Random Reviewer

[friedger]

What is the problem you are seeing? Please describe. Lots of energy is used to game the app mining program and to define non-gamable rules. Furthermore, the top apps in the ranking are perceived as the best Blockstack apps while there are non-top apps that can provide higher value to other users.

How is this problem misaligned with goals of app mining? Results are perceived as unfair if they do not reflect the personal expectations about what a good Blockstack app is.

What is the explicit recommendation you’re looking to propose? Add a reviewer that assigns a final score between -1 and 1 using a standard pseudo random function. No theta function is applied.

Describe your long term considerations in proposing this change. Please include the ways you can predict this recommendation could go wrong and possible ways mitigate.

• This reviewer encourage app publishers to not rely on rewards as a business model.
• This reviewer encourage users to also investigate on non-top apps.
• This reviewer encourage news publishers to create new/other rankings.
• This reviewer is easy to implement.

Additional context

175

181

[njordhov]

TMUI already provides substantial "randomness" by using only five reviews for 75% of its score, but is plausibly dependent on choice of audience for TMUI. The proposal of a Random Reviewer will make the randomness... less arbitrary, and formalize the extent to which app mining is a lottery. It will distribute the rewards among more apps, and give more apps a chance to shine in the limelight of high rank.

[njordhov]

Support for investing more broadly rather than funding only the already most successful.

[ntheile]

Why can't token holders become app reviewers again? I really liked democracy earth and it aligned with the Blockstack vision.

[hstove]

I really think it would be a bad outcome if bad apps performed better than good apps because the bad app got a 1 and the good app got a -1. I simply don't think this is a good idea.

If you wanted to spread the wealth, you could change the payout function - but this is already happening. We're moving from a 20% decay rate to a 3% decay rate. That gives way more money to non-top apps.

[friedger]

@hstove I think there is more to it (being top or lowest ranked) than spreading the wealth, like satisfaction, desillusion, ... But more importantly the random reviewer "institutionalize" the fact that algorithms do not share your expectations. I think it is already the case that bad apps perform better than good apps (and I share your feelings about it).

The current definition of good and bad apps in the algorithms are just the minimum that we all can agree on. Adding randomness will bring more satisfaction now and then but in the long won't harm anybody (law of large numbers), especially if the decay rate is low.

In proof of burn, the leader is also selected in a "weighted" random way.

It might be interesting to hear what game theorists would say to this.

[hstove]

Saying "randomness will bring more satisfaction" is a huge assumption. It's clear that if you're a higher ranked app, and this score brings you down, you will not be more satisfied.

[friedger]

I would like to have more satisfaction for the good apps, not for the higher ranked apps.

Hank Stoever notifications@github.com schrieb am Di., 3. Dez. 2019, 16:40:

Saying "randomness will bring more satisfaction" is a huge assumption. It's clear that if you're a higher ranked app, and this score brings you down, you will not be more satisfied.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/blockstack/app-mining/issues/186?email_source=notifications&email_token=AALBYWP5IGUUPI4FDI4YU3TQWZ4WLA5CNFSM4JQ6RB52YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFZZU4A#issuecomment-561224304, or unsubscribe https://github.com/notifications/unsubscribe-auth/AALBYWOVVW52OII7WU6YKYLQWZ4WLANCNFSM4JQ6RB5Q .