I put together an Observable Notebook with preliminary payouts for the first week of the CredSperiment. Please take a look at the documents; it contains an overview of cred, of the payouts, and some context.
Per the discussion here, we pay 80% based on total cred, 20% based on last week’s cred.
I still need to build a proper ledger for SourceCred as per @Beanow’s suggestions here, but I think that using the notebook for the first few payouts will be OK.
The payouts will be considered “canonical” when they are committed into the sourcecred/cred repository.
This is the first time I’m really looking at the weekly cred for any particular week (in this case, the last full week). It feels to me like @wchargin’s cred is pretty high? Likely because he sent in a flight of pull requests, each of which make a tiny, neatly scoped change. It’s a nice practice, and it shows again that we need to get more sophisticated than minting cred directly based on the node type.
I’m planning to make the notebook itself accessible to a broader audience, so that we can circulate it and build some enthusiasm about SourceCred. I’ll probably take another pass at it tomorrow with this in mind. All suggestions are welcome.
Scores are looking good to me. Only thing I would say is that just looking at the rows of the spreadsheet, the scores have lost some context/meaning for me. I like being able to see my line on the chart compared to others. But that may also just be me liking visuals.
Eh. Don’t know enough to make a judgement, but he’s an OG that’s done far, far more than I so np for me.
While the 80/20 rule has me making $7, I still like it. The long game of building an income stream (which continues to taper even if I take some time off) aligns well with my goals.
For a simple solution I would suggest committing the final output into a repo too. As archiving/replicating and tamper detection are much better understood vs notebooks.
You may want to include a stronger hash and commit that too, like sha256sum as git’s sha1 isn’t particularly great against collision attacks.
For a future ledger format, you may want to cryptographically sign the results, in order to provide authentication on top of tamper evidence.
In the notebook I noticed the payouts are based on a $250 budget. I remember a discussion about pros and cons of fixed bugets and at the time didn’t seem preferable. However that was before the slow & fast split and mana. Can you share some rationale for this number and what your plans are for the rest of the experiment?
Are there any caveats surrounding transaction fees? And / or options to defer payouts if it saves costs?
Yeah, my plan is to write a simple JSON spec for payouts, and commit the payout record to the repo.
That will be a good consideration to take in mind for the eventual productionized ledger system. For now, I’m not too worried about our records getting maliciously re-written (esp. since it’s permissioned and only @wchargin and myself are admins on the GitHub).
We talked about this a lot on our office hours call today. To summarize: there are some big benefits to having a non-fixed budget, and paying based on cred issuance instead. It:
avoids the feeling of “splitting a fixed pie”, and regretting when more people come. the pie should be growing
allows us to do “bounties” or collective incentives, e.g. if we make a release or merge an important feature, there’s more cred for everyone
avoids games around trying to contribute when other ppl are on vacation
However, it has some disadvantages. Right now, I don’t have great confidence that the absolute issuance of cred is super meaningful, and we might make weight adjustments that materially change the cred distribution. Also, it would be relatively easy to inflate the total cred right now (just post lots of stuff) and presents the risk that SourceCred spends its budget too quickly. So for now, let’s go with a fixed amount each week, with an intention to switch over once cred is a bit more mature.
For this week, let’s make the budget $500; I think $250 is a bit low esp. considering that @wchargin and I are both planning to take mana instead of cash. My intention is that we’ll pay out at least $500/week through the end of 2019; possibly more each week if the experiment is going well and budget permits.
By default, the payout doesn’t actually “pay out”, it just accrues into your balance, which can then be redeemed for cash or for mana. For transaction fees, we’ll probably have a minimum withdrawal, e.g. $5. It looks like we’ll have to pay 2.9% and $0.30 per transaction (PayPal fees) for the invoicing, which kind of sucks. We’ll need to think a bit about whether this is paid by the contributor or the project. Hopefully before too long we can add crypto payments for reduced fees .
Been quietly following along over here, thanks for laying everything out so clearly!
As someone with an ~$8 payout, I’m thinking about how to handle smaller sums like this in a way that doesn’t demotivate minor contributors. (This isn’t an immediate feature request, just long-term brainstorming!) It’d be cool if I could have a setting like “for sums below $XX/month, automatically take mana” (where $XX is customizable per user, and the sum is calculated across all projects using SourceCred where I’m a contributor). Kinda like reinvesting dividends. And then I can participate in projects without thinking about extrinsic motives, and only have to think about it as “real” money when it reaches an amount that’s meaningful to me.
My immediate thought is, hold out for a few weeks. This is just week 1’s payout and we’ve got something like 12 weeks left to go from there that are part of the experiment. Even if you can’t / don’t earn more cred during the experiment (though this forum post gives you some ), you’ll receive more payouts. You can let them accumulate and do the actual withdrawal later, saving a bit on transaction fees.
You’re right that mana is a type of reinvestment approach. However as of now, mana isn’t implemented or finalized in what it will do. One often talked about feature that would let you grow your stake like a reinvestment, is to use it to boost particular tasks or past contributions. For the project that is an indicator you believe this will be an important task / contribution, which may be reason to prioritize it and may grant you some returns if you end up being right about that.
However, because this is still very much up for grabs. I think defaulting to $ payouts makes the most sense for now. Saving up a while until withdrawals make sense in terms of tx fees.
So a few thoughts on your situation. Your current cred isn’t all that minor in relative terms. But right now is an experiment, $500/w going in at the top is a careful (training wheels on) budget, you can’t realistically hope to feed a dozen families. So as SourceCred matures and there’s more confidence in the system, one part is to fundraise and increase the budget. In the future if we can give such a budget a 10x, we’re talking a few paychecks worth going in at the top and would land you on $88 instead, for one week. Things like transaction fees are going to be less problematic.
Another noteworthy thing here is, most of your payout comes from cred you’ve earned around august/september last year. This is what the fast payout vs slow payout split does. To sketch a rough timeline. Last year, your work would have gotten you a fast payout first. Something on the order of $35 (assuming the same $500/w with 20% going to fast payout) in the week you did the work, then all those weeks between then and now you would have gotten slow payouts, starting in the 10s of dollars, and slowly decreasing over time as other people earn more cred and take more of the share. So over a longer period of time, we’re talking 100s of dollars over time. 10x the budget as mentioned before, now it’s 1000s of dollars. I think that’s in the respectable wages ballpark. Though I have to add, SourceCred as a project is still relatively small in that time period, this gradual shrinking of your relative share would go a lot faster if there were more people devoting time and earning cred.
What about the meantime? Something I find important is, between last years contributions and now, hopefully you’ve enjoyed working on a whole bunch of other awesome projects too , it’s what I would call being a “nomad”, contributing to a project for some time and moving on to different ones following your interests. Something I’d personally love to be able to do in the open-source space while being able to sustain myself. Maybe 10s-100s of dollars per week for one project isn’t significant enough to pay bills. But after roaming between projects for a while and building cred there, you might have 8-10 projects all paying you weekly in the 10s-100s of dollars region for multiple weeks. Now we’re getting somewhere!
Suggestion I’ve seen used in other places for folks who are intentional about donating their time/energy and may want that reflected: also offer options to convert to an EOY donation to a good cause. This way, folks who have a minimal output can join forces with other contributors to put that $$ toward a good cause. This may make more sense down the line when sourcecred (or a project using sourcecred) is fully funded by 1 or many organizations, but the I really like the idea of pooling small amounts of cred into larger bundles to become significant for an area or cause. =]
Currently there’s 2 ways to get mana: donate or contribute. My contribution has resulted in a $15.8 total payout for week 1. Does this result in the same level of mana as someone who donated $15.8 to SourceCred?
Nope. According to this notebook, at time of writing you have 3,081¤.
Yes, possibly modulo transaction fees (e.g. if you donate $10, and we get $9, I’m not sure if it should be 10¤ or 9¤).
It’s much cleaner to say that Protocol Labs (the big sponsor of the CredSperiment) is offering to buy mana at a rate of $1 = 100¤. This makes it more natural to think of mana as a digital currency, replace the project arbitrarily setting a mana price with a market, and have Protocol Labs support the project by offering to be a consistent buyer of mana.
However, we should study the legal and regulatory implications of conceiving of mana in this way. I intend to write a thread elsewhere exploring this in depth.
Hmmmmm this is deeply discouraging. I’ve spent hours reading about, writing about, and promoting SourceCred. This is because, obviously, SourceCred is awesome. That being said, if after all that I only have the reputation of someone who’s donated $15, it feels a bit wrong. Ideally early contributors would be rewarded with either something very rare and that has the potential to appreciate in value.
If mana is cheap and easy to get, then the “thrill” of contributing and playing the SourceCred game is is significantly dampened. TBH I’d be happier if mana had no monetary value, but was rare and coveted. Then I would feel special lol
Anyways, still happy to play because I was playing before there were rewards, but… what if?
there was a lockup bonus where the longer you waited to withdraw your mana the more valuable it got? (like a multiplier so it would increase by 20% for every week it’s vested or something)
mana earned via contributions was increased by a multiplier? (so if you buy mana you pay the market rate, but if you contribute to get mana you get “mana options” that allow you to purchase mana at a discount price)
buying mana was not a fixed price, but was dynamically priced on a bonding curve? (this way early contributors/investors could get mana at the bottom of the curve for cheap, and then as they contribute to SourceCred and drive demand mana prices would go up, aligning interests for all parties involved)
there were different payouts for mana vs cash? (so if you take mana you get a fixed amount of mana, not the cash equivalent - so as the value of mana increases the “value” of contributor mana payouts vs donations increases)
Likewise, @decentralion working on this approaching 2 years full-time, is worth more than $412.
You’re confusing Cred and Mana scores with the $ budget that has been put into the CredSperiment so far. Which is exactly the risk of tying it to fixed exchange rates.
How much has been payed out is just an arbitrary number that has no real influence on how the SourceCred algorithm values your contributions. When the budget is there to pay people more, your mana and $ equivalent will increase accordingly. For example as would happen with the $100k airdrop idea.
Yeah, @Beanow is exactly right — Cred represents reputation as a contributor, mana is sort of a measure of how much economic value you have in the project.
Right now, the project has essentially no economic value (we’ve only issued 100k mana) so someone could spend $30 and have as much economic value as you. But they won’t have any of your reputation. And if we do the $100k airdrop, you would wind up having orders of magnitude more mana than the person who only spent $30.
I think it is often harder to make meaningful contributions to a project than to throw money at it. If you agree, then you should be happy that you’re more “cred-rich” than “mana-rich”
Yeah, I think the bonding curve idea is definitely promising. I might prefer just having it be a market where for now we have a single buyer (or set of buyers) promising to buy at $1=100¤.
Oh good! I was hoping that was the case, but was not sure
Totally! Incentives should be aligned with value creation so this is very good
My main concern with the supply and price discovery brainstorm was that it felt like contributions were being undervalued if people could just buy their way in with lunch money and be on equal footing as community members. Now that the cred/mana (we’re calling it grain now right?) connection has been clarified, that’s less of a concern
but bonding curves are awesome! so that could/should def be explored in the future