Re: BYTEBALL: Totally new consensus algorithm + private untraceable payments

()

Decentralized? He looks pretty "in-one-piece-ish" to me...

12 public witnesses, explained in whitepaper

Yes, I am aware of that and that's exactly what I was getting at.

The witnesses themselves are not decentralized; they are anything but. You have a maximum of 12 very "central" entities securing the network. And while I believe that the witnesses do not have the same amount of power as, say, EOS delegates have, it is at the very least intellectually dishonest to speak of "decentralized witnesses".

Now, maybe you could actually decentralize a witness, by making it a group of entities, playing merry-go-round or something. I think this has been discussed before.

TL;DR:
Calling a single person a "decentralized witness" is misleading.

I spare us all an inappropriate joke about certain news items and the decentralization of a human being, for obvious reasons.

We could also have 100 or 1000 witnesses but the platform will become less secure if you have more witnesses. A single witness has no power at all and can easily be replaced by users. Users have the real power in the Byteball network, not witnesses. Only when 6 or more witnesses collude they can harm the network, but they still can't change anything in the past. In fact they have very limited options for abuse.

This is no answer to my original point, which is "a single witness is not decentralized unless the witness consists of multiple entities".

You are trying to make it your point so that you can fire off what you have said earlier, so I'll play along:

Users having the real power sounds nice, but has some serious flaws. I'm too lazy to go into this, but very simplified, choosing witnesses is not much different than choosing delegates in a DPoS system. Go take a look at Lisk and EOS to see how that is going.

A WITNESS IS A CENTRALIZED ENTITY AND WE NEED TO TRUST IT.

In what way do we need to trust a witness? A witness just sends transactions which (by definition of the protocol) cannot be fake or wrong or invalid in any way. And these witness txs are simply our way of attaining consensus on which transactions are final.

With Bitcoin, you don't really have to trust miners: if there are conflicting txs pending, then it doesn't matter which one they confirm. Whichever one they choose to put in their block becomes final. The only risk is with enough hashing power they could roll back and restart mining from an earlier block, thus changing recent history in a way.

Likewise, you really don't have to trust witnesses with Byteball: if there are conflicting txs pending, then it doesn't matter which one they choose. Whatever they choose to append their tx to becomes final. And the roll back risk does not even exist here.

So, Byteball is even MORE trustless than Bitcoin.