Timescale matters
Right now I think Initiatives make a lot of sense to me as a supernode. But it doesn’t do well for this need to have “ongoing tasks”.
And TL;DR I think the difficulty to capture it lies in time.
Here’s my thought process.
Imagine this ongoing task: Managing the Twitter account. The way I would go about that is probably make a wiki that gives a description of this task. What we’re trying to achieve, some guidelines about how to do it, and who the current Champion is. Then probably we should have a log of the contributions. In the case of Twitter that would be the list tweets, who authored it, when they… wait a minute! This is just a Twitter plugin, we can add this to the graph! Right? Well yes, a Twitter account would fit very snugly into the graph. Perfect example of what would be a good plugin. What’s the problem then?
Granularity is the problem. We’re back to the situation where if you have a bunch of tweets in the graph. You either need to value individual tweets (too fine). Or the entire twitter account (too coarse). There isn’t a natural node in between that that’s just the right granularity.
Meet Initiatives; just the right granularity. Initiatives are exactly that size that has all the benefits @decentralion outlined. They’re a natural fit to this granularity issue, because they are larger and finite tasks. But even though they’re finite, the retroactive aspect is till important. I’ll get back to that.
Ok so what’s the granularity of a Twitter account? That depends on how long the Twitter account has been going for, in other words time. Obviously to answer the question: how do you rate the all-time value of the Twitter account? Is something you might be able to reason about if the account has seen a month of activity. You can go through all the tweets, get some stats about views and retweets and make a pretty sound argument why you think it should have a certain score. But if it’s 5 years old that is too much. You probably can’t remember 5 years worth of tweets at the same time. And putting a Cred score to that is going to be subject to all sorts of gymnastics our brains are capable of, such as vividly remembering things that never happened.
So here’s a suggestion. Why not group the tweets by time? I think the earlier point about being able to give a decent argument for a score for a month worth of tweets could be applied to any of the following months as well. We could just evaluate every month separately, couldn’t we?
Promised I would get back to needing the retrospect. Of course evaluating a month of tweets makes most sense when that month is over. But imagine this scenario. A few months in there’s a discussion happening and we have some different ideas about what the tweets are for and want to change the guidelines for it. You could imagine someone saying “the kind of tweets we did back in august are a good example of this and were really helpful”. Having a “Tweets in Aug '19” supernode would make it really easy to go back and give extra Cred.
Champions would make sense again too. The same granularity and time issues apply to Champions. I think from the Initiatives and Champions discussions so far, it’s safe to say that having a Champion increases the value of the Initiative beyond the sum of it’s contributions.
But saying who our Twitter Champion is for that 5 year old account sounds like it would cause a civil war, while I think agreeing on who the Aug '19 Twitter Champion was should be achievable without bloodshed 
.
The best time frame probably depends on what we’re talking about. Maybe weekly is better for high volume things. Maybe quarterly or annually for low volume things?
I think a nice side effect of arbitrary time frames is that they’re logical moments for a Champion to step down if they’re getting overburdened for example or if there’s a new candidate offering to help. For a quarterly sliced task, the end of the quarter is a great moment to decide whether you want to do this for another quarter.