Consensus Methods Article

Hi all! I just found this article in the network and I feel a pain not seeing Datachains mentioned in this article:

Maybe with a little effort of the community and maidsafe marketing team we can create a small text describing the diferences between Datachains and Tempo (links below) and give a poke to the writer so that the article can be edited and Datachains added.

Main page related to Tempo:

Consesus mechanism white paper:

Nodes incentive white paper:

@mav sorry for summon you. But could you take a look to these whiterpapers and say what you think? (your freshness contributions lately puts you in a very good position). I understand that maidsafe devs are very busy and under pressure, for that reason I don’t expect nothing from their side, however they are more than welcome to participate.
Thank you all!!


From my understanding the SAFENetwork consensus model is Close Group Consensus. Data chains is the mechanism for reforming groups in the event of a network failure event which disrupts the group.

So, it is more precise to say the SAFENetwork uses close group consensus with data chains.


I think the sharding strategy is questionable (this is analogous to the split / merge algorithm on the SAFE network).

As the network grows each Node will be unable to maintain all shards and will need to prune shards until the resource requirements matches their own available resources.

Each Node must calculate which shards they wish to maintain and which they wish to drop. A good strategy for this is to select the shard set in which you have the highest aggregate probability of being selected to help create a Temporal Proof.

Nodes will naturally select shards with the lowest number of active Nodes as a ratio to the activity on those shards.

I can imagine for large networks there’s a reasonable chance of some shards having no activity and the data is pruned entirely off the network because of that. The value of data (to miners) is tied directly to the frequency of the data being changed. Sounds a bit backwards to me.

The sharding algorithm is quite clever, but I think the ratio of active to inactive data means data loss is inevitable, especially if the network becomes very large.

Haven’t read the temporal stuff in detail but the network is definitely interesting.


Hi :slight_smile:

how is it with pruning of unused, or rarely used data in SAFE?

No pruning. Data is persistent

1 Like

There will be highly trusted archive nodes. I think the idea is that such data might take a little longer to access, but will always be preserved. TBD.


thank you for the info, are those "highly trusted archive nodes some kind of “master nodes” ore bring some negative recentralisation problems?

I just discovered this great project and are super excited, and have a lot to study…
Can someone point me to good resources? (precise and detailed, but in plain English, understandable for a non mathematician/developer :wink: )

  • I want to learn about the architecture and technology behind the SAFE network.
  • I am also interested why it needs a coin, and cant be a moneyless network.
  • And how to become a farmer. (because i want to support the network)

Thank you very much!

btw: this is a nice forum! Does anybody know if its open source software?

sorry for spamming with unrelated questions…
but i have another one: how does it utilizes IPFS?
edit: i didn’t see the primer on the website…

They are simply nodes with large storage capacity. And they stay online for most of the time. They will likely have one time write media that can store massive amounts of data.

The only centralisation is that there are fewer of these than normal sized nodes/vaults. The data is stored redundantly the same amount so as far as the actual data is concerned there is no difference. There will still be the same number of copies of each chunk and the archive nodes will still need to follow all the rules any other node follows. I’d expect over time that there will be a lot of archive nodes so the centralisation will not really be there. They hold no special powers or anything.

There is a wealth of info here, read away. Use the search feature of the forum to help you

To have persistent data you need a way to prevent spamming and incentivising people to farm. The coin allows farmers to get paid for their efforts and by charging (even a tiny amount) for storing you prevent massive spamming (big discussions on this elsewhere) of data filling up the network before more farmers come online.

Follow the dev updates in #development:updates and in one of the beta releases will be testsafecoin. But it is still a while yet before that.

It a “discourse” forum

It doesn’t if you have not found out yet


Thank you very much! :slight_smile: