46
General Discussion / Re: Raising the quantity of BTS witnesses & the introduction of vote degradation?
« on: March 06, 2017, 09:18:50 pm »
Do you want to rewrite your outline with the changes? In the meantime, I'm asking @Chris4210 if he can put this on his radar.
These are rewrited propositions:
The maximum possible total score to achieve 1000.
TOP 21 gets 3 BTS
TOP 79 gets 1,5 BTS
The first criterion: proof of geographical dispersion.
The maximum possible score to achieve 300
we want a large scatter in different countries and on different continents
maybe we could use net latency (to serwer X, Y, Z [some form of triangulation] to prevent VPN fraud) or some form of existing geolocation if we can trust it
We need a developers discussion how to achieve this
The second criterion: proof of ability
The maximum possible score to achieve 300
Threshold of 50%, if not achieved, the candidate is eliminated completely.
missed blocks(last month, last week, last day), server uptime
The third criterion: proof of commitment
The candidate must provide prices feeds for MPA
The maximum possible score to achieve 300
probably someone would have to evaluate the quality of the scripts – If the script is new and it is not widely used.
The fourth criterion: proof of reputation
The maximum possible score to achieve 100
Points as a result of vote
The first (eg.) 21 candidates will be witnesses, and receives 100% salary.
The next (eg.) Of 79 candidates will not be witnesses and they will receives some less pay (maybe 50% ) for it that they are ready to become a node in case of failure of other nodes
If 1 of 10 witness is down, next 10 takes his place and recive 100% salary.
if such a change would be accepted, I will definitely run one or more the backup nodes in Poland on variety of service providers
Proof of commitment, that is another topic, I see it in this this way:
Thanks for restarting this thread. Hopefully the idea of more formally applying witness criteria will get more traction.
Looking at your scoring metrics, it looks like you have proof of commitment twice, once for price feeds and once for collateral. I think it makes more sense for collateral. And I think the price feeds can be grouped along with missed blocks under proof of ability. I would also add latency to that. So under proof of ability you could have: a) price feeds, b) missed blocks, and c) latency. What do you think?
It would be great if someone could write some code to measure these metrics and publish them in real time as a resource for proxies and individual voters to assess the quality of witnesses. Perhaps if this works well it could even be built into the blockchain such that if minimum thresholds are not met, then witnesses can automatically be made inactive (or standby) until they meet thresholds and get voted in again.
As for paying standby witnesses, I assume you mean in a scenario where they would randomly get to produce some % of the blocks. In the past I have suggested that it might make sense to let standby witnesses produce perhaps 10% of blocks. This would obviously require a hard fork, but maybe it should be considered. Either way, it would be nice if we could make some progress on applying criteria as discussed above.