Imagining new Discord Props Experiences

With the upcoming changes to cred weighting, many devs are riled about the limitations of the props/did-a-thing solutions. At risk of getting ahead of ourselves with the Creditor Design Sprint only weeks away, and hopefully in service to that process, I want to start an ā€œidea soupā€ thread for imagining how the discord plugin props/did-a-thing solutions could be different. I request that we stay high-level and accessible. I have 2 ideas to put in the soup:

1) Weight Confidence

Currently, a postā€™s weight is a sum of emojis, which creates strange UX where I might see a post I think is super valuable, but decide not to add an emoji because it already has plenty of emojis that seems like a fair evaluation. This also causes popularity effects, where posts from more popular people are likely to get more cred than a post of equal value from less-popular people.

We could instead imagine a plugin where emojis contribute to a ā€œconfidence scoreā€ for different valuations. Emojis indicate a value estimate, but reactor role and emoji count contribute to a confidence estimate instead of a value estimate. Example A: 1 person uses a ā€œsuper valuableā€ emoji, and 10 people use a ā€œsomewhat valuableā€ emoji, and so the system resolves on a slightly-more-than ā€œsomewhat valuableā€ weighting. Example B: 1 person uses a ā€œsuper valuableā€ emoji, no one else reacts. The system resolves on ā€œsomewhat valuableā€ because of the low confidence score.

2) Category emojis

We could implement a cultural tool of categorizing contributions with emojis ( :robot: for dev work, :gift_heart: for care work). This could be paired with tech tools like cred minting budgets per category, or weights per category. This would also help us build reporting tools that could give us insight into how much of our overall resources goes to different categories, or how much contributors in different categories are getting compensated.

10 Likes

We could create a discord channel with a github integration that posts all PRs from tracked repos in said channel. The community can then react to the PRs that are exceptionally valuable.

6 Likes

@META_DREAMER in the other thread:

4 Likes

This perspective definitely eases my concerns to a large degree. I think category based cred is a huge win, as we can more precisely tune the system to weigh care work (in this case) higher, without potential collateral damage to other cred flows.

Also, as you know Iā€™ve been a big fan of the confidence score style of reactions that you bring up here.

I also like the DisCo approach of a ā€œtaxā€ on those who do not to care work but who benefit from it. I enjoy this space because of the community and the work put it to make it what it is, and Iā€™m certainly willing to flow a chunk of my cred towards those who make the space what it is.

4 Likes

I think it would be awesome to create new emojis that visually represent the weightings. Like emojis that are actually +1, +3, +5, +8 icons. I think it would be easier for newcomers to learn a new emoji system layered on top of the way they already use emojis, rather than relearning how to use emojis they already use.

Some existing discussion of this: When to ā¤ something? - #3 by wchargin

7 Likes

To the github PRs in a discord channel :

  1. my mind is like, it could work

  2. then I have this example that comes up:


    Reading this is already a thing but why not making an effort. But if you need on top of that to evaluate it, that seems really tricky for non dev.

1 Like

Iā€™ve done some research on this and havenā€™t found a solution that exactly meets our needs yet, mostly seeing a lot of noisy bots like this one. Writing a bot that does this (and only posts to a channel what we actually want) wouldnā€™t be a large lift though.

1 Like

I like the common/rare/epic/legendary (or similar) type setup since it allows changing what the underlying weights are vs being stuck with the +8 on the emoji label.

3 Likes

MetaGame has been running into problems recently as our community scales where this solution would solve a lot of them. Cred inflation as the community rapidly grows is real and since the number of reactions on a post linearly impact its cred thereā€™s a huge variance with how much XP something gets and looking at the data shows us that the most valuable contributions arenā€™t actually getting the most cred, its just the ones that evoke the most emotions from people.

1 Like

Really liking both confidence weighting and category emojis.

Iā€™ve already kind of starting to do category emojis. Mainly just to be playful, but also with a growing realization that emojis could become an important data source for categorization. E.g. for docs-type tasks and note taking, I often use :open_book:. Generally nurturing, mentoring or self care I usually use :seedling:.

Iā€™ve also seen this in the wild. For instance, in 1Hive, when their token mooned, and the #memes channel was being gamed with low-quality memes, the ā€œmeme policeā€ emerged to call out said memes. Other contributors were putting :police_car: memes on the callouts.

image

This was pretty funny. But also potentially valuable. Weā€™ve talked about Cred Defenders before. Would these police emojis not be a good measure of that? Donā€™t think SourceCred wants cop cars rolling around lol, but we could do another playful emoji.

Decred has a bot that posts PRs to its #dev channel. People found it useful. Could be a nice way to automate ā€˜reportingā€™ for devs. Presumably PR authors could also ā€˜annotateā€™ posts explaining context to wider non-technical audience.

Using a Fibonacci scale here could be super powerful. +1.

Love these ideas!

Lmao, thatā€™s me :eyes:

I think this is a great idea, its also a good way to quantify (in a way) the amount of ā€˜workā€™? being done throughout a cycle.

In theory, this is a good idea, but the problem that I see is that people are more likely to react with the same emoji that was already there instead of selecting a new one.

3 Likes

I really like this discussion about going more granular with the props channel. It also echoes a good point made by @eeli on having better metrics to assess cred policy changes.

The format for the prop message would be :

  • the text : props to blueridger & metadreamer for the ideas brought up

  • a category emoji in relation with the work: :robot: or :gift_heart: or ā€¦

  • confidence emojis

I like these proposals, particularly the idea of designating category-specific emoji for particular types of recognized work. I feel less certain that creating budgetary buckets for different categories of contribution is in keeping with overall SourceCred ethos, but I do think that for both individuals and the community, it would be really nice to be able to track what type of work is getting done & what type of work is getting valued.

Iā€™d suggest, if possible, creating custom emoji for this purpose, because they would be searchable by specific labels, which would be helpful for newcomers and probably not a bad reminder for everyone as the community grows.

Along those lines, Iā€™m wondering if this proposal would mean that all non-assigned emoji lose their cred value? My gut says that going back to a only-specific-emojis-mint-cred system would offer some UX advantages across the board, particularly in the category of ā€œI value this contribution and want to react, but Iā€™ve done the math and think itā€™s been adequately credited already.ā€ I think that general trend, which Iā€™m sure is even more widely practiced than discussed here, is going to be a hard one to overcome, and Iā€™d be curious whether anyone has attempted an analysis of behavioral shifts before/after the changed emoji rules last year.

I feel like I missed where the Dev-bot-posts-for-emoji-ing came into the convo but in general Iā€™m in favor of the idea, especially because it helps to put development work in front of the whole community.

As Iā€™m going through areas of the project that have asked for ops support, one thing Iā€™ve been thinking about is whether it makes sense to try shifting non-dev task tracking to GitHub. Nowhere near a proposal at this stage but just noting that itā€™s sort of the inverse of this automate-pull-requests-to-Discord approach.

1 Like

Thereā€™s been much discussion about this in the past, and the consensus up to this point has been that github isnā€™t particularly accessible to non-devs. Weā€™ve made some progress with the docs team, but itā€™s involved a lot of tooling to abstract away the more tedious and technical aspects of git and github.

Are you mostly interested in project boards? We havenā€™t really built out the github plugin to capture project management-related activities, but if we can find a good project-management workflow, it would be a huge step forward for capturing that type of work (either with the github plugin or some more generic PM -like plugin).

One thing Iā€™m noticing, is that we donā€™t really handle the number-of-recipients vector very well. If 10 people get a single props for each doing X, then I would want to mint more cred for that props than if 2 people get props for doing X. Right now, itā€™s this weird eyeballing of how many people are mentioned and how many sum emojis are added. Itā€™d be nice if new experiences had a way to articulate this in the reactions. Not sure how. Maybe differentiation between ā€œtotal impact estimationsā€ and ā€œindividual impact estimationsā€. Like if 10 people did X, I could either say ā€œThe total impact of that was 20 credā€ or I could say ā€œEach person involved had an impact of 2 credā€.

I thought the Creditor designs did a good (first) job of tackling this with the ā€œ% contributionā€ component of the new-contribution-post. I feel like thereā€™s a lot more UX to dig into before a feature like that one can be functional, but I really appreciated the space for it and it seemed like a not-bad MVP to me.