Abstract

Authors: VERELST, David (DTU Wind Energy) BERGAMI, Leonardo (DTU Wind Energy) TIBALDI, Carlo (DTU Wind Energy) BLASQUES, José (DTU Wind Energy)

The paper discusses the principles behind the Open Science (OS) movement, and in particular its relation with the wind energy research community. Open Science refers to scientific research that is conducted, reviewed, and disseminated under the principles of transparency and “openness” to the public domain. Under an OS framework, all steps of a scientific research process, from creation to dissemination, evolve in their “open” counterparts: open source software models, open data sets, open peer review, and open access publishing. The paper will define and evaluate critically each of these steps while also relating them to the wind energy research community. Further, the authors will briefly argue why OS could be beneficial from a broader societal perspective.

Modern research often utilizes complex software models and large data sets not disclosed to the public domain. This approach hampers independent review and the reproducibility of research results. By removing such barriers, OS ensures that advancements are achieved through transparent and robustly validated steps. The review of scientific results in an open framework (open peer review) guarantees the quality of the results by enhanced transparency in the review process, and encourages active enduring contributions both from qualified reviewers and members of the research community. Dissemination and publication of scientific knowledge in an open access (OA) format reduces access barriers to published material, and enhances knowledge sharing. OA benefits the scientific community with the availability of a broader knowledge base, and the possibility to reach a wider audience, thus improves knowledge usage, increases citations, and amplifies the impact on society. The advantages of OA publications are recognized by mayor public funding programs, and they encourage actively to distribute the results from their sponsored projects as OA.

The shift towards an OS framework is occurring unevenly among disciplines. While some research fields are embracing more open practices rapidly, wind energy appears to lag behind. The paper aims at raising awareness, and stirring the debate within the wind energy research community about OS and its fundamental methods and tools. It is the authors’ hope that the paper will initiate a collaborative and “open” discussion forum on Open Science using the Torque 2014 conference as its kick-off event.

The Open Science of Making Torque from Wind

Thoughts and idea's towards increased collaboration and accelerated research progress for Wind Energy

Authors and collaborators: David Verelst, Leonardo Bergami, Carlo Tibaldi, José Blasques (DTU Wind Energy)

About the conference:

The Science of Making Torque from Wind is one of the most important scientific wind energy conferences. This is the fifth time that this conference is organized, and the second time to be organized at the Technical University of Denmark.

This is an IPython notebook, source available on Github, view using the online notebook viewer

Outline

  • What is Open science?
  • What is currently happening
  • Tools
  • Discussion

What is Open Science?

  • Open data repositories
  • Open source software and model development
  • Open peer review process
  • Open access publishing
  • Standard practice other fields such as economics, mathematics
  • Back to the basics of the scientific process
  • Science 3.0

Why Should Open Science Matter?

  • Reproducibility of research results
  • Publicly funded, publicly accessible (US, EU partially)
  • Accelerate progress:

    • more sharing, less double work
    • more checks
    • increased quality
  • Increase societal impact

  • For the benefit of all of society: distribute knowledge without discrimination
  • University top level strategy seems to align perfectly well with Open Science dogma's:

    • international collaboration with fellow researchers
    • accelerate innovation
    • collaboration with industry
    • address relevant problems in society, for the benefit of society

for instance lets look at the recently updated DTU strategy:

What Open Science is Not

  • Thread to the "established order", this is a renewed opportunity to proof our (increasing) relevance to society
  • A buzzword that will attract new venture capital and makes us researchers even more rich
  • The road to heaven

Open Data

  • Infrastructure: university servers, dropbox, and others
  • Data retention: duplicate data, commit to keeping data
  • Accessibility for easy sharing
  • Documentation
  • Usage of standard storage formats and protocols: database, compressed archives
  • Funding

Open Source

  • Software implementation of models are open
  • Usage of version control system: Git, Mercurial (Hg)
  • Code is an important part of our work, proper review is required
  • Not Invented/Implemented Here Syndrom: re-doing labours but otherwise not so interesting model implementations slows down progress.
  • Learn from the (open source) software development community:

    • geographically distributed teams, highly collaborative (international), transparent, creating high quality complex and widely used software
    • consistency: coding guidelines, code commit workflow
    • code review practices

Open Source (2)

  • But...what if other scientists "steal" my software and compete with me for the same grants?

    • Fraud and copyright violations also apply to any other context.
    • When the sources are public it is extremely easy to establish plagiarism
    • Why would a funding agency prefer a group that merely uses certain models/software over a group that actually develops the software?
    • The developers will have a much deeper insight into the code, and this will always gives the them the proverbial edge

Open Access

  • Different levels: gold, green
  • Do scientists care about open access?

    • Some quotes from Phil Davis blog:
    • Scientists have many identities, and tapping into the fundamental desire of the scientist-as-reader (access to everyone else’s work), does not always coincide with the fundamental desire of the scientist-as-author (recognition by one’s peers).
    • Consider page 11 of the Canadian study. While 97% of respondents relied on their institution’s subscription to access journal articles, the next most frequent response was contacting the author(s) directly for a copy (91%), followed by institutional repositories (76%), ignoring the article (75%), and social networking sites (50%). Pay per view took last spot at 27%.

Open Access (2)

  • Is open access relevant?

    • Yes, if we want research to be accessible beyond wealthy institutes.
    • From the open access book on Open Access by Peter Suber, rich Libraries have up to 100,000 subscriptions (Harvard), top Indian institute 10,000, and down to zero for many other institutes in the development world.
    • Why wouldn't it be relevant?
    • Innovation florishes when knowledge is wide-spread and easily accessible
    • Publicly funded, publicly accessible

Peer Review

  • Quality is an issue: many poorly reviewed open access publications
  • Open Access journals can be a scam, and there too many of them
  • A dire need for more clear and transparent review practices
  • Central review platform?

Open Peer Review

  • The review process is open and transparent
  • Reviewer can chose to remain anonymous or not
  • Review and discussion can continue after the publication!
  • Give reviewers incentives:

    • citable reviews
    • include citations to the reviews that helped improving the publciations
  • For a more detailed discussion, see an initiative in computer science here, and here

We should go back to what is the core task of a universities. Are we really creating for knowledge for the benefit of all of mankind, or just targeted industries and interests? What is society expecting from us, beyond the industry lobby groups, and beyond professors wanting to secure the next round of funding to boost a career.

Citing Peter Suber's Open Access book:

Even the wealthiest academic libraries in the world suffer serious access gaps. When the Harvard Faculty of Arts and Sciences voted unanimously for a strong OA policy in February 2008, Professor Stuart Shieber explained that cumulative price increases had forced the Harvard library to undertake “serious cancellation efforts” for budgetary reasons. Access gaps are worse at other affluent institutions, and worse still in the developing world. In 2008, Harvard subscribed to 98,900 serials and Yale to 73,900. The best-funded research library in India, at the Indian Institute of Science, subscribed to 10,600. Several sub-Saharan African university libraries subscribed to zero, offering their patrons access to no conventional journals except those donated by publishers.

For example, how is it that we publish in the AIAA while DTU does not even have (digital) access to those libraries?

Open Science Challenges for Wind Energy

  • Scientitif careers are by default not made with an open approach. So how can we realistically expect researches to value and follow open science guidelines?

Cultural and political changes

  • Code sharing, open source software development methodologies
  • Open data: sharing of experimental and simulation results by default (for example see the US, soon EU too?)
  • Agreements on software model interfaces and protocols

Funding and infrastructural challanges

  • Open data: deployment and maintenance of tools for sharing data
  • Retaining data: persistant funding for infrastructure
  • High impact Open Access journal: demand from publishers availability for gold open access license (Horizon 2020 requirement)
  • Open source models: securing long term funding across projects

Problems and Pitfalls

  • Reduced funding success rates: competing groups can use the same tools we developed and have access to the same data?

    • In the European wind energy context, how often is their real compition for the same project?
    • Collaboration instead of competition
  • Funding: expensive open access journals, small compared to project budget (small projects, institutes?)

  • Unconftarble working in the public eye
  • Currently no high impact open access journals for wind energy

What is currently happening

  • IPython (tool for collaborative science) and a data science, open source angle:

    • UC Berkeley, 1.15M$ from the Sloan Foundation link

    • 3 US universities (UC Berkeley, University of Washington, and NYU) secured 37.5M$ from several funding bodies link

  • EU Horizon 2020 call requires open access publishing (see press release)

  • EU Open Data agenda

What is currently happening (2)

  • HackYourPhD, collective of young researchers, developers and citizens calling on an open and transparent scientific process

  • An innovative concept to share open data with collaborators, under development: dat

  • Tesla (Elon Musk) opens up all its patents: the real competitor is else where, to fight climate change sustainable energy needs to grow much faster than a single company can

  • NASA is increasingly opening up code, and works actively on engaging with the community

  • Alternative ways of (crowd) funding research, for example Experiment and SciFund Challange

  • Center for Open Science, Open Science Framework

  • Preparing the UN climate meeting 2015 in Paris, US, China willing to finally commit to emmission reductions, Finland adopting 80% reduction scheme by 2050, ...

  • Climate change is our real challenge, in order to maximize renewable energy adoption we must work and share more openly

  • Parralel on NASA and community engagement: how to deal with negative wind turbine sentiment, subjective noise polution? Opening up will help dealing with this challenge.

Tools

  • Some open tools as an example
  • Ideally, the IT infrastructure is jointly funded by a group of universities, no need for unnecessary duplication
  • Similar propriety tools could also be used, not listed here

Open Data

  • As long as it is accessible, documented and versioned!
  • Cloud server software: owncloud
  • Cloud storage service providers: box, dropbox, spyderoak, and many others

Tools (2)

Open Source Software Development

Open Peer Review, Open Access Publishing

What's next?

  • Establish a network of interested collaborators with an as varied range of backgrounds as possible
  • Continue under the umbrella of the EAWE?
  • Funding/support for a course/practical guideline on open science methodologies and practices (version control, data sharing)
  • Publish only in open access journals
  • Identify collaborators for open source model development
  • Documenting data sets
  • Publically sharing data sets
  • Convince publishers and editors to experiment with open peer review

Discussion

Notes for future presentations

What about patents? Do they matter? Pattents are originaly intendet to "promote the progress of science and the useful arts". Is that still the case?

  • Open Access Barometer project at DTU (ended Dec 2013)