Author Topic: Profits, Performance, Trust & Efficiency  (Read 58050 times)

0 Members and 1 Guest are viewing this topic.

Offline mint chocolate chip

Generating the next block should be as efficient as possible to maximize dividends.
Transaction validation should be as quick as possible.

What are we talking about here in terms of increased efficiency and faster transactions? If the advantage is it will be a significant amount more efficient and significant amount faster, which in terms makes it much more usable for actual users of the system and profitable for shareholders, I say go ahead and find someone.

Would it be possible for i3 to be the (initial) trustee, they probably have the most BTS so their stake in the success of the operation makes them a good candidate and they have the technology means to set up the backup systems to make sure there is no disconnect with the network.

Offline jae208

  • Hero Member
  • *****
  • Posts: 525
    • View Profile
When and whatever Bitshares looks like when it officially gets released many things will happen that weren't 'according to plan.' The best that can be done is trial and error. A hard reset of Bit shares may be required but that is OK. We simply get back up and try again.
http://bitsharestutorials.com A work in progress
Subscribe to the Youtube Channel
https://www.youtube.com/user/BitsharesTutorials

Offline luckybit

  • Hero Member
  • *****
  • Posts: 2921
    • View Profile
  • BitShares: Luckybit
Assumptions:
With the shares system we have an easy way to vote on anything.
With the TaPOS system the transaction ledger becomes immutable automatically over time as the ledger is confirmed by everyone on the network.
Generating the next block should be as efficient as possible to maximize dividends.
Transaction validation should be as quick as possible.

Solution:
Shareholders vote (off chain) on a trustee.
Trustee generates blocks every 30 seconds (or at will..... no need to require any particular rate)
If trustee is compromised or shutdown, shareholders can elect a new trustee by broadcasting their vote.
Once a new trustee has 51% of the shareholders support the network continues.

As a trustee you cannot double spend (you will be caught and fired).
As a trustee you cannot perform Denial of Service without being caught and fired.
As a trustee you cannot be coerced without being let go.

As a shareholder this maximizes your value (dividends, transaction speed, no potential of forks).

A trustee is not a paid position and requires almost no resources to run.  A trustee could even operate behind a tor node.

The result is like a 'constitutional company' where the laws are entirely defined in the constitution and the 'president' can be recalled at any time and has almost no power even when in office.

This same process can be used to resolve when a hard fork goes into effect.   

Thoughts?

The trustee seems to be a single point of failure. Why do we want that?
Speed isn't everything and even if you want speed, you can always optimize for speed later.

Trust? I thought the whole point is not to have to trust. If we have to trust it's starting to look a lot like Ripple.

The one way I would support this idea of having a trustee is to make it a conglomerate of trustees. These trustees would have to be anonymous, numbering in the hundreds rather than just 1, and they'd have to be selected completely at random from a pool rather than voted on.

Random select 100 or so from a pool to be the trustee conglomerate. Then randomly choose a trustee for different transactions. We should want as many trustees as possible, spread out around the globe. So perhaps a cellphone app?

The problem is if you don't pay someone something then why would they take the risk?
It reminds me of Tor nodes, it could work if there are enough of them I suppose but it's not decentralized enough if there aren't plenty of nodes.



« Last Edit: March 28, 2014, 03:37:06 pm by luckybit »
https://metaexchange.info | Bitcoin<->Altcoin exchange | Instant | Safe | Low spreads

Offline jae208

  • Hero Member
  • *****
  • Posts: 525
    • View Profile
 
There is always a window of opportunity before one gets fired. How much damage can be done before one gets fired?  You are essentially making someone President or CEO of the chain, which is a step away from decentralization.


51% is too much. See MemoryCoin. They could only get 5% to participate in voting.

Whoever gets that position "could" take advantage and do damage before 51% rally to correct it.

 +5%
I agree that we probably won't get 51% of people to vote. Great damage will probably be done before 51% vote. We might as well just go with POW for.
http://bitsharestutorials.com A work in progress
Subscribe to the Youtube Channel
https://www.youtube.com/user/BitsharesTutorials

Offline oco101

  • Hero Member
  • *****
  • Posts: 586
    • View Profile
If Bitshares x could reach that kind of speed for validating transaction will be amazing and it is worth taking in consideration.
 
The problems I see :

- No more decentralized
- One trustee one point of failure, the DDOS attack will be hard to prevent 

Quote
If trustee is compromised or shutdown, shareholders can elect a new trustee by broadcasting their vote.
Once a new trustee has 51% of the shareholders support the network continues.

Not clear. How long it takes the voting process , meanwhile all transactions are stooped ?  Can you elaborate more on the voting process ?

Offline NewMine

  • Hero Member
  • *****
  • Posts: 552
    • View Profile
There is always a window of opportunity before one gets fired. How much damage can be done before one gets fired?  Firing someone is reactive.


51% is too much. See MemoryCoin. They could only get 5% to participate in voting.

Whoever gets that position "could" take advantage and do damage before 51% rally to correct it.
« Last Edit: March 28, 2014, 04:40:32 pm by newmine »

Offline jae208

  • Hero Member
  • *****
  • Posts: 525
    • View Profile
IMO we probably shouldn't be trying to prevent every single possible inefficiency. The law of diminishing returns kicks in. There is no way to engineer anything into perfection. Sometimes you just have to release something into the wild and see what happens. Also, computer hardware is increasing at an exponential rate, the recent slash of Google drive storage prices shows this, what I mean is that every year that goes by we have more storage capabilities at a cheaper price so having something akin to a full Bitcoin node storing say 30 GB of information is not an issue in the longterm. As internet speeds are getting faster as well as storage speeds. Google fiber gives you 1,024 Mega Bits per second and only costs $70 and it will be deployed to more cities here soon.
http://bitsharestutorials.com A work in progress
Subscribe to the Youtube Channel
https://www.youtube.com/user/BitsharesTutorials

clout

  • Guest
Maybe I don't get the difference between security and block production, but I don't understand why whatever node with the most cdd can't produce the next block. Or make it a more ambiguous standard for block production that may not be probabilisticly determined. Which ever node has the median cdd in broadcasted set of transactions produces the next block. Or would it even matter if this was randomly determined out of those destroying coin days at given time?

Offline wasthatawolf

This might be too much to ask for common shareholders.  I don't really want to have to worry about trusting someone even if they will eventually get fired if they double spend.  Isn't this defeating the whole purpose of a decentralized system?

Offline HackFisher

  • Hero Member
  • *****
  • Posts: 883
    • View Profile
What's the concern to post this solution, is it to solve efficiency and performance?
Anything said on these forums does not constitute an intent to create a legal obligation or contract between myself and anyone else.   These are merely my opinions and I reserve the right to change them at any time.

Offline xeroc

  • Board Moderator
  • Hero Member
  • *****
  • Posts: 12922
  • ChainSquad GmbH
    • View Profile
    • ChainSquad GmbH
  • BitShares: xeroc
  • GitHub: xeroc
Side note: Maybe different types of TaPOS / POS should be considered for different applications. Some applications (Lotto, Me, Music etc.) might need more efficiency, some might need more trustlessness/no central point of failure.
+5%

clout

  • Guest
What happened to the ideology that everything that can be decentralized will be decentralized?

clout

  • Guest
Does Tapos plus pow not work?

Offline santaclause102

  • Hero Member
  • *****
  • Posts: 2486
    • View Profile
Here is another proposal (not sure if this would work):

Every stakeholder locks in a vote for a trustee which can be changed at any time. The chance for finding the next block is random but depends on a mix of your stake size and your trust rating you got from others. This could be combined with a ripple consensus: As far as I know with ripple there is a list of trusted nodes who vote on which transactions are included in the next block. Here such a list would be all stakeholders that turned (that turned mining/forging on) but the "weight" of your vote depends on your stake size and your trust rating. 

Wasn't a combination of ripple style consensus and POS considered before?

Offline santaclause102

  • Hero Member
  • *****
  • Posts: 2486
    • View Profile
Based on what would voters decide the on the trustee(s)? If real world identity is not revealed their is not much fundamentals which you can base your vote on. If real world identity is revealed there is the danger of trustee shut-down.

There would also be more potential for corruption than with NXT, because a briber would have time to approach the trustee. I don't know how long in advance the next one to find a block is known with NXT but the short time window would make it very unlikely to get a deal going...     

Was there a solution for 'who finds the next block' with TaPOS other than this one up to now?

Side note: Maybe different types of TaPOS / POS should be considered for different applications. Some applications (Lotto, Me, Music etc.) might need more efficiency, some might need more trustlessness/no central point of failure.
« Last Edit: March 28, 2014, 04:11:01 pm by delulo »