For Data Sake – An insider’s view of a GovHack team

By Muhsin Karim

How we met

Our team of five were all seated at the restaurant before the launch of GovHack 2017. It was the first time we occupied the same space together. Over the next 47 hours, we would share a meeting room, staring at our Macbooks as we reused and mashed-together open government data to build something new and ideally useful. Failing that – movie-marathon!

Mabelle sat next to me. She and I had discussed entering GovHack over a year ago. We used to work together as part of an innovation team. I was the team data analyst, she was the team UX/UI designer. When the rest of our team would disappear in meeting rooms, Mabelle and I would swivel our chairs and chat. User experience and user interaction designers are highly sought-after in this digital design world of apps and websites. She must field calls from recruiters every week asking if she is happy with her current place of employment. A response of “No” would make the recruiters drool.

Seated in front of me was Marcin, the team developer. “Front end”, specifically, which in my limited knowledge meant he built anything the user could see or interact with. Marcin worked at my company, yet housed in a separate building where the developers coded with minimal exposure to sunlight. He and Mabelle used to work with each other. Mabelle would provide the design images and describe product functionality then Marcin would incorporate these into the product build. Marcin had prepared for GovHack for a few weeks leading up to the event. His enthusiasm overcame the flu that had struck him the day before and he sat sniffling but excited for the weekend.

At the end of the table was Charmaine. A year and a half ago I spoke at a data scientist Meetup group. I described my then journey from “data analyst” to “data scientist” and asked the audience “How do you know when you’re a data scientist?”. I had hoped a gremlin would materialise before me, throw confetti made of shredded textbooks at my face and declare “Thar be a data scientist, matie!”. Charmaine approached at the end of my presentation. “Can I contact you later and ask you about your data science journey?”. Yes, I said, happy to. We exchanged emails and I learnt that Charmaine was in the midst of a career transition – from a postdoc medical researcher to data scientist. A few days leading up to the GovHack launch, she messaged and asked if she could join our team.

Next to Marcin was Ellecer who I was meeting for the first time in person. Marcin was afraid the flu would keep him in bed, and his former colleague was looking for a team and could substitute. Ellecer joined as our “back end” developer. I assumed he was responsible for the things the user could not see behind the app or website, which made his profession sound mysterious. From my perspective, I consider devs as wizards – they code and create functional products where once there was none. Ellecer said he had permission from his wife to dedicate the weekend to the hackathon. He made us a team of five – team “For Data Sake”. I was proud of our team name. A name born from thoughts fuelled by white choc macadamia cookies.

We cannot leave until we have an idea

An hour after the launch, we sat in a meeting room at Charmaine’s offices at the University of Sydney. I was grateful we could use the space. The crowded Sydney hackathon location at Fishburners held teams huddled together all hitting the GovHack website at once, slowing the Wi-Fi to a crawl. We had the room, a whiteboard, chocolates and treats and an oversupply of chairs that were perfectly in the way when we walked around the table.

We reviewed the datasets made available by various National and State Government agencies. “I’m feeling uninspired”, said Charmaine as if she lifted the thought from my mind. “I thought I’d have all these ideas but I can’t think of one”.

“I feel like there are two broad types of projects”, I said. “The first is to take a dataset and uncover interesting insights. Like the Opal tap on tap off data. We could highlight peak times and locations. But I don’t know if that’s useful knowledge for a user. Even if you knew when there was congestion, what could a person do about it? The insight could be meaningful once we started joining it to other datasets, and we can show some interesting correlations”.

“Yes, but anything can be an interesting correlation”, said Ellecer. He sent the team a link via the messaging app Slack showing the spurious correlation between the number of people who drowned by falling into a swimming pool and the number of films Nicolas Cage appeared in.

“Agreed”, I said. “The other type of project is that we build a useful tool around the data. A product that the judges could imagine the public engaging with. I’d prefer to build a useful product”.

We bounced around ideas and quickly refuted them. We could use the Opal travel data to uncover where most tap on and tap offs occurred in the morning then in the evening as a measure of where people lived and worked. What if we built an app that compared a user’s home and work suburb? If we merged census data, we could compare demographic features.

“I think you’re onto something with comparing suburbs”, said Marcin. “It would be cool if we could give each suburb a measure based on liveability”.

“At the very least, we can count what sites and services each suburb has and rank them. The more parks, the higher the score”.

Marcin widened his eyes. “Yeaaaah!”. He walked up to the whiteboard and started to draw rectangles within a rectangle. “We could have this card and the user sees the suburb they’re in and it displays all the suburb’s features”. I felt relief as we progressed from discussion to making our thoughts visible on the whiteboard.

“You all saw the data that the ACT released. They have locations of skate parks, barbeques, beaches. They’re practically asking for an app that lets users see what’s around them. Here, I’ll list out all the datasets.” I held my laptop and scrolled down the screen of the supplied ACT data and wrote each site or service on the whiteboard based on the available datasets”.

“Woah, there are so many”, said Charmaine.

“And I’m not even halfway through the datasets yet”.

It was nearing 11 pm and we settled on this: Our app would be targeted for people living in the ACT or visiting the territory. They could enter a location and see what was available in the area. “We need to run this by Mabelle”, I said. Mabelle had left earlier. “She’s the best person to think about the target audience. I’ll make sure I can load and join all these datasets tonight and we can discuss it tomorrow”.

On Saturday morning, we discussed our tentative idea as the team gathered. The app could allow users to view an ACT suburb and have all the sites and services listed as a profile. Surely there was a similar concept out there in the world. There was. We looked at the Microburbs site. The site created a suburb report, pulling in various features including commute times, cafes and school locations. The report also included various liveability scores.

My heart sank. We concluded the previous night believing we had an idea. We were not keen on building a poorer version of Microburbs. What else could we do? We could still map everything from the various ACT datasets. How was this different to Google Maps? We searched for ACT skate parks on Google Maps and a number of pins appeared. Then we included barbeques. A number of pins disappeared leaving a few remaining. I wasn’t convinced that Google Maps could display multiple sites and services as we desired. We needed to press-on so we clung to this general idea.

“Who are our users?”, asked Mabelle. Who would use an app that could tell them what was around their suburb? We thought there could be two types of users, residents of the ACT and visitors. We refined the users down to those who did not know what they wanted to do in the given moment. People who were on the move could take out their phones and use our app for inspiration.

“Is that something you guys could build?”, I asked our devs and designer.

“Oh, hell yes”, said Marcin.

Getting to work

“OK, let’s think about the app from the user’s perspective”, said Mabelle. “Do we have post-it notes?”. I had “borrowed” some from work. Mabelle wrote “User assumptions” and “Business assumptions” on post-its and stuck them to the wall. I stacked chairs to clear some space. “Can everyone write down what you think our users will want from our app?”.

“What do you mean?”.

“Just write anything that comes to your mind. I’ll be asking you questions centred around the user and based on our assumptions we can design the functionality of our app”.

We wrote down answers for each of Mabelle’s questions. I collected post-its and helped her stick them on the wall. She rearranged the post-its into themes to understand our users and how the app could serve their needs.

“I’ve never done this before”, said Charmaine.

“My boss loves this stuff”, I replied. One cannot have an innovation team without post-its and markers. The post-its started to peel away from the wall. I guess we cheaped-out on the post-its at work.

Mabelle and Marcin were at the whiteboard drawing designs and discussing the app’s look and functionality.

“What should I be working on?”, Charmaine asked me.

“Well. Those guys are their own team for the moment”, I said looking over at the other three. “I have some data wrangling ahead of me. There’s a bunch of admin work that needs to be done, like filling out our team details on Hackerspace. But you shouldn’t have to do the majority of the admin. It would be good if you could do some data work at some point”.

“No, I’m happy to do whatever”, said Charmaine.

“What’s the name of the app?”, asked Mabelle, “I’ll start on the logo design”.

I got up and joined her at the whiteboard. “Um – I’m just going to start writing things that pop in my head – it’s better to see it than just say it”. I started to write. “ACT on this!”. “Around ACT”. I was reading each one aloud. “ACT About”, which I read as “A-C-T about”.

“ACTabout”, squealed Mabelle with a big smile. She had read it as “Act about”. We high-fived to make it official – the high-five being the lawful gesture permitting us to decide the project name on behalf of the team.

Over the next few hours there were long stretches of silence. We fixed our eyes on our Macbooks as we each contributed a piece to the final product. A product which was not 100% defined. The app was being built in parts and we all worked with the assumed knowledge that it would fit together and just work. If we believed otherwise, we would have gone to the movies.

We broke the silence with questions and requests for help. I asked Marcin about the required format for the ACT data. Ellecer would then upload the data to his database. There was frequent discussion between Marcin and Ellecer about software and approaches. I could see Mabelle’s eyes over the large monitor as she worked on her designs in silence. Charmaine and I would chat about how to best describe our project. However, without seeing the end product it was difficult to describe. The “ACTabout” app would include categories such as “Arts” and “Amenities”. Based on the user’s current location and category selection, they would be directed to the nearest relevant site or service.

On Sunday morning Mabelle messaged that she was unwell and would be working remotely from home. She would be editing the three-minute video showcasing our work as part of our submission. I worried about how we would work on this together with Mabelle not being physically present. We would figure it out.

In the meeting room, Charmaine filled me in on the discussion she and the devs had the previous night. “We thought that since our users are those that are not sure what to do in the moment and who are out-and-about, they could be inspired by our app with suggestions. So, we renamed the categories”. The new categories included “Be inspired” with arts facilities and public art displays, “Be wild” to find past location sightings of wildlife, “Be smart” for schools, TAFEs and libraries and “Be safe” for hospitals. Mabelle and I really liked the new category names – they were more active than the prior names and suited our app’s function for ACT users who were out-and-about and needed inspiration. I updated my code to include the new category names. Mabelle, on speaker phone, said she would create new category tiles for the app.


Crunch time

I was eager to work on the script for our video with Charmaine. “You really should see what the guys have done with the app first”, said Charmaine. “It will help you write the script”. I asked Marcin if he had something to show me. After a delay, he gave Charmaine and I a quick tour of the current state of the app. The designs from Mabelle looked great. The user would see the categories for selection. They would click on a category then based on the user’s current location, the next screen would display the sites or services ordered by distance.

“Are we able to see the map? Pins for each site?”, I asked.

“Ahh, no”, said Marcin. “I’m having some issues with the maps”.

We wanted a demonstration as part of the video. A quick run-through of how the app would work in practice. Charmaine and I took the script into her quiet office to start recording the voice-over. I wanted to insert jokes into the voice-over. Earlier we had shared the hilarious “Not Hotdog” scene from the show Silicon Valley. I desperately wanted to have Ellecer provide the voice-over in a lackadaisical deadpan manner like the Silicon Valley app-developer. He was not keen. Instead we would try to persuade Ellecer to cut-into the video to oddly ask if he was able to find snakes with our app since “ACTabout” held data on wildlife sightings.

After Charmaine had completed her recording and sent files to Mabelle to edit, we requested Ellecer’s contribution to the video. “We just need you to say something like, ‘What about snakes?’”. I recorded Ellecer at his laptop, looking up and saying the line multiple times. I kept ending the recording too soon and made repeated attempts. Charmaine was laughing. I was laughing. Ellecer was professional and I wondered how many times he had been asked to do this sort of thing. Marcin had a tense smile on this face and requested we keep it down. He needed to dig himself out of a coding hole. We selected the best Ellecer videos and saved them for Mabelle.

Mabelle created a draft video. Charmaine, Ellecer and I gathered to view her work. We loved it. She included music playing over her logo and designs as Charmaine’s voice-over described “ACTabout”.

“You’re a good narrator”, I said to Charmaine. “You should do voice-over work professionally”.

“You could work for the ABC”, said Ellecer.

“Marcin, are we able to get a demo of the app to include for the video?”, I asked.“Ah – give me a sec”.

Based on previous experience, GovHack recommended that participants upload their videos no later than 4 pm to make the 5 pm deadline. Uploading takes time. Mabelle said that it would take her 45 minutes to render the video after editing. That gave us a cut-off time of no later than 3:15 pm.

At 3:30 pm I made a call. “We’ve run out of time. If we can’t include a demo of the app then the best we can do is have some screenshots with Charmaine’s voice-over. Charmaine, we should start recording that now. Marcin, can we get some screenshots of the app?”.

“Hang on dude. Let me just hard code this”. Marcin switched to his text editor and entered some code. He switched back to the app then typed the name of an ACT location.

Marcin selected a category tile. The next screen displayed a list of nearby sites of interest. Since time was critical, that’s what we decided to display. A tiny snippet of how the app functioned.

We moved quickly over the next few minutes. Charmaine was set-up to record her voice as Marcin steered. He clicked on the app as Charmaine discussed the app’s features. We had a practice run, then a take-two. I waved my hands. “That’s it, that was great, we’re using that!”. I hopped back onto my Macbook to export the video for Mabelle. It was almost 4 pm.

“Oh man, I really wanted to show how the app could tell the user’s current location instead of having to type in their suburb”, pined Marcin.

“No time! Mabelle doesn’t have enough time to render and upload. It’s past four o’clock!”.

I saved the video and called Mabelle for the final edit. Marcin and Ellecer continued to develop the app. Charmaine and I scanned through the submission guidelines making sure we had neglected nothing.

Mabelle sent a Slack message saying the rendering was complete. She asked if we were able to download the video. I wrote back, “You mean upload it?”. A few more minutes passed. I called Mabelle. “Hey, how’s the video upload going?”.

“Upload? Was I supposed to upload the video? I sent it to you”.

Oh no.

It was 4:48 pm. We had 12 minutes to upload the video to YouTube. “Mabelle, can you upload the video now?”. She commenced the upload. “OK, bad team captaining on my part – there was a bit of miscommunication between Mabelle and I as to who was uploading the video”.

Just before 5 pm, Mabelle messaged with a link to the uploaded video. “It’s done!”, she said over the phone. “I was panicking so much that we would run out of time”. We gathered around Marcin’s laptop to watch our work. We laughed when Ellecer appeared asking about snakes.

“Oh, no!”, exclaimed Marcin. “The video is four minutes long!”.

We looked at the video length. We were not permitted to submit a video that was longer than three minutes. “We can edit it with YouTube’s editor and cut the time down”, said Marcin.

“Will the link to the video be the same? Because that’s the link we included on Hackerspace”. It was past 5 pm and I was unable to make changes to the submitted link.

“Yeah, it will be the same”. Marcin asked Mabelle for her password to log into her account to edit the video. We played the end of the video and made a call on the best time to snip the end. We waited for the changes to take effect. We kept waiting. It was well past 5 pm.

“It’s done!”, Marcin said. “Check the original link, it’s two minutes forty-six seconds long!”.

We raised our hands in the air in relief as Marcin told Mabelle that we were done. We successfully completed and submitted our first GovHack project with no minutes to spare.


Links of interest

Our Hackerspace project page with video.

The four-minute version with some thrilling dead-air at the end.

Spurious correlations.

Finally, “Not Hotdog”.