Witness competition

[xeroc]

Why not apply this reputation to their payout.Lets say once a week the witness performance is checked.  If the witness was only 87% percent accurate for the time period, they receive 87% of their pay for the period. Pay for performance. The 13% not paid can be a net savings or used to payout a bonus to those who are 99.9% consistent over say 2 periods. This will motivate witnesses to do their job.

IIRC witnesses are payed on delivery anyway .. if they miss a block .. they don't get a pay ..

[Pheonike]

Why not apply this reputation to their payout.Lets say once a week the witness performance is checked.  If the witness was only 87% percent accurate for the time period, they receive 87% of their pay for the period. Pay for performance. The 13% not paid can be a net savings or used to payout a bonus to those who are 99.9% consistent over say 2 periods. This will motivate witnesses to do their job.

[xeroc]

I now understand your reputation idea and like it .. but it can be implemented as UIA(s) outside of the core protocol (depending on the required amount of "trust" and "decentralization")

For now, I would say, let's try all of this without bonds .. basically for two reasons:
a) extra time it costs to have this implemented
b) additional barrier of entry for "yet unkown" witnesses ..

Anyway, we should recap this topic when BTS2 is out and has proven itself stable ..

[theredpill]

Thank you very much @xerox

The reputation points meaning is to compare witness with themselves, for sake of sorting the reward.

For example:

Block produced 1 pts
Block missed -10 pts
Block unvalidated by sequence of blocks  - 10 pts
Block unvalidated -10000

Others criterias?

With low reputation one would still be able to produce blocks (unless they are voted out by the community or a delegate or the points reaches same arbitrary low value), but the reward will make no sense as business

I read the @arhag post, is almost the same, by I not like the fee or lock money to be fixed, also not like the lost fund part, just stop paying then and ban.

For response time problem we can put on each block a report of the response time of the witnesses of the last two or three blocks. Then define that the response to be considered of the witness is average reported time, plus his own response time, which is the deviation of the vote that he made from average results of others.

[xeroc]

Hey theredpill, welcome to the bitshares community,

IIRC, @arhag proposed to have a bond of some sort for witnesses. That being, you need to put up a collateral to be able to be voted to an active witness .. If you screw up you may loss your collateral.
I like the idea much because it allows that anyone can participate ..
re reputation: I am to so sure reputation is supposed to be required to become a witness .. IMHO it would significantly raise the barrier of entry for that that would participate honestly and does not prevent fraud at all ... It is also very difficult to put "reputation" into a metric (e.g. number) so that you can compare it ..

Besides that, I like your way of taking a closer look at how to improve the existing system .. and therefore: !

edit: this is arhags post: https://bitsharestalk.org/index.php/topic,17308.msg220313.html#msg220313

[theredpill]

Hello everyone

I would like to suggest a change that I think will get our project even more
attractive to move and improve security, reliability, decentralization
and response time from de witnesses, goes like this:

Competition:   

Competition is what keep people motivated to do their best, If you have
better ideas or resources you may be happy to use and be compensated fairly, but how to
make competition between the witnesses when they need to be voted
and mine in a predictable manner for our amazing speed and precision like this in the beautifully BM idealized system? Changing the reward.

Every consensus round, the witnesses will be positioned in an escale where the best would receive much better reward during the next sequence of blocks (witness will be normally randomly selected)

The criteria:

The criteria for ordering the witnesses is divided into 2 parts:
   1 - Reputation (60% weight suggestion)   Reputation will be most important and point based, pointing up block mined and great response time, point down missing blocks and very hard down invalid block
   2 - How much they are willing to invest in the system (40% weight suggestion)   The measuring of how much BTS/year they lock or the value equivalent of PoW that they have done multiplied by the price per Ghs of the specific hashing algo (price choose by voting) please keep in mind that I'm not fan of mining myself, I was in love to PoS until I discovered DPoS but let me try to explain why this:

Burning energy is stupid, but the option is of the person who are burning, but we cannot deny that is a very effective way of distributing tokens to people that will immediately give value to, because have spent energy into, and have contributed to the network a bit. Also are a great entering point for those who want to try, but do not want to spend money or not has, may have access to a school computer and start making fan.
This also generate interest for those who already have some equipment or cheap energy. We just need to make sure that investing money will always be better than burning.

Since the PoW is not inherent part of the protocol, we can protocolize lots of problems in each range of 100 blocks lets say, in varios popular hashing algos and within a range of difficulties, the first who sign a block with the specific solution gain the proportional reward that will be valid to they order on the next witness change round.

Advantages:
   Witness will get incentive to be more trustworthy, reliable and fast
   More people will came in because of reduced entering barrier
   We will be the first blockchain to be have the best of DPoS and PoW
   Lots of witnesses
   Demand for BTS because to invest locking
   Stable energy price reference

I'm a software developer and I'm thinking in implementing this myself, please hit me suggestions