large part this is now a significantly simplified code base. So much so that core dev may be a large % paying some libp2p guys as that is what we need as the foundation to build Safe on. The Safe part of the code base in core now is tiny in comparison to a month ago and the complexity has all but gone.

In terms of not only speed to launch but certainty to launch, I am 100% certain now that it will happen very soon. At least Fleming imminently I would hope.Maxwell and upgrades should now be much much simpler too.

To me the most important thing is we can all understand the code, but even more so, I can talk of this network as the network that is influenced by nature. So if you like a pivot back to Safe as it was intended.

It feels really like a 10,000 ton weight has lifted of the team here.

30 Likes

You intuitively understand the ancient truth, that chaos is order, because order cannot exist without chaos, and together it is a dance. And this dance is the essence of nature… And thus the Safe Network is natural.

7 Likes

DBC’s do not use Raft consensus, right?

Bard appears confident that it does, but I think Bard is smoking crack.

3 Likes

It sure is. No DBC’s all use group consensus now.

7 Likes

so are we still going for the payment network first approach or has everything become so much simpler now that we can go back to launching the full network?

17 Likes