Dhaka: Hazard City

Category: Best data-driven reporting (small and large newsrooms)

Country/area: Bangladesh

Organisation: The Daily Star

Organisation size: Big

Publication date: 29/03/2019

Credit: Zyma Islam, Shaer Reaz, Nasir Uddin, Nurul Ferdous

Project description:

Dhaka’s growth is almost entirely unplanned urbanisation. It is very common to find chemical warehouses in the same building as residences, and huge skyscrapers without fire exits. As a result, Dhaka saw two of its most horrific fires in 2019 that left over 100 dead. We collected the fire service department’s internal survey of how vulnerable the city’s buildings are to fire hazards, and disclosed it to the public in an accessible format so that they can choose wisely which school to send their children to, or find out whether a chemical warehouse can blow up their entire neighbourhood. 

Impact reached:

Our project had two components: a map where people can zoom into their neighbourhoods to find structures vulnerable to fire, and a correspondent newsreport. The map acts as a timeless guide and is still referenced by our audience. As a result of the discourse generated by our project, fire safety became a priority among the general public. The government cracked down on corruption within the agency that issues building permits, filing cases against 23 people. Meanwhile, our audience reached out to us with questions on the building code, and in subsequent reports, we included visualisations and explainers for building safety. 

Techniques/technologies used:

When a chemical warehouse in an old part of the city blew up a whole neighbourhood on February 20, 2019, the fire service department leaked a survey they did of over 2500 buildings in the city. While this is not an exhaustive list by any means, it was still the first time we had any information on what neighbourhoods, or what structures are dangerous. The data was in Bangla and in a .pdf format. Bangla OCR is still in a developmental stage, meaning we all just had to sit down and do basic data entry. The next step was to geocode the addresses, which was a problem because far too many of these addresses did not exist on Google maps. We had to overcome that problem by manually finding latitudes and longitudes of different addresses. Because of the amount of manual work involved, the project took around a month.  

What was the hardest part of this project?

We all worked on the project as a side-gig in addition to our respective day-jobs at the newspaper. Datajournalism is still in a very nascent stage within the newsroom, and every bit of effort put into this project represents off-the-clock hours. The only reason this project even materialised is because a group of young reporters believe in the power of datajournalism and how effective it is in conveying important information to the public. The developer who helped us out with the front-end of the project is actually from the office IT team, who has never done datajournalism in his life. A graphic designer who constructs visualisations for print, contributed with the design. If this project is selected, it will help pioneer datajournalism within The Daily Star, and convince our editors to invest more time and resources into it. 

What can others learn from this project?

We believe that our project has the capacity to teach others in Bangladesh how to think beyond covering just breaking news. We could have simply investigated the two fires and left it at that. Everybody knows that the lack of building safety is an epidemic–but nobody has yet presented the breadth of the problem in the way we did. The government is the largest repository of data in the country, but most of the data is unusable, being preserved in formats that cannot be processed or analysed. In addition, most of government data is hidden to the public. We believe that this project can show people that we have a right to the information that the government holds, and that given the right skills, we can process and present them on platforms to make an impression on the public. 

Project links: