# MetaCartel Grants Analysis

## Introduction

The goal of the analysis is to understand what are the key themes of the types of projects that MetaCartel funds. This analysis is meant to be somewhat repeatable, so it could be performed on future funding rounds, or on corresponding data from other DAOs.

In order to understand what types of projects MetaCartel funds, we also need to look at what kind of projects it *doesn't* fund.

Thankfully we have a lot of great information available about each project on the forum posts. It's not really in any standardized format, but the text description of each project is a great starting point to work with.

## Data

I wrote a Python script to scrape the [MetaCartel forum](https://forum.metacartel.org/). I wanted to get the full project description from each post in the forum that was a request for funding.


Then, I matched the forum data with the data in the [Airtable](https://airtable.com/appWsfebn79bII12n/tblA2NXylnlVE9eIW/viw5H4Aaxv3XsyI0e?blocks=hide) tab called "MC Grants Proposal Portfolio" to get the list of funded projects.

In the end, I had a dataset with each projects' full description, and whether or not that project ended up getting funded by MetaCartel. In total there were 151 projects.

(Note: I didn't find a matching forum post for the following four projects: ['Overlay', 'MyMeta', 'A proposal for 5000 DAI to sponsor the NFT hackathon', 'Fork The World']).


## Process

I used Natural Language Processing techniques to build a model that predicts whether or not a project would be funded based on the description, using the words in the description as the inputs to the model. I used a technique called TF-IDF to represent the words as numerical vectors so that I could train the algorithm on the text. I used a type of Naive Bayes classifier as the algorithm because it has certain qualities that allow us to infer which words are most predictive of the outcome.

All that being said, the point of this was to figure out which words were most associated with getting funding. This should indicate common themes in what types of projects are funded by MetaCartel. The results were pretty interesting!

## Results

Here are some of the top words associated with each funding status:

| Funded | Not Funded |
| ------- | ---------- |
| creation | time |
| transactions | think |
| deployment | resources |
| provided | idea |
| tested | small |
| v1 | fees |
| contributors | validation |
| engineer | potential |

Let's look at some of these examples in more detail.

### v1


When funded, it was used in the following contexts:
* *... Plan: 1-month: Build **v1** product and onboard one community ($JAMM)...*
* *... Features of the **V1** mainnet release include:*...
* *Moloch **v1** was a simple, MVD for permissioned, grant-making communities...*

Project descriptions that contained the word "v1" were about 4x more likely to get funding. My interpretation of this is (1) MetaCartel tends to fund early stage projects, and (2) it is preferrable when these projects are well thought-out, and have plans for expected stages with different versions.

### deployment


When funded, it was used in the following contexts:
* *...the initial **deployment** operated as a proof of concept...*
* *...with this grant the key team will prioritize the completion and **deployment** of key elements...*

Project descriptions that contained the word "deployment" were about 6x more likely to get funding. Based on the contexts in which it is used, I think MetaCartel prefers projects that state that they will use grant money to actively deploy new features, and again, have clear plans for what they want to create.

### community

Although not a top predictive feature, another interesting word that stood out to me was the word "community." This word was used in 80 different project descriptions (over half of all posts), and those that used the word "community" had about 1.7x more chance of getting funded.


Examples of usage in funded projects:
* *...we would like to invite any and all **community** members who are interested in contributing to reach out...*
* *...make it easier for the **community** to manage its treasury...*
* *...we hope to work closely with the **community** to get feedback and support...*
* *...in order to meet the needs of a particular **community**...*
* *...maintain the dashboard as a neutral **community** public good...*

This aligns with MetaCartel's values: "We are people oriented and community first."

### fees

22 project descriptions used the word "fees," and none of these were funded.

Usage examples:
* *...we use metatransactions to fund tx **fees**...*
* *...organizing a voting process was tricky plus while dealing with small amounts of money we were losing out with each transaction due to **fees**...*
* *...mainly to cover the audit **fees**...*

It seems that it is not desirable for grant money to be put towards paying off transaction fees.

### art

I thought it was interesting that both "art" and "artists" showed up as words associated with not being funded.


Context examples:
* *...a generative **art** nft project with a focus on activism...*
* *I’m a jack of all trades master of none. I work as an **art** director and writer...*
* *...handle all product production fulfillment on behalf of our design partners so they only need to focus on their **art**...*

Either art is not an area of interest for MetaCartel to fund, or perhaps it is an area that has been overlooked. Or maybe there just haven't been enough good art-related projects proposed yet.

Lastly, I want to point out a few other words associated with being funded that seem more specific and could be of interest. (I don't know what everything means since I'm new to the field.)

| Funded |
| ------- |
| Solidity |
| L2 |
| DAOhaus |
| Mainnet |
| pet3rpan |
| ETHDenver |
| Gitcoin |

## Analysis/Conclusions

Overall, there are several salient themes here. Projects with clear plans, a well-thought-out path to success, and a focus on community were much more likely to be funded. Words like "tested," "deployment," and "growing" suggest the former, while "community" and "collaborative" suggest the latter.

However, having a good plan seems like an obvious requisite. What actually distinguishes MetaCartel specifically from other grant giving DAOs? It would be helpful to repeat this analysis for Moloch DAO and see if different themes come up in the types of projects they fund. This could help us identify if there is a funding gap, and where the primary differences lay. Some potential distinguishing characteristics include preferring earlier stage projects ("v1"), using specific tools such as "Solidity" and "L2", and being associated with specific events ("Mainnet", "ETHDenver").

So far, MetaCartel tends to not want to use funding for transaction fees and art-related projects. This could be on purpose, an oversight, or simply a lack of interesting relevant projects.

More investigation could definitely be done depending on what people think is interesting. See full examples of top 200 words associated with [funded](https://github.com/maia-adar/MC_grants/blob/master/top200_funded.rtf) and [not funded](https://github.com/maia-adar/MC_grants/blob/master/top200_not_funded.rtf).

## Links to Code

https://github.com/maia-adar/MC_grants/blob/master/Scrape%20MC%20forum.ipynb

https://github.com/maia-adar/MC_grants/blob/master/MC%20grants%20analysis.ipynb