Data Explorer Mission Recap

BACKGROUND

From October 19-21, the School of Data (a collaboration between Open Knowledge Foundation and Peer 2 Peer University) held a Data Explorer Mission focused on Garment Factory Data–collecting it, collating it, mapping it, and telling a story with data. This course had several points of innovation and experimentation, including:

• Time: instead of a longer online course, this was more of a “sprint” held over a weekend.

• Tracks: Learners chose one of three different tracks: (1) geocoding garment factories, (2) visualizations to show supply chains, or (3) connecting the dots between factories and brands.

• Trainings: Over the course of the weekend, Liana Foxvog (International Labor Rights Forum) Nick Rudikoff (Warehouse Workers United) and Jasmine Du (GIS Expert) ran trainings via Google Hangout.

• Tools: We at P2PU have been prototyping and testing our Unhangout tool (https://unhangout.media.mit.edu/) to spawn small conversations around a main video event. Think an unconference, but for video.

WHAT WAS ACCOMPLISHED DURING THE EVENT

• Geocoded the bulk of a list of over 2000 factories from the BGMEA using CartoDB and OpenRefine;

• Made data from the Alliance for Bangladesh Worker Safety available as machine readable data;

• Consolidated the data we have into a central repository;

• And began prototyping some of the data into visual presentations

Read the full blog post from Chad Smith and Matt Fullerton here: http://schoolofdata.org/2013/10/29/findings-of-the-investigation-of-garment-factories-of-bangladesh/

 

FEEDBACK AND LESSONS LEARNED

Both P2PU and OKFN folks huddled after the Data Explorer Mission to reflect on what we’d learned. We’d like to take lessons away from this experience in order to scale our learning experiences beyond offline workshops and appeal to a more diverse audience. Also, offline experiences are labor-intensive and not everyone can come to them, and we’d like to maintain the quality of our offline expeditions. Finally, we wanted to understand to what extent the learning events scale, and find out if team learning can happen asynchronously.

GREENS (Things that went well)

• The trainings & tracks. They were interesting and useful.

• Group projects where people could break off a piece and do it together. Track 3 did this well: https://trello.com/b/G0NNzqCx/bangladesh-garment-industry-database-and-mapping

• A handful of volunteers connected and decided to follow up on the investigation beyond the data expedition.

• Some actual work was accomplished

• People elected to choose themselves tasks and it was easy to see what they were working on.

• Recruiting trainers and having meetings together beforehand. These were great, and I plan to implement in the future.

• Unhangout as a mission control. The tracks and participants in them were clear and easy to use.

• Facilitation. These folks were great, supportive, took interest in people’s projects.

YELLOW: (Questions)

• Was this a learning experience or a data project? Most of the participants were well-versed in data wrangling, so it’s difficult to tell what was “learned.”

• The difference between Hangout on Air and using Unhangout as a way to simply organize synch hangouts was unclear.

• Diversity. The audience was pretty much NGO/data folks who already knew what to do with data. They were helpful to n00bies, but it was difficult to work on the bigger projects without presupposed knowledge.

REDS (Things that could be improved)

• Our shiny new tool had some hiccups to work out: definitely need to use calendar-farmed hangouts; our one big technical issue was with people getting rejected from hangouts: https://github.com/drewww/unhangout/issues/189

• Feature request: chat history for folks who are jumping in to help orient themselves.

• Attendance was low. There were more “insiders” than outsiders/new to data. Perhaps because the ask was “get involved with a problem” as opposed to “Learn data wrangling in the service of solving a problem.”

• Need a clear agenda of the day in the lobby (or linked to from the lobby)

• No good introductions to people entering first time.

• No clearly announced scheduled of catch up sessions.

• Promotion could have been completed earlier.

• Content was important to a particular slice of folks–less interesting to folks outside the NGO world.

• The more complex a topic, the more time is required. Thus a weekend was likely not long enough for the tasks presented.

• Team coordination was a real challenge and a more formal appointment of volunteer would have added some structure for arriving newcomers.

SHOUTOUTS: Who Was a Data Rock Star?

• Anders did an awesome job recruiting volunteer facilitators

• It was great to connect the online expedition with the offline expedition in Brazil – thanks Milena

• Chad Smith and Matt Fullerton were really helpful (these should both be mentor candidates)

• The learning environment setup was valued for NGOs

IDEAS FOR NEXT TIME

• A bit more like an IRC channel than a real offline event: how do developers work to get something done…

• Mass events with basic training OR smaller events (e.g. 10-30 people) which are actually investigations.

• Make more engaging and inviting for new people, ie: the “Show How Your Country is Unique with Data” Philipp helped design.