CredSperiment Progress Report

[burrrata]

@decentralion

We’ve made a lot of progress on getting Discourse and GitHub both working together in SourceCred. The official SourceCred prototypes now include:

• cred for sourcecred/sourcecred
• cred for sourcecred/research
• cred for sourcecred/widgets
• cred for this discourse
• cred for all of the above, combined

I’m getting 404s on the cred links?

@s_ben

While I appreciate the crediting, I would ask to have my pseudonym disconnected from s_ben (and edited from the comment please). Still have the same reasons for using the pseudonym.

If you don’t mind me asking, what reasons?

@s_ben

pseudonymous participation is important, especially in DAOs. Something to keep in mind from a design perspective.

This is so important. I could dump a huge pile of links here to explain why, but a lot of them are fairly complex/boring. TL;DR: the ability to differentiate between A and B allows you to discriminate between A and B.

By applying the principles of least authority to data disclosure, users are in control and able to share the information they want. This prevents discrimination based on location, age, race, gender, status, wealth, etc… Also, if everyone does this, it creates a level playing field for brainstorming where all parties are treated equally vs just reading/replying to posts by community leaders.

To make this more concrete, here’s a great Wait but Why article that explains this phenomenon with pictures.

Now (assuming you read the article) imagine that each megaphone represented not just the media, but cred. Then imagine that cred is not just a fun rewards token, but a governance token that is used to establish reputation in a community. In this example having cred = power, influence, and access to credit, jobs, admission to schools, etc… (like the social credit system - fun infographic!). Furthermore, anyone who disagrees with those in power or goes against the norm would then be at a disadvantage. This incentivizes group think, stifles innovation, and creates a really effective dystopia. SourceCred is capable of amplifying all of this.

AFAIK the goal of SoureCred is to transparently recognize and reward contributors, but you cannot transparently and seamlessly recognize and reward community contributions if the people in those communities are afraid to say, do, and support the things they actually want. Optimizing for psuedoanonymous participation (via all data points, not just user names) reduces the ability for anyone to censor anyone else while also increasing the diversity and authenticity of ideas. This helps balance of power between those who create the cred algorithms and those who play the cred game. If anyone anywhere can earn cred, and thus influence in the system they are participating in, the system will more likely reflect the values and goals of that community as a whole vs any particular group. This key point governance, will (IMHO) be the factor that determines if SourceCred is used as a tool that empowers people or as a tool that suppresses them.

@s_ben

This is basically what I’ve imagined all along as the best way to pay people generally. Now that the cred is exponentially time-weighted, and with a decay that feels about right, I think a robust measure of current cred could be simply the addition of all cred over time. If we were starting from t0, I’d say just paying based on last week’s cred makes more sense to me. The total money paid out would then always match the total sum cred. If you wanted to reward older contributions more, you could maybe just decrease the exponential decay parameter?

In this example is the lifetime cred affected by if/when you cash out, or just a separate variable used for cred weighting that tracks reputation earned regardless of if you cash out your tokens? (this might be important because otherwise the system would favor those who have the financial freedom to accumulate cred while disfavoring the “workers” who need to cash it out to pay for stuff)

@s_ben

But since we’re starting in the middle, it feels like there should maybe be an “airdrop” or something (another potential use of SourceCred)?

That seems reasonable. If the project has a history of rewarding contributors in a fair and transparent way it would encourage more people to contribute.

@s_ben

it’s never too soon to think about governance

By the time you need governance it’s too late. The system needs to be put in place before there’s contention. If everything was okey dokey we would never need governance because everyone would just get along. Governance exists to prevent and manage disputes. This is essential. The best time to work on governance is yesterday, but the second best time is now. @decentralion Maybe we could start a governance category in the forum to start working on this?

@decentralion

@burrrata is a psuedonym, so they can be our canary in the coal mine for ensuring that psuedonyms can use SourceCred just fine.

That is what I do best, albeit often at the frustration of everyone around me lol

@decentralion

For the CredSperiment, they will likely need to connect a bank account to OpenCollective to take dollar payouts, and as an administrator of the collective, I may be able to see some of that information. But it still wouldn’t be public. And this won’t be an issue in the same way for DAOs; also burrrata could just take mana payouts.

Oh… yeah this completely restricts my participation to mana payouts lol. Not sure what the latest iteration of the cred/grain/mana system is, but eventually these will all trade on decentralized markets right?

@decentralion

However, I do want to have the explicit “pay based on last week’s cred” term in the payout function so that people get some fast feedback, which is a lot more engaging.

YES! It seems silly, but the fast feedback makes things feel much more like a game and much less like “work” lol

@decentralion

I agree that we should be thinking about governance early, and planning for how to migrate out of benevolent dictatorship and into collective governance. Having some kind of cred-weighted voting scheme around the parameters seems promising. We’ll need to think about what kind of weird factional dynamics and gaming would emerge, though.

Yeah we actually want to try out cred weighted voting with some Aragon experiments so happy to test drive that and provide feedback. Also a governance category would help to start organize ideas and conversations around this

@s_ben

Have some governance ideas, but will save them for another thread.

Looking forward to hearing your thoughts. Please do share! (in another thread)