Author Topic: Gridcoin community discussing potentially utilising Graphene to scale massively!  (Read 4236 times)

0 Members and 1 Guest are viewing this topic.

Offline R

  • Hero Member
  • *****
  • Posts: 1004
    • View Profile
Decentralized voting (continuous) would also be needed for voting project in/out, or do you think that should be up to the elected statistics committee? (That may provide them too much power though).

https://followmyvote.com/its-alive-follow-my-vote-launches-blockchain-voting-software-on-the-bitshares-blockchain/ Pretty cool graphene voting developments - related to this post.


Offline R

  • Hero Member
  • *****
  • Posts: 1004
    • View Profile
I don't know how GRC distributes rewards, but from what you wrote I gather that you need something similar to a price feed. Let's call it a "work feed" - an (elected?) group of people gathers external data and publishes that on the blockchain. The blockchain has a built-in algorithm that looks at the feed data and from that it decides how to distribute rewards.
Currently, every windows client gathers BOINC statistics directly from the whitelisted BOINC projects, extracts team Gridcoin stats & produces a local table of stats for every registered (within the grc network) user. Once a day, the statistics (which the most clients agree upon) are included in a 'superblock' to which rewards are based upon when end users stake a block.

The problem with the above is that if we were to remove the team requirement to increase our userbase, the gridcoin network would be overwhelmingly consuming BOINC web server resources. Elected delegate roles for producing these statistics would solve this issue.

If it was possible for these stats to be fed into a built-in algorithm which paid BOINC users (rather than the end users having to stake a block to access their owed rewards) that would be pretty epic! To counter the issue of "I don't need a stake to earn rewards" (bad for marketcap), the algorithm could vary frequency of payouts to end users based on their base asset holdings.

There is no technical reason why the witnesses should be the ones to produce that "work feed". From a decentralisation perspective I think it would be a good idea to separate these roles.
It would make sense to separate the responsibilities out to different entities, however, voting participation needs to be incentivized somehow (or really in your face) - currently we're getting like.. 15% max voting weight participation on the GRC network.

The tricky part will be defining the "work feed" data so that it can be included in the blockchain in an efficient way. You may already have solved that on the current GRC chain.
The Gridcoin network has solved this to a certain degree, we could probably scale to 10k+ users with the current block storage mechanism, however if we were to aim for the current max ceiling of 560k+ active BOINC users we would be in trouble. On the other hand, users could use light clients & witnesses/comittee's would have the burden of storing this data..

I don't know how GRC distributes rewards, but from what you wrote I gather that you need something similar to a price feed. Let's call it a "work feed" - an (elected?) group of people gathers external data and publishes that on the blockchain. The blockchain has a built-in algorithm that looks at the feed data and from that it decides how to distribute rewards.
Pretty much what steem does .. except that EVERYONE decides on who gets the block rewards :)
Only on a very abstract level, IMO. Steem rewards a very subjective notion of "quality" (that I don't seem to agree with most of the time), whereas GRC tries to reward an actual measurable quantity of work.
You're right in that it would follow the same lines of Steemit - where rewards are issued by an algorithm (which is fed by data, in steem's case votes & GRC's case NN statistics) then distributed to users without end-users staking blocks & without witnesses having the ability to steal these issued funds.

However, as PC states - Steem doesn't proportionally reward users for their work completed (Steem's really a lottery whether or not large users see/upvote your post/comment) where as (gridcoin+graphene) would continuously & proportionally reward verified BOINC computation.

Offline pc

  • Hero Member
  • *****
  • Posts: 1530
    • View Profile
    • Bitcoin - Perspektive oder Risiko?
  • BitShares: cyrano
I don't know how GRC distributes rewards, but from what you wrote I gather that you need something similar to a price feed. Let's call it a "work feed" - an (elected?) group of people gathers external data and publishes that on the blockchain. The blockchain has a built-in algorithm that looks at the feed data and from that it decides how to distribute rewards.
Pretty much what steem does .. except that EVERYONE decides on who gets the block rewards :)

Only on a very abstract level, IMO. Steem rewards a very subjective notion of "quality" (that I don't seem to agree with most of the time), whereas GRC tries to reward an actual measurable quantity of work.
Bitcoin - Perspektive oder Risiko? ISBN 978-3-8442-6568-2 http://bitcoin.quisquis.de

Offline xeroc

  • Board Moderator
  • Hero Member
  • *****
  • Posts: 12922
  • ChainSquad GmbH
    • View Profile
    • ChainSquad GmbH
  • BitShares: xeroc
  • GitHub: xeroc
I don't know how GRC distributes rewards, but from what you wrote I gather that you need something similar to a price feed. Let's call it a "work feed" - an (elected?) group of people gathers external data and publishes that on the blockchain. The blockchain has a built-in algorithm that looks at the feed data and from that it decides how to distribute rewards.

There is no technical reason why the witnesses should be the ones to produce that "work feed". From a decentralisation perspective I think it would be a good idea to separate these roles.

The tricky part will be defining the "work feed" data so that it can be included in the blockchain in an efficient way. You may already have solved that on the current GRC chain.

Pretty much what steem does .. except that EVERYONE decides on who gets the block rewards :)

Offline pc

  • Hero Member
  • *****
  • Posts: 1530
    • View Profile
    • Bitcoin - Perspektive oder Risiko?
  • BitShares: cyrano
I don't know how GRC distributes rewards, but from what you wrote I gather that you need something similar to a price feed. Let's call it a "work feed" - an (elected?) group of people gathers external data and publishes that on the blockchain. The blockchain has a built-in algorithm that looks at the feed data and from that it decides how to distribute rewards.

There is no technical reason why the witnesses should be the ones to produce that "work feed". From a decentralisation perspective I think it would be a good idea to separate these roles.

The tricky part will be defining the "work feed" data so that it can be included in the blockchain in an efficient way. You may already have solved that on the current GRC chain.
Bitcoin - Perspektive oder Risiko? ISBN 978-3-8442-6568-2 http://bitcoin.quisquis.de

Offline R

  • Hero Member
  • *****
  • Posts: 1004
    • View Profile
https://cryptocointalk.com/topic/24622-dposbitshares-tool-kit/

Back when the Gridcoin community was looking for a safe haven away from emerging Scrypt ASICS (when GRC was Scrypt POW based), I proposed that we should have switched to using DPOS (back in December 2014!).

Gridcoin ended up picking Blackcoin's POSv2 mechanism instead of Bitshares' DPOS mechanism, but I've begun restarting the discussion over at cryptocointalk.

We're needing to scale from 2k users to 560k+ users, our main scalability issue is the neural network - currently each windows user gathers user XML files (some 1-2GB in size) from each whitelisted BOINC project & extracts team gridcoin stats to form the neural network with (to base GRC rewards with).

If we were to remove the team requirement, we could quite easily scale to 10's of thousands of users, but if we've got 10k+ (or 100k+) users all downloading these XML files, we'll be effectively DDOS'ing the projects. Several BOINC projects are based at universities or are home servers, not many could withstand this scenario.

If we had witnesses, they could gather the statistics to form the neural network instead of all users - drastically reducing the load our system places onto BOINC users.

We've been discussing this in the cryptocointalk thread and in the 9th/10th Gridcoin Community Hangouts (currently being edited, not live) - difficulties would lie in ensuring witnesses are providing identical neural network stats (punishing/catching witnesses deviating from legit stats) and in having witnesses pay BOINC users their owed rewards (rather than currently the case where end users are rewarded once they stake a block).

I'm open to discussing this in detail & would love to hear your thoughts on this idea.

Cheers,
CM.