Open blueridger opened 3 years ago
A better tool for comparison: https://observablehq.com/@sourcecred/grain-config-simulator
This looks like what i was expecting, honestly. Newer folks who have been spending more time in SC are seeing ~2x what they've otherwise been receiving
One thing to note is that 0.6 still catches a decent amount of contributions from up to 3 weeks back. Here's a chart showing the curves for the 0.16 and 0.6 decay.
There is also some amount of 'smearing' if I understand correctly between epochs. In Timeline Cred, Cred minted on a contibution would take 3-4 weeks to exponentially decay to negligible amounts. If in CredRank, Cred minted 'smears' have a similar effect, then a non-negligible amount of Cred earned in the last week will be leaking into previous weeks?
Yet another complicating factor could be variability in Cred inflation? Below is a chart showing total Cred created per week, calculated from data from the Cred and Grain CSV Downloader. It seems the Cred inflation has been pretty stable the last few weeks (not counting April 10 partial week, which I'm assuming Grain is not distributed for). However, since you're distributing a flat amount of Grain, presumably you'd need to normalize contributors' Cred earned to more accurately compare contributors' % Cred earned per week.
If we're achieving the main objective (boost newer contributors' pay), then it seems this is achieving that. If we want to have a better understanding of what's happening with the scores, perhaps we could try a more extreme case that should be easily predictable? Set decay to 1 (equivalent to Immediate policy) and see if it corresponds to expectations?
I am curious as to how important it is for us to make such pretty significant changes to our foundations in order to, what, placate new people and make them see more instant gratification? What is the value-add for us there?
Note that "incentivizing new people" also means potentially incentivizing gaming with newly created accounts.
pretty significant changes to our foundations
Not sure what you mean by this. This thread is exploring a 1 line config change, not a code change.
The value add is fairer compensation for newer contributors. Working 2+ months before achieving payout equity can seem pretty extreme. It also incentivizes old contributors to stay active, though with the downside of sharper pay cuts if you take a week or two off.
I'm not particularly concerned about gaming right now, since our platforms are pretty well moderated and there are several manual, reversible checkpoints between cred-flowing activity and money leaving our treasury. I definitely wouldn't want to miss out on opportunities to improve pay equity and help the newcomers we do want to fold in, just on account of the always-present dice roll of whether gamers are going to try and get past our many-layered moderation.
I haven't come to a strong opinion either way.
I tried seeing what it would look like if we increase the recent policy decay to make it so newcomers can have faster access to income. The results aren't fully what I had imagined:
0.16 decay (current)
0.6 decay (arbitrary higher number)