# Let's improve our Cred scores

**URL:** <https://discourse.sourcecred.io/t/lets-improve-our-cred-scores/914>\
**Category:** Governance\
**Created:** [November 17, 2020, 10:20pm UTC](https://discourse.sourcecred.io/t/lets-improve-our-cred-scores/914 "2020-11-17T22:20:27Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![panchomiguel](https://yyz2.discourse-cdn.com/flex032/user_avatar/discourse.sourcecred.io/panchomiguel/32/839_2.png) [@panchomiguel](https://discourse.sourcecred.io/u/panchomiguel)\
**Post date:** [November 17, 2020, 10:20pm UTC](https://discourse.sourcecred.io/t/lets-improve-our-cred-scores/914/1 "2020-11-17T22:20:27Z")

</div>

The Creditor is Coming (soon)

Meantime, there are things we can do to improve our Cred scores and our weekly Grain allocations. We don’t really have ways of proposing these changes yet, so this might also be a good opportunity to discuss how our community can propose changes to these instance parameters.

Things we can do:

-channel specific weights (raise the amount of Cred flowed in Props; lower the amount of Cred flowed in Random Chit-Chat etc.)

-trust level based cred minting

-futz about with the grain distribution so that Dandelion gets less of a lion’s share of the weekly Grain distribution and smaller contributors get more

The first 2 changes have been implemented by 1Hive, and my understanding is that it would be trivial to implement them in our own instance. The third change has to do with the arcana surrounding Grain allocation, would be great to hear feedback from @decentralion on this one.

What can we do to get changes of these kinds implemented, and what is our strategy for measuring consensus? What kind of consensus do we want? What other suggestions do people have about simpler, pre-Creditor fixes?

---

<div class="post-metadata">

**Author:** ![decentralion](https://yyz2.discourse-cdn.com/flex032/user_avatar/discourse.sourcecred.io/decentralion/32/7_2.png) [@decentralion](https://discourse.sourcecred.io/u/decentralion)\
**Post date:** [November 18, 2020, 1:22am UTC](https://discourse.sourcecred.io/t/lets-improve-our-cred-scores/914/2 "2020-11-18T01:22:55Z")

</div>

For changes to the Cred configuration, I propose that we use the following process:

1. Make a Discourse thread suggesting a specific, concrete change to Cred configuration
2. Include rationale on why we should make this change
3. Show analysis of how the Cred scores change, i.e. a before/after view
4. Discuss the change and see if there’s consensus in favor

We can view the following two topics as examples of this strategy

> [@Changing the CredSperiment weights](https://discourse.sourcecred.io/t/changing-the-credsperiment-weights/540/):
>
> Currently, SourceCred puts a very high weight on Discourse activity relative to GitHub activity; for example, the weight on a Discourse topic is 8x higher than the weight for a pull request. Node Type Weight Discourse Topic 8x Discourse Post 2x GitHub Pull Request 1x GitHub Issue 1x GitHub Comment 1/16x GitHub Review 1x At [the time I chose these weights](https://discourse.sourcecred.io/t/preliminary-credsperiment-cred/219), we had much less Discourse activity than GitHub activity, and I deliberately pumped the Discourse weights so that Discour…

> [@Minting Discourse Cred on Likes, not Posts](https://discourse.sourcecred.io/t/minting-discourse-cred-on-likes-not-posts/603/):
>
> Currently, we mint Discourse cred for activity; every topic and post creates new cred. I recently threw together a [prototype](https://github.com/sourcecred/sourcecred/pull/1587) which allows us to mint cred based on heart likes instead of posts. This post explores how this changes our cred. In keeping with the practices we came up with during the [CredSperiment weight changes thread](https://discourse.sourcecred.io/t/changing-the-credsperiment-weights/540), I’m not going to mention any individuals by name. However, the analysis happens at the level of individuals, so you can argue this is just a fig leaf. Let me know i…

---

<div class="post-metadata">

**Author:** ![s\_ben](https://yyz2.discourse-cdn.com/flex032/user_avatar/discourse.sourcecred.io/s_ben/32/816_2.png) [@s\_ben](https://discourse.sourcecred.io/u/s_ben)\
**Post date:** [November 20, 2020, 1:43am UTC](https://discourse.sourcecred.io/t/lets-improve-our-cred-scores/914/3 "2020-11-20T01:43:04Z")

</div>

+1 starting proposal convos in forum posts. Easy and accessible.

> [@panchomiguel](#):
>
> -channel specific weights (raise the amount of Cred flowed in Props; lower the amount of Cred flowed in Random Chit-Chat etc.)

Like the idea of flowing more Cred in #props, but when is the logic being implemented that gives most of the Cred to the person mentioned? As long as we don’t have that, we’re building up a big Cred shift once we turn that on, and we don’t want to add more volatility there.

> [@panchomiguel](#):
>
> -trust level based cred minting

We haven’t turned this on already? I suppose we don’t need this as much as Maker did, as I don’t think we really have a Sibyl problem. But might as well implement it. I think we’ll see a slight increase in Cred score quality just due to higher trust level users minting a little more Cred.

> [@panchomiguel](#):
>
> -futz about with the grain distribution so that Dandelion gets less of a lion’s share of the weekly Grain distribution and smaller contributors get more

Just learned 1Hive has implemented a new distribution scheme that uses the Recent policy. From @sandpiper (what is their name on here??) on Discord:

> Oh, btw, 1hive started using recent with a 30% decay factor on Monday and seemed fairly happy with it …switched from 20/13 HNY immediate/balanced to 25/8 recent/balanced

Will be doing some simulations for Maker as well, comparing different payout configurations. So we should have a bit of data and analysis to inform our decision soon.

---

<div class="post-metadata">

**Author:** ![Santigs](https://yyz2.discourse-cdn.com/flex032/user_avatar/discourse.sourcecred.io/santigs/32/858_2.png) [@Santigs](https://discourse.sourcecred.io/u/Santigs)\
**Post date:** [November 22, 2020, 8:44pm UTC](https://discourse.sourcecred.io/t/lets-improve-our-cred-scores/914/4 "2020-11-22T20:44:48Z")

</div>

Sandpiper has been doing a great job showing the impact cred changes introduce in every account whenever 1hive has decided to change them, especially to fight gaming. They create a spreadsheet file showing the impact on every change and the total impact compared to cred distribution before changes were introduced.