# Collected things ###### tags: `gitcoin` `Notes` --- would be interested in how we decide what communities to plug into this analysis. i think that collections ( https://gitcoin.co/grants/collections?featured=true&collection_id=14 ) might be an interesting way of grouping grants for this analysis is it easy to make this self service, ie @frank@gitcoin.co or i want to analyze a group of grants, we can just run the analysis? Show less Michael Zargham Michael Zargham 2:49 PM Today probably need a few iterations before trying to make it self service -- the first pass we simply picked some algorithms and described what we saw -- to make this reusable we need to evaluate some alternatives and understand what kinds of results are robust to changing the community detection algo, and the params of that those algos. Show less Michael Zargham Michael Zargham 2:52 PM Today after some of that evaluation/iteration scoping a self-service inspection tool would be viable -- starts as an internal tool and maybe becomes a product feature eventually ("CLR explorer"); disclaimer: i am not recommending we use our time that way, just pointing it out as a thing that could be done. R&D-> PoC-> internal application --> external application. (we're usually just R&D and PoC) --- i'd really like to see someone who has a large twitter follower count in this analysis, to measure the impact of "popularity" on gitcoin grants. we often are criticized that its just a popularity contest. also ETH2 researchers - since the stated goal of gitcoin grants is to fund infrastructure and these people are actually building the infrastrucutre --- love this. 1. maybe worth noting that we already have human flagging system, and we've caught colluders in the past with that. 2. i think what youre suggesting is an algorithmic flagger. is it too early to suggest algorithms + parameters for that algorithmic flagging? Show less Michael Zargham Michael Zargham 3:01 PM Today yes it is too early. this is the right "architecture" but the specific of combining human + machine intelligence to accomplish a goal (and to do so in a way that is transparent and fair) is exactly the kind of algorithmic policy making challenge i've been writing about with @kelsie.nabben@gmail.com -- we need to start by educating people about this in and around a community they are deeply vested in (eg gitcoin). this starts we being really careful to clarify what policy making problem we are addressing, why it matters (to them), what it means to take it seriously including better understanding the system itself before trying to hard to intervene. interventions should be intentional, and it needs to be possible to evaluate the intervention itself (including but not limited to its outcome -- keep in mind that how we measure the outcome itself is subjective, which makes this whole process challenging -- and again why it need to be treated as policy making -- rather than just 'optimization' -- a common theme in my talks is "is optimization really objective if the metrics you are optimizing for are subjective choices encoding trade-offs" --- fwiw; weve done a bit of this in the past RE: pairwise matching https://ethresear.ch/t/pairwise-coordination-subsidies-a-new-quadratic-funding-design/5553/24?u=owocki --- we don't need to include this in the article but i popped it in to make a point about brightid driving usership from many different language groups which i think really fits with the idea that "new communities" might look like attacks when they are new before they have a chance to begin to integrate over time. Show less Jeff Emmett Jeff Emmett 2:52 PM Today Cool, will give BrightID a shout out for their role in identifying and acting on/understanding these kinds of graph anomalies, but will prob drop the image. Kevin Owocki Kevin Owocki 2:54 PM Today image feels awkward where it is, idk if thats the best way to drive the point zargam mentioned in comments. fwiw, you could just add a paragraph about new communities might look like attack" and ways to mitigate that. Michael Zargham Michael Zargham 2:56 PM Today @jeffemmett@gmail.com i don't think we need the image -- i dropped it in more to make the point to you all -- feel free to cut it out once the point is made in the text.