Data Science @ GitcoinDAO

[nollied]

it’s interesting you post about this, i was just thinking about the feasibility of launching an AI workstream, similar to moonshot collective.

@owocki

perhaps one day even simulating agent-based contributor behaviour

this is what the matrix squad in FDD is working on btw.

@simba-sandy

Have a dedicated analytics person working with every team- MMM, DAOops. While central teams have their advantages, core teams will either always be stuck in prioritization queue or forever unaware of how data can help and wont be as impactful as they could have been with data.

i think this point should be wrestled with a little more. one interesting benefit of having a dedicated AI workstream would be that these services could be provided, with the goal of creating data unions/revenue streams (carefully of course to avoid negative incentivizations).

if we did have an AI workstream, it should play a supportive role, facilitating collaborations and acting as a service to the DAO.

@jonas

A powerful link would be data to be set in comparison with grants data from Metabase to compare and identify new opportunities for growth.

this is another benefit, the ability to cross-correlate data for insights.

@jonas

This also needs to align with the Web 3.0 ethos to collect as little data as possible

and for the data that we do collect, the reports should be openly available.

@nategosselin

In my mind, #1 should be a data-focused product team with the remit to build a scalable data warehouse/system that can make data more easily available to analyst-level data users. In other words, how do we knit all of our data sources together and let SQL users get value out of it (given they have the business context)?

beautifully put.

[tigress]

@owocki

1. What should the data science practice at Gitcoin look like?
2. If Data Science was an area of practice at Gitcoin, what would it look like?
3. How could it span multiple workstreams or squads(teams) & cross-pollinate between them?

Having Data Science as an Area of Practise or sometimes also called Community of Practise in larger web2 IT organizations is a common pattern. However, this DS-Area would not do any particular work for projects, OKRs or initiatives. Their goal is to reflect upon and improve their craftmanship by ways of learning from each other and presenting or socialising at events. If they get really productive they might develop and provide trainings for other Areas. For example: enabling developers so they learn from Data Scientists.

There is this idea of “Team Topologies” out there that could be helping us design and visualize how we work together. And also discuss several options how Data Science could look like and interacts with others.

Beyond the Spotify model - Team Topologies - DevTestNorth - 2019-09-25 -...

[Fred]

@ivanmolto from Builderband helped the MMM Workstream create a Dune Dashboard with an overview of the GitcoinDAO governance and finances. These visuals greatly increase the transparency of our DAO - and they look super cool!
Feel free to reach out with feedback.

[umarkhaneth]

100% support this

Giving people access to quality data, with a data dictionary, and the ability to use tools like SQL and Python will probably drive them to generate insights on their own. Data-savvy contributors who want answers will look for them – they just might need the infrastructure/access to do so.

This is what AWS is really good at – although a centralized technology – it makes it easy to have your data somewhere people can find and run more advanced analytics on using tools like SQL, & Jupyter Notebooks. It also allows for granular role-based access control. I don’t know what the decentralized alternative is but would love to

[DisruptionJoe]

Great work here. I’d love to see the number of delegators each steward has delegating to them as well. Something like this:

[Fred]

Number of delegators has been added to the table. @ivanmolto have also been working hard on an additional deep dive into our Governor Alpha set up.

Some very interesting data can be found here: Dune Analytics

[owocki]

@Fred

These visuals greatly increase the transparency of our DAO - and they look super cool!

was just messing around with breadcrumbs.app tonight and was able to come up with this sankey diagram that allows ppl to follow the flow of GTC out of the timelock + to the workstreams + to their contributors. pretty neat!

ivan, would love to show u breadcrumbs.app sometime.

[Fred]

This is cool! We have discussed creating a subgraph for the Steward Health Card-project as a potential metric of engagement. The theory being that Stewards receiving GTC less than 4 hops from the multisig is a strong indicator of engagement. Especially if it’s occurring month after month.

[ivanmolto]

Yes, of course. I would love it. I didn’t know about this tool. It is a result outstanding visually!

[Pop]

this is great - I would love to use some of this for the workstream and treasury health dashboard to be added as DAO tools alongside the steward health cards

[Fred]

@ivanmolto and I have been looking into the correlation between GTC outflow from GitcoinDAO and GTC spot price. A common belief is that Workstreams compensating contributors with GTC would somehow negatively impact the price of GTC. We put this theory to the test through a number of queries in this Dune Dashboard to see if it holds any water.

In order to produce a meaningful analysis we couldn’t just track GTC outflow from Workstreams and compare that to GTC price. To accurately test this hypothesis we need to take one step further and track GTC outflow from contributors in relation to price.

A common conception about contributors receiving compensation in GTC is that they will immediately sell their tokens for other assets, causing a massive sell wall and negative price movement as a result.

After analyzing 269 unique contributors over time we can see that this is far from the truth. The data show that contributors of GitcoinDAO are likely to hold their GTC and the average amount of GTC held by this group is increasing. The first figure below shows the number of contributors who hold more than 0, 100 and 1000 GTC respectively. The second figure displays the amount of GTC held (x100) in total by this group as well as their average GTC holdings. These metrics are all increasing.

We could not find a correlation between GTC outflow from Workstreams and GTC price movement:

As we looked further into what could impact GTC we found a correlation between GTC and other governance tokens. In general, GTC seems to be following the movements of the market sector. The figure below show the price movement of other governance tokens and GTC, relative to their price on May 25th 2021:

In conclusion; the hypothesis that sending GTC to Workstreams who then pay contributors would negatively impact GTC price can not be proven through the data. GTC is however correlated to similar tokens in the sector.

We encourage you all to look through the dashboard and we welcome feedback on ways to improve these further. A couple of additional visuals and tweaks will be added to the dashboard in the coming days.

Enormous shoutout to @ivanmolto for the amazing job.

Our two previous dashboards can be found here:
GitcoinDAO: Governance & Financial overview
GitcoinDAO: Governor Alpha & Timelock

[ivanmolto]

The same to you @Fred It has been an amazing job side by side with you. Please all you enjoy the insights!

[Lunacat]

Great to see some data! Thanks for taking effort and time to put together.

After analyzing 269 unique contributors over time we can see that this is far from the truth.

Does this mean data set is limited to these 269 individuals over the entire time frame? So any contributor that may have joined in Feb 2022 would not be counted?

In conclusion; the hypothesis that sending GTC to Workstreams who then pay contributors would negatively impact GTC price can not be proven through the data.

Assuming the population is constant throughout the period, the conclusion is undeniable. But concerns right now are more forward looking – it’s easy to hold an appreciating asset, especially given most contributors have likely been sitting pretty from wider market during this time. Will be interesting to revisit this after each round and see how trend develops in response to wider market turmoil

[Fred]

I could have clarified the “269” number a bit better.
269 is the current number of total contributors we are able to identify on-chain.

On May 25th 2021, when GitcoinDAO launched, we had 0 on-chain contributors. Now we’ve reached 269. A visual for the number of total contributors over time can potentially be added to the dashboard for clarity!

In my opinion we have seen some pretty big market turmoil in the last months but I absolutely agree, it will be very interesting to see how things evolve. Thanks!

[kishoraditya]

This is great @Fred and @ivanmolto
It would also be great to see if or not rounds exhibit any trend in pricing of GTC!

[Pop]

Totally missed this in March - what an interesting insight into representation

[Pop]

Great work @Fred and @ivanmolto - It is always important to look towards data, especially in a highly charged dynamic like a market downturn. It’s easy to make assumptions yet we must always keep ourselves in check to ensure governance decisions are not made emotionally but rather objectively and from a fully informed place of intention.

When we were discussing this and even though I had an inkling the contributor sell could not affect things as much, I really did not grasp just how much hodling was actually going on. The same with the pattern of other gov tokens which is why I made the suggestion to add that chart in order to give place this analysis against the broader market background.

I am keen to see these analyses progress and become a tool in the gov process. Perhaps, something we look to include in the improved versions of workstream accountability flow or in any voter matrix we may choose to adopt - provided they achieve balance between objectivity and personal takes.

[myceliumcoordinator]

@Fred

In conclusion; the hypothesis that sending GTC to Workstreams who then pay contributors would negatively impact GTC price can not be proven through the data.

The inverse is also true! This analysis does not show that selling tokens to market DOES not create downward price.

@Fred

We could not find a correlation between GTC outflow from Workstreams and GTC price movement:

How can you draw that conclusion from that graph? You just showed a bunch of data that has a bunch of outflows and a large decline in price. There’s no rigor to this analysis method. There’s no control group, nor is there any analysis of the liquidity of the market in this analysis.

@Fred

In general, GTC seems to be following the movements of the market sector.

@Fred

GTC is however correlated to similar tokens in the sector.

GTC is down 50% vs ETH over the last 30 days, so I dont think you can say the outflows are in line with the rest of the market decline.

This chart is from coingecko:

I think doing an analysis about this could be important, but more rigor is needed to be able to form conclusions backed by data in my opinion.

[umarkhaneth]

Hey! I really enjoy the discussion here. I would love to get an hour on the calendar to pull together various people with data backgrounds or interest within the DAO. We could do some introductions and share projects we’re working on.

Please fill out this lettuce meet if you’re interested: LettuceMeet - Easy Group Scheduling

All are welcome!

[php]

Maybe this way:

the practise only deal with data anonymisation, and data infrastructure, and model performance evaluation.

With anonymised data, all the rest can be put as bounty so be publicly worked on by any contributors