Below is an example of a filled out template of an “Ideation Doc,” which we’ll be asking people with ideas that they want to incubate to prepare prior to starting the Ideation Pod (we’re happy to help). It serves two purposes:
1) For people to show serious intent. Since the pod doesn’t pre-judge ideas on their merit (any Cabin-related idea is welcome), preparing such a document serves as evidence that the person really thought the idea and the 1-month goals through and is unlikely to waste own and Steward time.
2) For people to get the most out of the 1-month incubation. 30 days is such a short time, it’ll serve each idea best to have a plan of action in terms of focus areas, vision for success, awareness of what can go wrong, etc.
*Without further ado, here’s a blank template to use:
And here’s the template for the idea to scrape and interact with Cabin-relevant Twitter keywords and key accounts:*
Social Scraping and Interaction
Team so far: Dahveed and JonBo
Table of Contents
1-month goals
Expected expenses
Focus areas
Relevant Twitter accounts
Relevant Twitter keywords
LLMs
Squads
Content
What does success look like?
What could go wrong?
Looking forward
1-month goals
- Compile and refine a list of Twitter accounts with relevant tweets
- Compile and refine a list of keywords that catch relevant tweets
- Test LLM use to increase keyword and account relevance
- Start building Cabiner squads for each relevant keyword
- Define our strategy for creating pro-Cabin articles and guides to link to in our interactions
After 1 month, we will submit a proposal to the DAO for a 3-month funded trial so we can reward squad members for regular and engaging interaction with relevant accounts and keywords — measuring and analyzing the results.
Expected expenses
During this 1 month, we expect the following expenses:
- $100 for Twitter API access
Now let’s get into the details:
Focus areas
Relevant Twitter accounts
We already started compiling a list of accounts posting about urban planning, placemaking, and walkability. So far, the ones we’ve found are on point but have little interaction with their tweets. Of course, we want to find and interact with the ones that have hundreds of comments and thousands of views — but there is also an advantage to starting with less prominent accounts to build up our reputation and approach before getting into the “big leagues”. In that first month, we want to build a database of relevant accounts across a number of Cabin-relevant keywords.
Relevant Twitter keywords
At first, our keywords returned over 100 tweets per day per keyword, which is too much for anyone to actually interact with. So we’ve been refining and refining them. And it’s been getting better but still often hit or miss with keywords. For example, our early use of “loneliness epidemic” got some good results, but also a lot of politically charged tweets. “Block party” got a lot of tailgating and promotional events. We’re already seeing some success in refining our keywords. For example, one of the variations of our “neighborhood” group of keywords scraped a tweet that was a reply to this treasure of a thread:
But there is still a mountain of work to be done in refining keywords so that the squad’s time is not wasted in clicking on irrelevant tweets for hours.
LLMs
We haven’t even gotten to this yet, but - intuitively - it’s highly likely that LLMs can analyze the caught tweets and suggest ways to refine our criteria (and maybe even help find more relevant ones; same for accounts).
Squads
Squads are topic-based: parenting, placemaking, friendships, etc. — within Cabin we each have topics of interest, expertise, and relevance closest to us, something we’d most be interested and qualified to talk about. Squads serve a dual role:
- Break down the volume of interactable tweets into smaller, more manageable chunks
- Give anybody at Cabin an opportunity to meaningfully contribute
It will also be very helpful to have people in different time zones in each squad — because the attention span of Twitter is measured in minutes, and it’s beneficial to get into the conversation early to maximize viewership and responses.
Content
For this entire endeavor to serve Cabin, our interactions with those tweets should lead to something Cabin-related. This can be our website, forum, Paragraph, or specific pages (NAP, census, neighborhood directory). But that is a bit salesy. So it would be very powerful to have content specifically on those keywords that also happen to mention Cabin and maybe even lead to our site/other content. Basically: provide value first; sell as a secondary consideration. Savannah’s writeup in Supernuclear is to date the most effective example of that.
How to run this content creation? Maybe it should be under a separate Content Guild. Maybe it can be part of Eileen’s proposal. Maybe something else entirely. The goal of our first month is to come up with an approach and to put out feelers to those willing to run and/or contribute to it.
What does success look like?
After this 1-month period, success would look like:
- At least 3 keywords catching engageable tweets relevant to Cabin
- At least 50 good key accounts (i.e. with engageable tweets)
- At least one person (not named Jon or Dahveed) working on interacting with each keyword
What could go wrong?
Of the things that could go wrong or be potential problems, the most likely are:
- Failure to find keywords and key accounts with engageable, Cabin-relevant tweets
- Our engagement with the tweets leading nowhere
- Failure to recruit people into squads
- Technical issues with the API, LLMs, etc.
- Us not giving enough time/attention to this b/c of “life happens” stuff
Looking forward
As such, at the end of 1-month, we want to have clarity on which keywords and accounts to scrape, who will help us to interact with them, and the nature of those interactions.
Our proposal for the 3-month trial will likely focus on the deliverables of:
- Volume of interactions
- Traffic to Cabin online properties
- Squad growth
- Content growth
But let’s get there first.
