Reduce Rewards for Regular Paid Contributors

šŸ—£ Advice Process
21

From what weā€™ve seen is that there is a consistently high correlation of praise rewards given to paid contributors - this is the broader issue that the case you just mentioned plays a part.

These suggestions seem sort of non-tangible and donā€™t, at least to me, show any clear approach. In my experience chasing after a community to change the way itā€™s been organically doing things requires a lot of manual effort and essentially boils down to policing behaviour.

We should ask ourselves -

Do we change the culture of praise to fit into the system or change the system of praise to fit into the culture?

22

based on feedback from @Griff I reworked the method of how we deal with contributors vs. non-contributors.

So now in this other iteration we donā€™t subtract rewards from contributors by a factor but rather give bonus rewards to non-contributors by applying a factor to their praise rewards (a factor greater than 1)

We then take the sum of all praise rewards with the bonuses and normalize it to fit inside of the specified amount of rewards to distribute -

check it out

basically the main effect of this change is more about optics - rather than reducing rewards for paid contributors we instead just apply a bonus non-contributors and then make it all fit into the amount we have to distribute.

1 Like
23

Iā€™d like to leave the next couple weeks open for testing but I would propose we come up with a solution to implement for quant 24 - which quantification would end for december 26th.

24

I like the change to optics. For future rounds, this could be done at the scoring step vs calculating rewards twice, right? So once the round is scored, the bonus score is allocated to non-contributors (which doesnā€™t change the instances of praise) and then the total rewards are still calculated with that instances of praise number.

We had also discussed a scale for contributors, since not everyone is a contributor at the same level. Was there a resolution on that? I donā€™t want to overcomplicate or delay, so we could re-visit a scaled contributor factor at some future date with some additional date from the initial implementation.

25

Well since this alternative method is providing a flat bonus to non-contributors there is no effective reduction to contributors.

In order to have a scaled reduction for contributors we would have to revert to the first method

26

That was the original discussion, but a scale could be on the bonus as well, so in this model for example a full time contributor would get no bonus, a part time something like 125% and non-contributors the 155%. (or whatever we land on). Like I said, I donā€™t want to overcomplicate it, and it seems like it would only make sense if there were wildly different levels of contribution (which I donā€™t think is an issue right now).

27

I made some changes anyway to reflect a ā€œcontributor scaleā€ using the 1st model

we identify three tiers of contribution

  • 1 = full time
  • 2 = part time
  • 3 = casual

we lookup the contributor name in master sheet and find their contribution tier number, then we apply a different factor based on that number.

this can be very easily adapted for the ā€œbonus onlyā€ model as well

2 Likes
28

Iā€™m glad youā€™re making this distinction as it makes it more fair for those who arenā€™t working many hours.

Iā€™d recommend being specific about the hours to leave less room for misunderstanding. Something like:

  • 20+ hours per week (no one is full time these days)
  • 5-19 hours per week
  • less than 5 hours per week
1 Like
29

yeah the hours are actually more specific inside the spreadsheet itself if you open it upā€¦

1 Like
30

Hey all, we just had the Quant Call earlier today, in which we discussed Mitchā€™s implementation on this matter as well as my own. Just giving an idea here on the difference - my idea was to extract outgoings from TEC multisigs in the last three months via a Dune query, and join that to the Praise data to set a sliding-scale bonus on those not being actively funded by the TEC.

Approach

I used data from rounds 1 - 16, extracted 3 months ago - and established two parameters: 1. the maximum rate of taxation; and 2. the percentage of multisig $xdai outgoings at which that threshold would be hit, explained by this chart -

image
image1605Ɨ969 76.2 KB

Links

Hereā€™s a resulting spreadsheet with the output:

Hereā€™s the process I took to extract said data:

Outcome

We all agreed Mitchā€™s approach is best for us, because an approach as above would take a bit of coding and finance overhead. Given the low number of paid contributors, the fact there is no concrete standard standards on how people get paid at the TEC, and that

  • some are getting paid via third party wallets, making these funds less clear;
  • multiple individuals use separate wallets for Praise and financial assets;

it would be more troublesome to implement a smooth line than an approach with graded steps.

1 Like
31

This is the final spec we rallied around:

Implementing 4 tiers of contribution (based on weekly hours worked) and applied variable bonuses to them (with full time contributors receiving no bonus). After we apply the bonuses we normalize it to fit inside of the assigned amount of rewards we have to distribute

There are a few parameters left to decide within the system:

  • what should be the contracted hours worked per week for each tier?
  • what bonus should each tier receive?

Please feel free to fork and play and propose some numbers that make sense. We will implement for calculating rewards for Quant 24, deadline is thus December 23rd!

2 Likes