SourceCred

Retroactive Initialization: Pre-requisite to distributing a lot more Grain

Currently, SourceCred’s cred is based almost entirely on activity-level data. As discussed at length in Supernodes: Moving past raw activity, this is a somewhat fragile metric, and misses contributions that don’t occur on GitHub or Discourse.

Once we have the supernode system (i.e. Initiatives, Artifacts, and Spotlights) working, I intend to re-orient cred so that it generally flows from the artifacts and initiatives to activity. This is likely to cause some material changes in the cred distribution (i.e. flowing relative cred away from people who have mostly been active via Discourse posts, and towards people who have been involved in key contributions). This may be an uncomfortable transition, but it will improve the robustness and accuracy of our cred distribution.

In practice, this will be a few weeks of work where we try to go through SourceCred’s history and write initiatives that describe the major workstreams that have happened thus far.

Once we’ve done this, we’ll be ready to kick the CredSperiment into a higher gear. I want us to start issuing much larger flows of Grain, so that the total Grain supply will better reflect the value of SourceCred. We’ll be able to do this with confidence once we have “retroactively initialized” the project.

Not quite sure how to interpret this.

Long term, it sounds like the goal is to flow cred to people who do things? (good)

  • If so, how does that take into consideration all the work that goes into formulating ideas (and the community) so that people are at the point where they are ready to take actions? Often it takes a lot of work to create a community. This involves reaching out to people, onboarding people, creating documentation for them to learn about the project, and keeping everything organized and running so that people can contribute. How will all this be factored in so that the model doesn’t just favor contributors who already know how everything works, but doesn’t encourage new contributors?

Short term, it sounds like the result will be reshuffling everyone’s cred? (sad)

  • This would be rather unfortunate as loss is emotionally weighted much more than gains. I’ve earned more cred than most, but frankly I wasn’t even aware of it until recently because I’m just having a blast talking to people. The idea of having my cred taken away, however, is not pleasant. This is because there’s no agency. I’m not doing anything different. I didn’t do anything to deserve to have my cred taken away. But someone is taking it from me. Not a good contributor experience.

I get that this is an experiment and that we’re learning as we go. Totally support updating the model to make it the best possible. That being said, it’s not just about the SourceCred model itself. It’s also about building the SourceCred community. Would highly recommend thinking through how such a move might impact the perception of the SourceCred community from those in and out of the community.

To build on this, if you were to increase the weight of some things moving forward that would be very different than retroactively changing the weight of things in the past. Humans evolved in a physical world. When you do work, there’s a result. You plant seeds, you water plants, you harvest grain. The grain in your silo doesn’t just disappear without reason unless someone takes it. If you wake up one morning and someone had taken your grain you might be rather upset.

If, however, you woke up and realized that your neighbor had twice as much grain as you it would be different. Your neighbor’s magical grain replication doesn’t affect your grain. You’ll still have enough to eat this winter.

Now, obviously we’re not literally eating our SourceCred grain in the winter, but… our brains don’t know that. All we know is that we did work, we earned rewards, and now those rewards might vanish into thin air!? Not cool. The caveman in me wants to move to a village that isn’t haunted by grain ghosts lol

So yeah, if you were to change the grain weightings might be better to do so in a way that is forward looking and increases contributions to things. It has the same effect in that it will dilutie current grain holders over time, but… it would feel different emotionally lol


Also, from a purely practical standpoint (which is obviously no one here lol), saying that you’re going to retroactively change people’s grain creates an incentive to cash out right away. If you know your grain will be worth less next week than today, why hold it? Why keep doing work to contribute when you’re better off just cashing out and moving on to something else? Again, not a good contributor experience lol

This is a key piece that this change is trying to address. Right now, cred is heavily biased towards activity on GitHub and Discourse, and misses a lot of “behind the scenes” contributions that are essential to the project’s health. By implementing the supernodes system (Supernodes: Moving past raw activity), we’ll have the ability to manually include or reference important but illegible contributions. Over time, I imagine there being “Cred Historians” who use this system to keep on improving the cred graph.

Cred isn’t really something that you “have” that can be “taken away”, similar to how a friend’s affection isn’t something that they are “taking away from me” if we part ways. Cred is a shared (and retroactively updating) social perception, not something you own.

That said, to make the psychology a bit easier, I expect that we’ll make the numbers go up. Right now there’s ~7000 cred across all contributors. I’m imagining giving important initiatives an average “cred bounty” of around 100, with the result that we would actually inflate cred quite a lot. So numerically, almost everyone’s cred will go up, not down.

This is exactly right. Grain is different from Cred! Unlike Cred, Grain is something that you own. Grain distributions never retroactively change. @burrrata, the only way your 357,532 ¤ is going anywhere is if you sell, transfer, or seed it.

Once we’ve done this cred re-initialization, we’ll have a more robust cred instance. At that point, we’ll start issuing substantially more Grain. However, I do not expect this will devalue the current Grain, because there’s still so much “head room” for issuing Grain to reflect SourceCred’s value. So far, at the $0.01 per Grain price, we’ve issued $60,000 worth of Grain. Since SourceCred’s actual value is much higher than that, we have plenty of room to issue more Grain.

Ok great :slight_smile: That addresses 80% of my concerns. Very glad that balances are going up and not down, even though that might be an irrational perspective. Thanks for explaining all that (yet again)

Would be great to think through some more accessible memes, narratives, and metaphors to make all these concepts “click” and “stick.” As is, it’s very awesome, but also very confusing. Maybe the Documentation Glossary could help us organize key ideas, esp the novel ones, and then make sure there’s easy ways to explain and think through them all. Like a mental models list, but for SourceCred.

Starting to catch up on all the recent activity. After reading 5-10 more posts everything makes a lot more sense. Wasn’t immediately obvious though. Not sure how many people will read this, but in the future whenever a big idea is mentioned (like say the supernodes) it would be good to provide a tldr on that big idea in the context of the post it’s being mentioned in. Then if people want to learn more they can follow the link, but if they don’t immediately have a few hours to reading multiple forum posts all the relevant info they need is right there.

1 Like