Topic: Obyte: Totally new consensus algorithm + private untraceable payments - page 1070. (Read 1234271 times)

I saw that the wallet was working fine just waiting for the launch, at least to see the community response to byteball

tonych, I return for one final clarification. Am I correct that if the 12 witnesses do not change, then they can only create one total order in the DAG even if they attacking? The only damage rogue witnesses can do is to stall (and indefinitely if they are the majority of witnesses) the forward movement of "stability points", i.e. finality of transactions?

They can stall by not signing units or signing branches that the other witnesses have not seen.

In your wallet, see Menu -> Settings -> Witnesses.
In testnet, I control all 12 witnesses, they are always online.

The testnet distribution will happen in late November (the bot was designed for livenet and still says 25 December, it doesn't apply to testnet).

pls let me know good..the first distrib will be done 25 december on bb tn or after 25 december on live net ?..we must paired btc addresse before end november or we can wait live net ?

tonych, I am going to leave this thread now, because I have other work to do. If there is anything you want me to read or respond to, please PM me at any time.

I suggest and I think you already decided to emphasize the asynchronous advantage of you blockless DAG compared to for example Steem's DPoS. This provides more robustness than a synchronous round-robin cycling through a set of witnesses that produce blocks in DPoS.

I hope my summary on the prior page of thread, exemplifies one possible wording of how to better summarize the algorithm of Byteball in fewer words for any cases that can benefit from a more concise summary.

Btw, your design helped me refine how I articulate my design in my white paper, because it helped me develop a general model for the taxonomy of consensus ordering designs. This is why I prominently cite Byteball in the current draft of my white paper as I consider it a keystone prior art in that taxonomy.

The stability point design concept is technically clever (and complex enough that I had to apply some effort to wrap my mind around it). You are obviously a smart guy.

hello tonych    gd mrg frds


pls question about = WITNESS  the 12

who are they ? how u select them ? when can we see them ?

what is there profile : capital, computer power, etc ......

are they up & running when ?

thanks
 

I agree. I remember a lot drama at CommunityCoin where the focus of the thread was completed diverted. Which is definitely not good for the coin. Here we need to focus on the coin development and features.

You can move the coin however you want but it should be on the linked address on the distribution date.

There's nothing absolutely fair. If they have BTC then they get BB, I see nothing wrong with it. I remember several years ago there was a coin called CommunityCoin, where it distributes according to the active members in this forum, evenly. Guess what, people created (or already have) 10s, even 40 accounts to claim the coin. Dev made huge effort trying to detect fraud, but many are well hidden until the time when they dump the coin, you realize that they are from the same wallet.

The way Tony proposed is not perfect, but at least fair using some information that people can not so easily fraud, which I think is a pretty good one which is simple and straightforward.

maybe 51% of witnesses should be fixed..this will a relative centralization, if the bb community voting is implicated.

i'm curious also of know if we must keep our coin on address up distrib or just proof our btc balance ?

ardor collect is a exemple.. users on the exchanges had no control over their nxt account and the funds raised were moved without user intervention for distribution..this prove tha devs have the super user power when they want.

Did not see the list in client. Now I saw it, will explore a bit.

I agree with iamnotback that we should use fixed witness list (or a pool of fixed list slightly greater than 12), this could avoid confusions.

Most of those ico addresses are multisig addresses which can sign a message instead of making a micro payment. Each of the private key holders needs to sign the message, but that could probably be arranged. The ico buyers would start complaining if all the Bitcoins suddenly moved out of a dev's address, but they probably wouldn't notice or care if that address signed a message to claim BB.

Loving this tech discussion, iamnotback has some serious knowledge.

Okay so basically as I said, same 12 witnesses globally (over long-term time window) unless some extraordinary global political will and consensus exists to replace 1 witness then again same 12 witnesses over long-term time window.

It seems the main advantage this has over Bitshares' DPoS+TaPoS is the asynchrony of the DAG and that exchanges can't vote for the witnesses. But I think DPoS normally has 100 witnesses, or at least it was 20 or so for Steem. And DPoS is in control of the whales so they can bypass politics to change witnesses quickly.

If the accounts are locked, how do they make the micro payment to confirm their address? they can't obtain the signature from their escrow in this case.

The system is designed based on assumption that the prevailing witness list is nearly static and evolves slowly.  Most often, we'll need to replace just one witness at a time, and before we need to replace another witness, the first change has more than enough time to diffuse into the system, so that when we replace the next witness we are doing just one mutation again. 

The need to replace more than one witness simultaneously is extremely rare, but still possible, and when it happens someone will have to coordinate the process so that users replace one of the defunct witnesses first, and soon after that, after the first change reaches stability point, they go on to replace the next witness.  This is not quite convenient, can prove chaotic at times, and requires a degree of short term centralization, but no system works optimally in all sorts of edge cases that one can reasonably believe are quite rare to happen.

To reach any significant mutation seems highly improbable.

For example to get 2 mutations, we need 1 of the 11 witness at the last stability point to agree with 1 of the 2 of the mutations we want to make. To get 3 mutations, we need that prior sentence plus 1 of the other 10 witness from the starting stability point to agree with 2 of the 3 mutations we want to make. To get 4 mutations, we need that prior sentence plus 1 of the other 9 witness from the starting stability point to agree with 3 of the 4 mutations we want to make. Etc.

Three mutations seems very unlikely unless those two witness were misbehaving, in which case it seems everyone globally will want to get rid of those two witness. But that presumes there can even be an objective truth about misbehavior and that agreement can be reached. Getting 9% of the witnesses to agree on 2 witness to expel and 18% to agree on 1 witness to expel seems improbable.

Four mutations seems to be extremely unlikely, requiring 9% of the witnesses to agree on 3 witness to expel, 18% to agree on 2 witnesses to expel, and 27% to agree on 1 witness to expel.

And that presumes that the replacements for those expelled witnesses agree with the subsequent mutations. So actually the odds are worse than I just stated. Four mutations actually requires 14% of the witnesses to agree on 3 witness to expel, 27% to agree on 2 witnesses to expel, and 41% to agree on 1 witness to expel.

If there is indeed a proliferation of witnesses, who in the heck is going to have any confidence that a majority of some set of 12 (of the multitudes of such sets) won't go defunct leaving all those addresses (money!) in limbo forever. The cognitive load and worry on the users is unfathomable.

The state of this system will either solidify on 12 witness every where which are highly trusted, or it will devolve into Swiss cheese. You still don't seem to understand the most basic element of monetary theory, which is that the value of money is solely based on the CONFIDENCE that it can be spent later and someone will accept it.

Rube Goldberg machines are intentionally more clever than they need to be.