-
Notifications
You must be signed in to change notification settings - Fork 15
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add first batch of CredCon and Partner initiatives #27
Conversation
Explorer previews
Score comparisonsFor lack of a good compare feature, here is a summary of the resulting score differences >1 cred. Reminder: please be mindful not to add possibly hurtful comments, such as by comparing people to each other! These scores are included to check, whether Cred is flowing from initiatives to contributors in a way that seems reasonable.
Some things which stand out to me.
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We should be setting the weight in the initiative as the fibonacci weight. Then, we can set a type-level weight (a prefix weight) of 200 on the whole initiative plugin. This way we can tune and rebalance the overall weight of the initiatives without needing to individually modify every initiative.
I think it's better to use the absolute weights and lean towards 1x multiplier on the type level. |
Based on a discussion we had and https://discourse.sourcecred.io/t/rolling-out-initiatives/698 for context, the proposal is to start out with 100x for this PR. Because we know there will be updates and improvements needed to complete it. Then increase this to 200x later as we're more confident we've got good coverage. As these following updates may cause overshooting, followed by dilution to happen for champions. Because more of the Initiatives Cred would flow to contributors who are currently not added yet. Which is not very enjoyable to see I think. So it's better to do another increase of weights at the end and make it look more like this: |
Based on the rationale explained in the last comment, added the commit which sets it to 100x. |
cbbe5e5
to
6ad8e51
Compare
## This is a preview This is a preview, as it's in need of another weighting round. Still missing from this, but identified at the team call: - CredCon: video interviews (need a user account for zak) - CredCon: CredRank prototype (have not decided fibonacci weight) - CredCon: Discord prototype (have not decided fibonacci weight) - Evan's explainer (have not decided fibonacci weight) Also great improvements to make from here: - Adding contributions and linking that to contributors Good but not as critical: - Making the timestamps more accurate ## Assumptions to validate ### Fibonacci to absolute weights: 200x The major point to debate is the weights we should set. In the team call we've decided on fibonacci scale weights, which I've included in these files as `_fibonacciWeight`. This PR includes a ballpark suggestion of a multiplier to use. The intention is to "inflate cred" by weighing initiatives more heavily than plugin-imported contributions so far. While keeping in mind that more initiatives will follow and inflate cred further. And adding more contribution edges will "dilute" the share currently flowing to champions. A multiplier which I think is in this ballpark is: 200x (e.g. a fibonacci weight of 2 = a node weight of 400) ### Dependency structure For the activities as part of CredCon (like hackathon, party, talks), I found they had very similar dependencies on logistical initiatives and CredCon itself. I've generalized this by depending on CredCon, rather than directly depending on logistics. Example: CredCon party, depends on CredCon, depends on car+food+lodging This will make the life of historians easier too, as adding more logicstics (maybe invites or scheduling) would only need to be set as a CredCon dependency, rather than for all activities as well. ### Example references / contributions For the main CredCon initiative, I've added several clearly related forum topics. These are both as an example, and to make sure this part of the graph is not exclusively connected through identities. ### Completion status Other than both Partner initiatives, the CredCon initiatives are set as completed. ## Technical aspects to validate The main point which needs careful manual validation, is to inspect the graph. For example using the "legacy explorer" to confirm the initiative nodes are added and have edges to *all* the URLs added. That means: - Champions: Identities / GitHub users / Discourse users. - References / Contributions: Forum topics (in CredCon). - Dependencies between Initiatives. Note: it's expected for links to an initiative to give a 404 not found, until the PR is merged to master. However they should be listed in the explorer.
The 200x multiplier seems in the right ballpark, however we want to avoid champions see Cred "dilute" because we'll add more contributions. Start with 100x now, and move to 200x as we have a more complete coverage.
70e7693
to
3065811
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good as a start! I'll note that @protocol's funding support is missing here, and will want to add that in. But we already know we need a system for allowing inlined contributions. Let's go with this for now!
Please update the title and description prior to merging.
In the merged commit: 3452820 I've slightly reworded and distilled the conclusions of the discussion in this PR, rather than have the commit message be full of open questions. |
Depends on #26
This is a first iteration
Edit: rather than letting this stall for too long, let's improve on this with follow ups.
Have a look at #29 for info on how to go from here.
This is
a previewa first iteration, as it's in need of another weighting round.Still missing from this, but identified at the team call:
Also great improvements to make from here:
Good but not as critical:
Assumptions to validate
Fibonacci to absolute weights: 200x
Edit: see #27 (comment) we'll start out with 100x for this PR, then move to 200x when we're more confident about having solid coverage of connected contributions.
The major point to debate is the weights we should set. In the team call
we've decided on fibonacci scale weights, which I've included in these
files as
_fibonacciWeight
.This PR includes a ballpark suggestion of a multiplier to use.
The intention is to "inflate cred" by weighing initiatives more heavily
than plugin-imported contributions so far. While keeping in mind that
more initiatives will follow and inflate cred further. And adding more
contribution edges will "dilute" the share currently flowing to champions.
A multiplier which I think is in this ballpark is: 200x
(e.g. a fibonacci weight of 2 = a node weight of 400)
Dependency structure
For the activities as part of CredCon (like hackathon, party, talks), I
found they had very similar dependencies on logistical initiatives and
CredCon itself. I've generalized this by depending on CredCon, rather than
directly depending on logistics.
Example: CredCon party, depends on CredCon, depends on car+food+lodging
This will make the life of historians easier too, as adding more logicstics
(maybe invites or scheduling) would only need to be set as a CredCon
dependency, rather than for all activities as well.
Example references / contributions
For the main CredCon initiative, I've added several clearly related forum
topics. These are both as an example, and to make sure this part of the
graph is not exclusively connected through identities.
Completion status
Other than both Partner initiatives, the CredCon initiatives are set as
completed.
Technical aspects to validate
The main point which needs careful manual validation, is to inspect the
graph. For example using the "legacy explorer" to confirm the initiative
nodes are added and have edges to all the URLs added.
That means:
Note: it's expected for links to an initiative to give a 404 not found,
until the PR is merged to master. However they should be listed in the
explorer.