Our decentralized future by a ZeroMQ founder

Our decentralized future

Pieter will talk about the urgent push towards a decentralized future. As founder of the ZeroMQ community, he will explain the vision, design and reality of distributed software systems. He’ll explain his view on the community itself, also a highly decentralized “Living System”, as Hintjens calls it. Finally he’ll talk about edgenet, a model for a decentralized Internet.

Continuing the discussion from Meshnets on Maidsafe:


I would love to watch a video between @dirvine and Pieter discussing the SAFE Network and the future of decentralized technologies.


Really like the idea of merge almost anything. I would really like to separate our code and tests in separate containers (which we are doing in the rust test) and then allow anyone to merge after CI says OK all tests pass, coverage up. I would really like to add, lines of code also less (whilst not counting huge lines with multiple semi colons and all that cheating jazz.

This would be neat as then a system of paying core dev could introduce an automated payout for a commit that stays in place for X months or similar (to prevent vandals getting paid).

It should be a goal for us, but we need to ensure it’s in such a way that the automatic Continuous Integration testing framework kicks in and also to ensure there is not a flood of commits to obfuscate a back door etc.

Key is as we are doing though, first reduce the codebase as much as possible, make it simple, make it readable. Then look at ways to fully automate and distribute development in a manner that allows simpler and very open collaboration and improvement.

We will get there I believe and remove the last parts of centralisation of SAFE. It’s a great goal and it is achievable, but folks do need to understand there will be cheats, thieves, scoundrels and worse. Any natural system will have this, as long as we know and can handle that then we are good.

Today and I see it especially in bitcoin/crypto type projects there is all this can you mathematically guarantee 100% and for infinity complete accuracy and God like capabilities every single statement you make. To me that’s the greatest weakness of any forward motion in this area and why I seem at times very flippant and uncaring about some things, like the safecoin farming (dampening) algorithms etc. (which you could provide an almost infinite amount of and achieve similar basic goals) These things find their way and do so naturally, so what we release will change for sure and should. Locking this for 40 years or similar is a mistake.

Entropy. arrow of time etc. shows exactly that change happens you cannot change having change and this means provide flexibility in your system and let them grow.

It’s a differentiator for me that many will argue about, but anoyone can fork us (but not make the code not free :slight_smile: ) and lock their algorithms for 40 years and as Pieter points out let the market decide. We can take it a little further though and see nature ultimately decides. i.e. the market in California decided to go water ignorant and uncaring to a degree (not picking on Californians here) and now they are being punished by nature and need to find a better equilibrium. I am sure we will all do the same with fraking and suchlike.


Anything worth sharing from the meeting with Pieter on Thursday?

Early days meeting really, but he has some good contacts for sure and we will chase these up when he is back home next week. We can be put in touch with the libsodium dev so that is good as well as looking at potentially putting udp and reliable udp in zmq. We need to see how important it is just now.

The longer goal though is his vision of mesh based networking which we will investigate further. Nothing concrete just now.

Main thing is now we have met and know of each others projects and goals just that bit better.


Maybe ZeroMQ is very different from the SAFE network, and it’s like comparing apples and oranges, but I prefer the SAFE model where the core is under tight control, and the development of apps on top is done with total freedom.

Did you check out http://zguide.zeromq.org/php:chapter8 …probably one of the best overviews of mesh networking you will find. If you do a search for UDP on that chapter, you’ll see the possibilities to be explored when David says

But doesn’t the SAFE network already have functions for things like messaging and finding other nodes? Higher level functions can be built as apps on top in competing and emergent ways instead of as a monolithic protocol stack like ZeroMQ, which will only be one of many possible solutions.

I believe it’s largely due to the early stages, that it feels as if the core is tightly protected. It will be much healthier for the SAFE network if there is a wide variety of core node implementations all talking to each other. Having only a single implementation would be an inherent weakness for the network.

So let’s not have a tight control over the core :slight_smile:


I see the SAFE network as a fundamental infrastructure, like TCP/IP although on a higher level. It would be chaotic if TCP/IP lacked tight control over its specification.

Yes, that is a good point: the specification has to be very precisely defined as a contract on the API. But there are countless of implementations handling TCP/IP

1 Like