Author Topic: CP-nets, algorithmic voting theory and Turing complete voting languages  (Read 2544 times)

0 Members and 1 Guest are viewing this topic.

Offline Geneko

  • Full Member
  • ***
  • Posts: 187
    • View Profile
I am not actually tech concerned but feeling this feature of DPOS as extremely important. 
How far is this from actual implementation?

Any time frame?

Offline luckybit

  • Hero Member
  • *****
  • Posts: 2921
    • View Profile
  • BitShares: Luckybit
Can this be build on top if DPOS exploiting the Turing completeness?

Yes this can be built and much more.
https://metaexchange.info | Bitcoin<->Altcoin exchange | Instant | Safe | Low spreads

Offline xeroc

  • Board Moderator
  • Hero Member
  • *****
  • Posts: 12922
  • ChainSquad GmbH
    • View Profile
    • ChainSquad GmbH
  • BitShares: xeroc
  • GitHub: xeroc
Can this be build on top if DPOS exploiting the Turing completeness?

Offline luckybit

  • Hero Member
  • *****
  • Posts: 2921
    • View Profile
  • BitShares: Luckybit
I think algorithmic voting is the direction that DACs will go for achieving consensus through Delegated Proof of Stake. My thoughts on some of the possibilities of algorithmic voting can be seen in these threads: https://bitsharestalk.org/index.php?topic=10804.msg142243#msg142243

Let me try to simplify it into core concept points.

1) Algorithmic hiring via smart contracts: This means the shareholders define a "policy" and this "policy" has an objective which must be met in order for an entity matching the "attributes" to receive the "automatic votes" to be made a delegate.

2) Attributes: This includes the criteria that the policy is looking for. So if the criteria is referrals as part of a marketing smart contract then the contract will only say the attributes are met when and if they surpass a certain amount of referrals to win the votes as a reward.

3) Hired: This is what occurs whenever the attributes are reached and the terms of the contract are met. This means any random person could refer a bunch of people and the moment they pass the threshold within a specific time period (or unlimited time period) then they are approved by the contract. Once this happens the contract gives them all the automatic votes necessary to make them into a delegate.


I hope these concept points are enough to make it understood what I mean. The result of these ideas should be a system where we give out votes to smart contracts instead of delegates. The smart contracts would then give out votes to whichever delegates match our criteria and automatically fire those delegates by removing votes when they don't meet the terms of the contract.

Conditional preference networks become possible once you have a voting language and other mechanisms to elicit the preferences of voters. Artificial intelligence can play a role here because it can suggest or recommend voting decisions based upon their known preference just as Google or Chrome can suggest websites as you type it in.

This would allow artificial intelligence to help voters improve the quality of their decision making. This would also allow voters to precisely define their preferences in a conditional manner so that their votes are shaped by events. For example if you have oracles which give feedback to the voting script then the feedback from these oracles
could determine the choice of the script which determines where the votes go.


References

"Computational Voting Theory: Game-Theoretic and Combinatorial Aspects" (CRCS Lunch Seminar)

https://www.youtube.com/watch?v=yKLy8uSy2SQ

CP-nets: A Tool for Representing and Reasoning with Conditional Ceteris Paribus Preference Statements
Quote
Abstract
Information about user preferences plays a key role in automated decision making. In
many domains it is desirable to assess such preferences in a qualitative rather than quanti-
tative way. In this paper, we propose a qualitative graphical representation of preferences
that reflects conditional dependence and independence of preference statements under a
ceteris paribus (all else being equal) interpretation. Such a representation is often compact
and arguably quite natural in many circumstances. We provide a formal semantics for this
model, and describe how the structure of the network can be exploited in several inference
tasks, such as determining whether one outcome dominates (is preferred to) another, order-
ing a set outcomes according to the preference relation, and constructing the best outcome
subject to available evidence.
http://arxiv.org/pdf/1107.0023.pdf

Making CP-Nets (More) Useful
Quote
Preferences have been studied in philosophy, economics,
psychology, and computer science and have a wide range
of applications, such as e-commerce, recommender systems,
decision support systems, and control of automated systems.
A variety of methods have been proposed for modeling pref-
erences. The one that I consider here is that of conditional
preference networks (CP-nets). First studied by Boutilier et
al. (2004), CP-nets exploit the power of conditional ceteris
paribus preference rules to enable (in many cases) a compact
representation of human preferences.
http://www.aaai.org/ocs/index.php/AAAI/AAAI14/paper/viewFile/8641/8743
« Last Edit: October 31, 2014, 11:45:06 pm by luckybit »
https://metaexchange.info | Bitcoin<->Altcoin exchange | Instant | Safe | Low spreads