Software that makes a difference

 

When Hugo Ledoux moved to TU Delft in 2006, he encountered a problem. He had just completed his PhD in computational geometry, focusing on modelling geoscientific data. “My PhD was very theoretical, and when I came to Delft, I had to deal with real-world geographic data”, he explained. “I often found errors in the data sets, and these prevented me from using them to extract information. There was no automatic way to repair them.”

Open software licence
Ledoux noticed that others had the same problem, but they seemed to spend hundreds of hours manually repairing their data sets. To avoid that, he decided to develop software to automatically validate and repair geographic data sets. This work led to a number of computer programs that became mature enough to be made freely available under an open-source license. For Ledoux, this was a no-brainer: “I solved my problem and I realised that I could help others by sharing the results of my work.”

“I solved my problem and I realised that I could help others by sharing the results of my work.”

Now an associate professor in the 3D geoinformation research group at TU Delft, Ledoux develops algorithms and methodologies to produce and extract information from 3D models of cities, buildings and landscapes. These models describe 3D features and objects found in cities and landscapes – such as buildings, roads, rivers, bridges, vegetation, etc. – and the relationships between them. The group’s work has applications in environmental modelling, crisis management, modelling of the interior of buildings, population estimation, amongst others.

3d-city-model

3D city model of Delft automatically reconstructed with Ledoux’s open-source software (https://github.com/tudelft3d/3dfier) and open data sets from the Dutch government.

Benefits of open science
“Most people using our software are in government agencies,” Ledoux explained. “We target users from industry and government and try to release software that will be useful to them. We make the extra effort of releasing Windows executables. The users don’t need to know all the mathematics behind the code; they can just use the software.” The group tries to offer tools that commercially available software do not provide. Because they believe in the benefits of open science, they always insist on developing open source software when working with external partners in government and industry. “We’re also choosing more and more to publish only in open access journals”, he added.

Ledoux estimates that he spends about 30% of his time creating, developing, documenting and maintaining software. Developing and maintaining high-quality software that can be shared and used by others is time-consuming, but Ledoux continues doing so because he enjoys doing this work and because he found that he could fit it alongside his research. “I have a permanent job, so I don’t necessarily need to write more papers,” he noted. “I can sacrifice one paper a year to devote time to software development.” Ledoux feels, however, that that may not be the case for everybody, particularly early career researchers who are still looking for a permanent post. “Researchers are still evaluated on their publications and h-index,” he said. “There is no formal recognition for software development.”

“There is no formal recognition for software development.”

New collaborations and grant opportunities
Nevertheless, developing and sharing software has raised Ledoux’s profile, and this has led to new collaborations and grant opportunities. Recently, Ledoux was contacted by a group in South Korea who wanted him to be part of their grant proposal. The proposal was successful, the funds will come from South Korea to be used in the Netherlands, and Ledoux may become an author of new publications. Through sharing software, Ledoux has also got in closer contact with practitioners in his field. “They often have other problems that sometimes open up new areas of research for me”, he said. He also likes to think that one of his PhD students recently got a job in Singapore because of the open source software he had developed. The software gave visibility to the student’s work: by using his code, others became aware of his research results.

For all these benefits, Ledoux doesn’t feel that it is always worthwhile to make software openly available. Not all the code he, or his PhD students and postdocs produce is well documented and supported, or made available through GitHub – the group’s favorite platform for software sharing and development. The group is not always rigorous about coding standards and documentation. “We’re not that structured”, he said. “We should be, but there isn’t enough time.”

“You get a PhD for your ideas, not for the software you write”, Ledoux explained. “Software can help you prove a point and show that something is possible, but it is always a means, not the end goal.” From that perspective, it is not always a priority for PhD students to make their code available. That takes extra time. The priority is to write and publish research articles. ”We do encourage PhD students to pursue open source projects if they think their code will be useful, Ledoux added. “They get time to develop and refine the code.”

“You get a PhD for your ideas, not for the software you write.”

hr_mks5846

More support
Ledoux’s group recruits students who are already skilled in computer programming. There is no formal training on how to create and release open source software. They learn from one another. In that sense, Ledoux thinks that it could be useful if the Graduate School at TU Delft would offer a course. Another area where formal support from the University would be welcome is on how to choose a license. Ledoux didn’t look for help – he figured it all out for himself – but admitted that it can be tricky to find the right licence.

In the future, Ledoux is considering to write software papers. Users don’t always cite the papers the codes are associated with. His group asks people who use the code to cite the respective paper, but that doesn’t always happens. Articles in software journals offer a potential solution to this problem. Acquiring a DOI for the code is also a consideration.

Ledoux would like to see his software development activities bring in extra funding opportunities. “If we develop software that is useful for the government, for example, we would like them to sponsor us directly to develop new features,” he said. Most of the group’s funding comes from NWO and ERC grants, but they also have contracts with the Dutch government to develop standards and software for the Dutch Kadaster (land register).

 


Library information box

As part of the TU Delft Open Science initiative, the Library together with ICT are looking at the issues related to open source software created by researchers at TU Delft – sustainability, career recognition, training, archiving, licensing and copyright. Interviews with researchers about their ideas and needs regarding open source software are part of the process that will help deliver a strategic framework for Open Science at TU Delft. One of the ambitions is to build a community of those actively using and interested in open source software and generate more awareness of the practical uses and challenges of open source software as output of the researchers at TU Delft. For more information see our Open Working blog.


More information:

Hugo Ledoux h.ledoux@tudelft.nl and  Maria Cruz m.j.MarquesdeBarrosCruz@tudelft.nl (Data Officer at the TU Delft Library)

Author: Maria Cruz

Photography: Marcel Krijger marcel@marcelkrijger.nl 

Don’t want to miss the next edition? Subscribe here!

TU Delft Library