Telling science stories with code & data

The live blog, from the MIT Media Lab

By Matt Carroll <@MattatMIT>, Saturday, April 28, 2015

Welcome to the live blog for today’s hackathon, “Telling science stories with code and data.” We’re tweeting at #hacksciwrite.

// ++ //

1:50 pm: People are intensely hacking on four problems. Here’s what they are focused on:

PROBLEM 1: Where is the solar-energy boom coming?
Yale’s Project on Climate Change Communication has surveyed public opinion to see where people support investment in expansions of solar energy. Meanwhile a US Dept of Energy project tracks and analyzes state-level incentives for renewable energy. Can these two databases be compared and combined to predict which metro areas might participate in a solar economic boom?

PROBLEM 2: Where goes the water?
The western drought has focused attention on water demand. Can we map data to show the split between agricultural, industrial and residential water demand crossed with expected changes in precipitation under climate change? California is an obvious focus, but it may be more informative to look at the response of demand to previous droughts, say in the southeast US.

PROBLEM 3: Secret studies
Every day, patients and their physicians are making treatment decisions while important information about the prescribed drugs is routinely and legally concealed from them, expert observers have charged. Evidence from clinical trials has been selectively withheld from publication in the scientific literature. Can we investigate this claim? The problem would use data from the US clinical trials database and two literature databases with a focus on multiple sclerosis (MS), where there is no cure but there have been 12 new drugs in the last 20 years.

PROBLEM 4: Cod booms and busts
The incredible abundance of cod that gave Cape Cod its name has been fished almost to oblivion, and now even Cape Codders eat imported fish. The goal of this project is to create an infographic representing the fluctuation of the Cape Cod cod population, as well as the origins of cod sold through out the US, for as far back as possible. The group may also look at ways to investigate the unresolved conflict between the catch data produced by the fishing industry and NOAA’s population estimates.

// ++ //

9:45 am: Brian Hayes is talking about telling stories using code and data. His first example is about a story he “has been trying to tell for 40 years.” His first data viz was explaining how refrigerator magnets work, in terms of physics. As he tells it, some poor artist had to interpret rows and rows of 1s and 0s, and turn them into a black and white chart. Next innovation: a little bit of color. But they fail to convey what is happening to the reading audience. But now we are able to do it much better, with dynamic visualizations, using JavaScript.

This ability to create these kinds of visualizations is now available to us for the first time. He’s showing off some visualizations that work well — one of the MBTA, which is well known among the local dataviz community. “This is magic,” he says.

Next: a cool WSJ graphic that shows how dramatically vaccines can knock out diseases. That’s followed by a viz that illustrates climate change. (Spoiler alert: The data shows 2014 was the hottest year on record.) For fun: a program that’s supposed to help teach robots how to walk. “Robot zombies”, he calls them, because they keep falling down.

Here are his examples.

He ends with: “Have fun.”

// ++ //

9:30 am: Rahul Dave explaining Python to attentive science writers. Good questions from the audience.

Next up will be Brian Hayes, who will talk about using code and data to tell a story.

// ++ //

8 am, April 18, 2015: We’ll be live blogging and tweeting (#hacksciwrite) today’s hackathon, “Telling science stories with code and data.” It’s still quiet, as people are just starting to filter in. It’s coffee and pastry time now. But plenty will be happening later, so stay tuned.

It’s a great idea for hackathon. There’s growing, intense interest among journalists about learning code and data to help their storytelling and reporting. While many people learn perfectly well on their own, others do better in a group atmosphere. This helps those people.