search

solana-labs / networkexplorer

Retired
13 stars 19 forks source link

Finalize tooltip language #566

Open cfl0ws opened 4 years ago

cfl0ws commented 4 years ago

I took another pass through the tooltip language.

cfl0ws commented 4 years ago

A few questions -

1 - I'm suggesting we stop using fullnode and only use "validator" or "lead validator" and have asked @mvines for his feedback on that in the spreadsheet.

2 - I think the Uptime measurement is still an open discussion in #425 The tooltip language may change as a result.

3 - I'm not sure how the top validators are ranked, in reference to row 19. Maybe @sunnygleason does?

CriesofCarrots commented 4 years ago

1 - I'm suggesting we stop using fullnode and only use "validator" or "lead validator" and have asked @mvines for his feedback on that in the spreadsheet.

Yes, definitely. We've made that change (fullnode>validator) through the solana code-base. I'm not sold on "lead validator" for anything, though. I think we can just use "leader" I'll take a pass through the doc with an eye to this.

CriesofCarrots commented 4 years ago

I also updated the "should click through to" column, and made a couple comments

sunnygleason commented 4 years ago

3 - I'm not sure how the top validators are ranked, in reference to row 19. Maybe @sunnygleason does?

Right now it's:

(currentBlock / stageDurationBlocks) + stakeWeightedBonus (expressed as score from 0-100)

The parameters of the function are totally placeholders - I haven't heard a final definition from anyone yet. When folks decide the real function & parameters, I'm happy to implement (or be part of the discussion if brainstorming is necessary)!

CriesofCarrots commented 4 years ago

I've been wondering whether top-validator ranking should have any correlation with the prize metrics @jstarry is developing here: https://github.com/solana-labs/tour-de-sol/tree/master/winner-tool @mvines , thoughts?

mvines commented 4 years ago

The formula @sunnygleason mentioned sounds fine to me. The final arbiter will be the winner-tool and it uses the entire ledger, whereas this is just a informal snapshot of where everybody ranks throughout the stage.

cfl0ws commented 4 years ago

There's a TdS leader conversation happening in #485

In it, I reference the reward program listed here.

Is the tool being developed by @jstarry consistent with that version of the rewards program?

Also, I'd be hesitant to display a leaderboard that's not consistent with the way rewards will be paid out. This can cause much confusion and frustration among validators.

This happened in GoS in a way. While there was no explicit leaderboard, people tried inferring who the leaders were, only to find out at the end, their inferences were inconsistent with the winners, as announced by the Tendermint team.

I'd suggest the leaderboard should be consistent with the winning logic or not shown at all.