By opensourceway on Flickr under CC-BY-SA 2.0

The Open Science Working group have long supported training for open science and early introduction of the principles of open and reproducible research in higher education (if not before!). This area was a focus in 2013-4 and grows in importance as we enter 2015 with the level of interest in openness in science increasing at a rapid rate. This post attempts to provide examples of training initiatives in which members of the working group have been involved and particular areas where work is lacking.

1. Openness in higher education
2. Strategies for training in open science
3. The Open Science Training Initiative (OSTI)
4. Developing Open Science Training Curricula
5. Incorporating Open Science Training into Current Courses
6. Conclusion
7. Getting Involved in Open Science Training

Openness has the potential to radically alter the higher education experience. For instance, Joss Winn and Mike Neary posit that democratisation of participation and access could allow a reconstruction of the student experience in higher eduction to achieve this social relevance, they propose:

“To reconstruct the student as producer: undergraduate students working in collaboration with academics to create work of social importance that is full of academic content and value, while at the same time reinvigorating the university beyond the logic of market economics.”[1]

Openness focuses on sharing and collaboration for public good, at odds with the often competitive ethos in research and education. This involves more than simply implementing particular pedagogies or publishing open access articles – as Peters and Britez state bluntly in the first sentence of their book on open education:

“Open education involves a commitment to openness and is therefore inevitably a political and social project.” [2]

This could equally apply to open science. Openness is a cultural shift that is facilitated but not driven by legal and technical tools. In open education, for instance, open pedagogy makes use of now abundant openly licensed content but also places an emphasis on the social network of participants and the learner’s connections within this, emphasising that opening up the social institution of higher education is the true transformation. In open science, a lot of training focuses on the ability to manage and share research data, understand licensing and use new digital tools including training in coding and software engineering. However, understanding the social and cultural environment in which research takes place and how openness could impact that is arguably even more fundamental.

This section will focus on three topics around open science training, offering relevant linkages to educational literature and suggestions for teaching design:

1. Use of open data and other open research objects in higher education.
2. Use of open science approaches for research-based learning.
3. Strategies for training in open science.

As openness is a culture and mindset, socio-cultural approach to learning and the construction of appropriate learning environments is essential. While the Winn and Neary [1] focus on the student as producer, Sophie Kay [3] argues that this can be detrimental as it neglects the role of students as research consumers which in turn neglects their ability to produce research outputs which are easily understood and reuseable.

Training in evolving methods of scholarly communication is imperative because there are major policy shifts towards a requirement for open research outputs at both the funder and learned society levels in the UK, EU and US. This is in addition to a growing grassroots movement in scientific communities, accelerated by the enormous shifts in research practice and wider culture brought about by pervasive use of the internet and digital technologies. The current generation of doctoral candidates are the first generation of `digital natives’, those who have grown up with the world wide web, where information is expected to be available on demand and ‘prosumers’ who consume media and information as well as producing their own via social media sites, are the norm. This norm is not reflected in most current scientific practice, where knowledge dissemination is still largely based on a journal system founded in the 1600s, albeit now in digital format. Current evidence suggests that students are not prepared for change, for example a major study of 17,000 UK graduate students [4] revealed that students:

• hold many misconceptions about open access publishing, copyright and intellectual property rights;
• are slow to utilise the latest technology and tools in their research work, despite being proficient in IT;
• influenced by the methods, practices and views of their immediate peers and colleagues.

While pre-doctoral training is just as important, the majority of open science training initiatives documented thus far have aimed at the early career research stage, including doctoral students.

OSTI photo courtesy of Sophie Kay, licensed under CC-BY.

Open Knowledge Panton Fellow Sophie Kay developed an Open Science Training Initiative (OSTI) [3], trialled in the Life Science Interface Doctoral Training Centre at the University of Oxford, which employs `rotation based learning’ (RBL) to cement the role of students as both producers and consumers of research through learning activities which promote the communication of coherent research stories that maximise reproducibility and usefulness. The content involves a series of mini-lectures around concepts, tools and skills required to practice openly, including an awareness of intellectual property rights and licensing, digital tools and services for collaboration, storage and dissemination, scholaraly communication and broader cultural contexts of open science.

The novel pedagogical approach employed was the creation of groups during an initiator phase where each group reproduces and documents a scientific paper, ensuring that outputs are in appropriate formats and properly licensed. Next the successor phase sees the reproduced work being rotated to another group who must again validate and build upon it in the manner of a novel research project, with daily short meetings with instructors to address any major issues. No intergroup communication is allowed during either phase, meaning that deficiencies in documentation and sticking points become obvious and hopefully leads to greater awareness among students of the adequacy of their future documentation. The pilot course involved 43 students and had a subject-specific focus on computational biology. Feedback was excellent with students feeling that they had learnt more about scientific working practises and indicating they were highly likely to incorporate ideas introduced during the course into their own practice.

This course design offers great scope for inter-institutional working and as it uses OERs the same training can be delivered in several locations but remains adaptable to local needs. RBL would be more challenging to mirror in wet labs but could be adapted for these settings and anyone is encouraged to remix and run their own instance. Sophie is especially keen to see the materials translated into further languages.

OSTI is one of the first courses to specifically address open science training but is likely the first of many as funding is becoming available from the European Commission and other organisations specifically aimed at developing open access and open science resources and pedagogies. Some of the key consideration for teaching design in this space are:

1. How to address socio-cultural aspects in addition to imparting knowledge about legal and technical tools or subject-specific content and skills training.
2. The current attitudes and perceptions of students towards intellectual property and the use of digital technologies and how this will impact their learning.
3. The fast pace of change in policy requirements and researcher attitudes to aspects of open science.
4. Additional time and resources required to run additional courses vs amelioration of existing activities.

Open science curriculum map at MozFest 2014. Photo by Jenny Molloy, dedicated to the public domain via a CCZero waiver.

There are numerous one-off training events happening around the world, for instance the series of events funded by the FOSTER EU programme, which includes many workshops on open science. There are also informal trainings through organisations such as the Open Science working group local groups. Open science principles are incorporated into certain domain-specific conferences or skill-specific programmes like Software Carpentry Workshops, which have a solid focus on reproducibility and openness alongside teaching software engineering skills to researchers.

There are no established programmes and limited examples of open science principles incorporated into undergraduate or graduate curricula across an entire module or course. Historically, there have been experiments with Open Notebook Science, for instance Jean-Claude Bradley’s work used undergraduates to crowdsource solubility data for chemical compounds. Anna Croft from Bangor University presented her experiences encouraging chemistry undergraduates to use open notebooks at OKCon 2011 and found that competition between students was a barrier to uptake. At a graduate level, Brian Nosek has taught research methods courses incorporating principles of openness and reproducibility (Syllabus) and a course on improving research (Syllabus). The Centre for Open Science headed by Nosek also has a Collaborative Replications and Education Project (CREP) which is an excellent embodiment of the student as producer model and incorporates many aspects of open and reproducible science through encouraging students to replicate studies. More on this later!

It is clear that curricula, teaching resources and ideas would be useful to open science instructors and trainer at this stage. Billy Meinke and Fabiana Kubke helpfully delved into a skills-based curriculum in more depth during Mozilla Festival 2014 with their mapping session. Bill Mills of Mozilla Science Lab recently published a blog post on a similar theme and has started a pad to collate further information on current training programmes for open science. In the US, NCAES ran a workshop developing a curriculum for reproducible science followed by a workshop on Open Science for Synthesis .

NESCent ran a curriculum building workshop in Dec 2014 (see wiki). Several participants in the workshop have taught their own courses on Tools for Reproducible Research (Karl Broman) or reproducibility in statistics courses (Jenny Bryan). This workshop was heavily weighted to computational and statistical research and favoured R as the tool of choice. Interestingly their curriculum looked very different to the MozFest map, which goes to show the breadth of perspectives on open science within various communities of researchers!

All of these are excellent starts to the conversation and you should contribute where possible! There is a strong focus on data-rich, computational science so work remains to rethink training for the wet lab sciences. Of the branches of skills identified by Billy and Fabiana, only two of seven relate directly to computational skills, suggesting that there is plenty of work to be done! For further ideas and inspiration, the following section details some ways in which the skills can be further integrated into the curriculum through existing teaching activities.

Skills map for Reproducible, Open and Collaborative Science. Billy Meinke and Fabiana Kubke’s session at MozFest 2014.

Using the open literature to teach about reproducibility

Data and software is increasingly published alongside papers, ostensibly enabling reproduction of research. When students try to reanalyse, replicate or reproduce research as a teaching activity they are developing and using skills in statistical analysis, programming and more in addition to gaining exposure to the primary literature. As much published science is not reproducible, limitations of research documentation and experimental design or analysis techniques may become more obvious, providing a useful experiential lesson.

There is public benefit to this type of analysis. Firstly, whether works are reproducible or not is increasingly of interest particularly to computational research and various standards and marks of reproducibility have been proposed but the literature is vast and there is no mechanism widely under consideration for systematic retrospective verification and demarcation of reproduciblity. Performing this using thousands of students in the relevant discipline could rapidly crowdsource the desired information while fitting easily into standard components of current curricula and offering a valid and useful learning experience.

The effect of `many eyes’ engaging in post-publication peer review and being trained in reviewing may also throw up substantive errors beyond a lack of information or technical barriers to reproduction. The most high profile example of this is the discovery by graduate student Thomas Herndon of serious flaws in a prominent economics paper when he tried to replicate its findings [5,6]. These included coding errors, selective exclusion of data and unconventional weighting of statistics, meaning that a result which was highly cited by advocates of economic austerity measures and had clear potential to influence fiscal policy was in fact spurious. This case study provides a fantastic example of the need for open data and the social and academic value of reanalysis by students, with the support of faculty.

This possibility has not been picked up in many disciplines but the aforementioned CREP project aims to perform just such a crowd-sourced analysis and asks instructors to consider what might be possible through student replication. Grahe et al., suggest that:

“Each year, thousands of undergraduate projects are completed as part of the educational experience…these projects could meet the needs of recent calls for increased replications of psychological studies while simultaneously benefiting the student researchers, their instructors, and the field in general.” [7]

Frank and Saxe [8] support this promise, reporting that they found teaching replication to be enjoyable for staff and students and an excellent vehicle for educating about the importance of reporting standards, and the value of openness. Both publications suggest approaches to achieving this in the classroom and are well worth reading for further consideration and discussion about the idea.

Reanalysing open data

One step from reproduction of the original results is the ability to play with data and code. Reanalysis using different models or varying parameters to shift the focus of the analysis can be very useful, with recognition of the limitations of experimental design and the aims of the original work. This leads us to the real potential for novel research using open datasets. Some fields lend themselves to this more than others. For example, more than 50% of public health masters projects across three courses examined by Feldman et al. [9] used secondary data for their analyses rather than acquiring expensive and often long-term primary datasets. Analysis of large and complex public health data is a vital graduate competency, therefore the opportunity to grapple with the issues and complexities of real data rather than a carefully selected or contrived training set is vital.

McAuley et al. [10] suggest that the potential to generate linked data e.g. interconnecting social data, health statistics and travel information, is the real power of open data and can produce highly engaging educational experiences. Moving beyond educational value, Feldman et al. [9] argue that open data use in higher education research projects allows for a more rapid translation of science to practise. However, this can only be true if that research is itself shared with the wider community of practise, as advocated by Lompardi [11]. This can be accomplished through the canonical scientific publishing track or using web tools and services such as the figshare or CKAN open data repositories, code sharing sites and wikis or blogs to share discoveries.

In order to use these digital tools that form the bedrock of many open science projects and are slowly becoming fully integrated into scholarly communication systems, technological skills and understanding of the process of knowledge production and disemmination in the sciences is required. Students should be able to contextualise these resources within the scientific process to prepare them for a future in a research culture that is being rapidly altered by digital technologies. All of these topics, including the specific tools mentioned above, are covered by the ROCS skills mapping from MozFest, demonstrating that the same requirements are coming up repeatedly and independently.

Use of open science approaches for research-based learning

There are several powerful arguments as to why engaging students in research-based activities leads to higher level and higher quality learning in higher education and the Boyer Commission on Educating Undergraduates in the Research University called for research based learning to become the standard, stating a desire:

“…to turn the prevailing undergraduate culture of receivers into a culture of inquirers, a culture in which faculty, graduate students, and undergraduates share an adventure of discovery.”

The previous section emphasised the potential role of open content, namely papers, data and code in research-based learning. In addition, the growing number of research projects open to participation by all – including those designated as citizen science – can offer opportunities to engage in research that scales and contributes more usefully to science than small research projects that may be undertaken in a typical institution as part of an undergraduate course. These open science activities offer options for both wet and dry lab based activities in place or in addition to standard practical labs and field courses.

The idea of collaborative projects between institutions and even globally is not new, involvement in FOSS projects for computational subjects has long been recognised as an excellent opportunity to get experience of collaborative coding in large projects with real, often complex code bases and a `world-size laboratory’ [12]. In wet lab research there are examples of collaborative lab projects between institutions which have been found to cut costs and resources as well as increasing the sample size of experiments performed to give publishable data [13]. Openness offers scaling opportunities to inter-institutional projects which might otherwise not exist by increasing their visibility and removing barriers to further collaborative partners joining.

Tweet from @O_S_M requesting assistance synthesising molecules.

There are several open and citizen science projects which may offer particular scope for research-based learning. One could be the use of ecology field trips and practicals to contribute to the surveys conducted by organisations such as the UK Biological Records Centre, thus providing useful data contributions and access to a wider but directly relevant dataset for students to analyse. NutNet is a global research cooperative which sets up node sites to collect ecosystem dynamics data using standard protocols for comparison across sites globally, as this is a longitudinal study with most measurements being taken only a couple of times a year it offers good scope for practical labs. On a more ad hoc basis, projects such as Open Source Malaria offer many project and contribution opportunities e.g. a request to help make molecules on their wishlist and a GitHub hosted to do list. One way of incorporating these into curricula are team challenges in a similar vein to the iGEM synthetic biology project, which involves teams of undergraduates making bacteria with novel capabilities and contributes the DNA modules engineered to a public database of parts known as BioBricks.

In conclusion, open and citizen science projects which utilise the internet to bring together networks of people to contribute to live projects could be incorporated into inquiry-based learning in higher education to the benefit of both students and the chosen projects, allowing students to contribute truly scientifically and socially important data in the `student as producer’ model while maintaining the documented benefits of research-based pedagogies. This ranges from controlled contributions to practice particular skills through discovery-oriented tasks and challenges such as iGEM, allowing students to generate research questions independently.

There are significant challenges in implementing these types of research-based activities, many of which are true of `non-open’ projects. For instance, there are considerations around mechanisms of participation and sharing processes and outputs. Assessment becomes more challenging as students are collaborating rather than providing individual evidence of attainment. As work is done in the open, provenance and sharing of ideas requires tracking.

This post has introduced some ideas for teaching open science focusing on the student as both a producer and consumer of knowledge. The majority of suggestions have centred around inquiry-based learning as this brings students closer to research practices and allows social and cultural aspects of science and research to be embedded in learning experiences.

Explicitly articulating the learning aims and values that are driving the teaching design would be useful to enable students to critique them and arrive at their own conclusions about whether they agree with openness as a default condition. There is currently little systematic evidence for the proposed benefits of open science, partly because it is not widely practised in many disciplines and also as a result of the difficulty of designing research to show direct causality. Therefore, using evidence-based teaching practices that attempt to train students as scientists and critical thinkers without exposing the underlying principles of why and how they’re being taught would not be in the spirit of the exercise.

Support for increased openness and a belief that it will lead to better science is growing, so the response of the next generation of scientists and their decision about whether to incorporate these practices into their work has great implications for the future research cultures and communities. At the very least, exposure to these ideas during under- and postgraduate training will enable students to be aware of them during their research careers and make more informed decisions about their practises, values and aims as a researcher. There are exciting times ahead in science teaching!

If you’ve found this interesting, please get involved with a growing number of like-minded people via the pointers below!

• Run an OSTI in your institution. Contact Sophie Kay for more info via the OSTI website.
• Contribute to the Mozilla Science pad on an open science curriculum. All are welcome!
• Contribute to the Centre for Open Science Collaborative Replications and Education Project (CREP).

More projects people could get involved with? Add them to the comments and the post will be updated.

References

1. Neary, M., & Winn, J. (2009). The student as producer: reinventing the student experience in higher education.
2. Peters, M. A., & Britez, R. G. (Eds.). (2008). Open education and education for openness. Sense Publishers.
3. For a peer-reviewed paper on the OSTI initiative, see Kershaw, S.K. (2013). Hybridised Open Educational Resources and Rotation Based Learning. Open Education 2030. JRC−IPTS Vision Papers. Part III: Higher Education (pp. 140-144). Link to the paper in Academia.edu
4. Carpenter, J., Wetheridge, L., Smith, N., Goodman, M., & Struijvé, O. (2010). Researchers of Tomorrow: A Three Year (BL/JISC) Study Tracking the Research Behaviour of’generation Y’Doctoral Students: Annual Report 2009-2010. Education for Change.
5. Herndon, T., Ash, M., & Pollin, R. (2014). Does high public debt consistently stifle economic growth? A critique of Reinhart and Rogoff. Cambridge journal of economics, 38(2), 257-279.

6. Roose, Kevin. (2013). Meet the 28-Year-Old Grad Student Who Just Shook the Global Austerity Movement}. New York Magazine. Available from http://nymag.com/daily/intelligencer/2013/04/grad-student-who-shook-global-austerity-movement.html. Accessed 20 Dec 2014.
7. Grahe, J. E., Reifman, A., Hermann, A. D., Walker, M., Oleson, K. C., Nario-Redmond, M., & Wiebe, R. P. (2012). Harnessing the undiscovered resource of student research projects. Perspectives on Psychological Science, 7(6), 605-607.
8. Frank, M. C., & Saxe, R. (2012). Teaching replication. Perspectives on Psychological Science, 7(6), 600-604.
9. Feldman, L., Patel, D., Ortmann, L., Robinson, K., & Popovic, T. (2012). Educating for the future: another important benefit of data sharing. The Lancet, 379(9829), 1877-1878.
10. McAuley, D., Rahemtulla, H., Goulding, J., & Souch, C. (2012). 3.3 How Open Data, data literacy and Linked Data will revolutionise higher education.
11. Lombardi, M. M. (2007). Approaches that work: How authentic learning is transforming higher education. EDUCAUSE Learning Initiative (ELI) Paper, 5.
12. O’Hara, K. J., & Kay, J. S. (2003). Open source software and computer science education. Journal of Computing Sciences in Colleges, 18(3), 1-7.
13. Yates, J. R., Curtis, N., & Ramus, S. J. (2006). Collaborative research in teaching: collaboration between laboratory courses at neighboring institutions. Journal of Undergraduate Neuroscience Education, 5(1), A14.

Licensing

Text is licensed under the Creative Commons CC0 1.0 Universal waiver. To the extent possible under law, the author(s) have dedicated all copyright and related and neighbouring rights to this text to the public domain worldwide.

Open Books Image: by opensourceway on Flickr under CC-BY-SA 2.0

It was a jam-packed week but I’ve tried to pick out some of my personal highlights. Puneet Kishor has also blogged on the meeting and there was an active Twitter feed for #SciDataCon2014 .

Text and Data Mining Workshop
Open Data Initiatives
Data from the People for the People
Summary
Bonus slide decks

Photo by Puneet Kishor published under CC0 Public Domain Dedication

On Sunday 2 November I ran a workshop with Puneet Kishor of Creative Commons as a joint venture with Open Knowledge and ContentMine. Armed with highlighters, post-it notes and USB-stick virtual machines, we led a small but dedicated and enthusiastic group of immunologists, bioinformaticians, plant genomics researchers and seabed resource experts through the basics of content mining.

Photo by Puneet Kishor published under CC0 Public Domain Dedication

We covered what it means, when it is legal to content mine and more broadly some of the policy and legal frameworks which impact access to and rights of reuse for the scientific literature. We hand annotated entity types in two papers about lion evolution and Aspergillus fungi. This aimed to get people thinking about patterns and how to program entity recognition – what instructions does a computer require to recognise what our brain categorises easily? Swapping over the papers showed 80-90% inter-participant agreement in entity mark-up suggesting a reasonable precision and recall rate for our content mining humans!

Photo by Puneet Kishor published under CC0 Public Domain Dedication

Everybody managed to scrape multiple Open Access publications and extract species names, we also discussed potential collaborations and had a virtual visitor from afar, Peter Murray-Rust. The overwhelming feeling in the room was once of dismay at the restrictions on reuse of academic content but optimism about the potential uses of content mining – we hope that an excellent collaboration opportunity around phytochemistry will come to fruition!

Several sessions over the conference highlighted how far we still have to go in terms of data sharing and particularly the challenge of gaining political will required for data sharing for global sustainability. Waltraut Ritter, a member of the active Open Knowledge local group Open Data Hong Kong, presented a paper co-authored with Scott Edmunds and others, making the case to policy makers that open data can support science and innovation. There is no guidance from the Hong Kong University Grants Committee on dissemination of research data resulting from its 7.5 billion HKD annual funding pool. Data sharing was explicitly flagged as low priority in 2011 and on enquiry in 2014 Open Data HK were informed that this assessment had not changed. Arguments to appeal to policy makers are clearly required in these situations and Waltraut expanded on a few during the talk.

Exploring the complexities of sharing data for public health research, Sanna Meherally reported on a qualitative study examining the ethical and practical background to potential research data sharing, involving five sites in Asia and Africa, and focusing on stakeholder perspectives. A key takeaway message was the importance of considering cultural barriers to implementation of funder data policies. Chief concerns raised in interviews were confidentiality, the potential for data collection efforts to be underplayed and the need to give something back to research participants. That the latter point was raised by so many researchers interviewed is encouraging given the title of the next day’s session ‘Data from the people for the people’, which was another focus of SciDataCon.

This double session focused on citizen science projects around topics related to sustainability, including biodiversity and climate change. Norbert Schmidt introduced projects in the Netherlands to monitor air quality while Raman Kumar from the Nature Conservation Foundation introduced a range of bird and plant ecology citizen science projects in India such as eBird, MigrantWatch and SeasonWatch. You can find the full session list here.

Cumulative hours of birding as of Sep 214 through the eBird India citizen science initiative

Tyng-Ruey Chuang tackled some of these issues in his talk on ‘Arrangements for Data Sharing and Reuse in Citizen Science Projects‘. He asked projects to compare themselves to Wikipedia in terms of openness, participation and tools. For instance, does your project retain or strip metadata from contributed images? Tyng-Ruey also emphasised informed participation – clearly state if citizen contributions are prima facie uncopyrightable or ask agreement for open licensing. This chimed with earlier points by Ryosuke Shibasaki about the need for citizen ownership of contributed data and agency to make informed decisions about its use.

The talk ended with a call to action, as the Open Definition was practically quoted and Tyng-Ruey called for raw data, now! He’s in good company at Open Knowledge!

The sessions above are only a small subset of the conversations happening across the whole programme and papers are available online for all sessions. There were many demands for more open data, from Theo Bloom using her keynote to call for the abolition of data release embargos to Chaitanya Baruo revealing that Indian geology students are using US data because India does not make its own data available for open academic research. However, there were also excellent case studies of the reuse of data and its value. It would have been interesting to see some more cross-cutting sessions including all of the data collection and sharing cycle, but that will need to wait for 2016! This is a thoroughly recommended conference for data scientists and managers as well as domain experts and has notable participation from the global South, which is excellent and enriches the perspectives discussed.

Finally, I can only apologise for not being able to report on the Strategies Towards Open Science Panel – I was giving a talk at IIT which clashed with the session, but I’ve no doubt some excellent points were raised which will soon be shared!

I couldn’t attend these sessions, but they’re worth a look! First up Susanna Sansone and Brian Hole on data journals:

Citizen science is nothing new

What do Benjamin Franklin, Johann Wolfgang von Goethe, and Francis Bacon have in common?

All were amateur scientists. Franklin invented the lightning rod, Goethe discovered the incisive bone and was moderately successful as an art theorist and Bacon can be considered as nothing less than the father of empiricism, or can he? Either way, the three shared a passion for discovering things in their spare time. None of them earned their pennies as professional scientists, if that profession even existed back then.

Discovery is a matter of thirst for adventure

Citizen science is in fact old hat. It existed long before disciplines existed and could be described as the rightful predecessor of all empirical science. It laid the foundations for what we know today as the scientific method: the rule-governed and verifiable analysis of the world around us. Still, amateurs in science have often become marginalized over the past 150 years, as scientific disciplines have emerged and being a scientist has become a real thing to do (read more here).

Citizen science’s second spring

Today, citizen science is experiencing a second spring and it is no surprise that the internet has had a hand in it. In recent years, hundreds of citizen science projects have popped up, and they’re encouraging people to spend their time tagging, categorizing and counting in the name of science (see here and here). Some unfold proteins in an online game (Foldit), while others describe galaxies from satellite images (GalaxyZoo and here) or count wild boars in Berlin and deliver the numbers to an online platform (Wild boars in the city). Citizen science has moved online. And there are thousands of people in thousand different places that do many of funny things that can alter the face of science. The Internet is where they meet.

Berlin Wall; East Side Gallery

The logic of Internet-based citizen science: Large scale, low involvement

Citizen science today works differently to the citizen science of Goethe’s or Franklin’s time. The decentralised and voluntary character of today’s citizen science projects questions the way research has been done for a long time. It opens up science for a multitude of voluntary knowledge workers that work (more or less) collaboratively. In some respect, the new kind of citizen science is drawing on open innovation strategies developed in the private sector. In their recent Research Policy article, Franzoni and Sauermann refer to this type of amateur science as crowd science. The the term is extremely effective at capturing the underlying mechanics of most citizen science projects, which involve low-threshold-large-scale-participation. Today, participation of volunteers in science is scalable.

The advantages of citizen science

When it comes to data collection, social participation and science communication, citizen science is promising.

For scientists, it is an excellent way to collect data. If you visit one of the citizen science directories (for example here and here) and scroll through the projects, you will see that most of them involve some kind of documenting. These citizen scientists count rhinoceros beetles, wild boars, salamanders, neophytes, mountains and trees. There is nothing that cannot be quantified, and a life solely devoted to counting the number of rhinoceros beetles in North America would indeed be mundane for an individual scientist, not to speak of the travel expenses. Citizen scientists are great data sensors.

For citizen scientists it is a way of partaking in the process of discovery and learning about fields that interest them. For example, in a German project from the Naturschutzbund (German Society for the Conservation of Nature), sports divers are asked to count macrophytes in Northern German lakes. The data the divers collect help monitoring the ‘state of health’ of their freshwater lakes. In follow-up sessions, the divers are informed about the results. The case illustrates how citizen science works. Volunteers help scientists and in return receive first-hand information about the results. In this regard, citizen science can be an excellent communication and education tool.

Citizen science brings insight from without into the academic ivory tower and allows researchers and interested non-researchers to engage in a productive dialogue. This is a much-needed opportunity: for some time now, scholars and policy makers have been saying how challenging it is to open up science and involve citizens. Still, what makes the new kind of internet-enabled citizen science science, is rather the context volunteers work in than the tasks they perform.

The honey bee problem of citizen science

The old citizen scientists, like Franklin, Goethe or Bacon asked questions, investigated them and eventually discovered something, like Goethe did with his incisive bone. In most citizen science projects today, however, amateurs perform rather mundane tasks like documenting things (see above), outsourcing computing power (e.g. SETI@home) or playing games (e.g. Foldit). You can go to the Scientific American’s citizen science webpage and search for the word ‘help’ and you will find that out of 15 featured projects, 13 are teasered help scientists do something. The division of roles between citizens and real scientists is evident. Citizen scientists perform honey bee tasks. The analytic capacity remains with real researchers. Citizen science today is often a twofold euphemism.

That is not to say that collecting, documenting and counting is not a crucial part of research. In many ways the limited task complexity even resembles the day-to-day business of in-person research teams. Citizen scientists, on the other hand, can work when they want to and on what they want to. That being said, citizen science is still a win win in terms of data collection and citizen involvement.

An alternative way to think of citizen science: Small scale, high involvement

A second way of doing citizen science is not to think of volunteers as thousands of little helpers but as knowledge workers on a par with professional researchers. This small-scale type of citizen science is sometimes swept under the mat even though it is equally promising.

Timothy Gower’s Polymath Project is a good case for the small-scale-high-involvement type of citizen science. In 2009, Gowers challenged the readers of his blog to find a new combinatorial proof of the density version of the Hales-Jewett theorem. One has to know, that Gowers is a field medallist in math and apparently his readers share the same passion. After seven weeks, he announced that the problem had been solved with the help of 40 volunteers, a number far too small to count as massively collaborative.

Nevertheless, Gower’s approach was successful. And it designated an form of citizen science in which a few volunteers commit themselves for a longer period to solve a problem. This form of citizen science is fascinating regarding its capacity to harvest tacit expert knowledge that does not reside in a scientific profession. The participation is smaller in scale but higher in quality. It resembles Benkler’s commons-based peer production or the collective invention concept from open innovation.

The core challenges for this kind of citizen science is to motivate and enable expert volunteers to make a long-term commitment to a scientific problem.

Both strategies, the large scale low involvement participation as well as the small scale high involvement participation have the capacity to alter science. The second however would be a form of citizen science that lives up to its name. Or did you never want to discover your own incisive bone?

Goethe and the incisive bone

Pictures

1. Franklin with kite: Franklin’s Experiment, June 1752
2. Wall: Own picture
3. Goethe dozing on stones: Goethe in the Roman Campagna (1786) by Johann Heinrich Wilhelm Tischbein.
4. Incisive bone, Gray’s Anatomy, 1918

Thanks to Roisin Cronin, Julian Staben, Cornelius Puschmann, Sascha Friesike and Kaja Scheliga for their help.

Understanding reproducibility in science

Reproducibility is fundamental to the advancement of science. Unless experiments and findings in the literature can be reproduced by others in the field, the improvement of scientific theory is hindered. Scholarly publications disseminate scientific findings, and the process of peer review ensures that methods and findings are scrutinized prior to publication. Yet, recent reports indicate that many published findings cannot be reproduced. Across domains, from organic chemistry ((Trevor Laird, “Editorial Reproducibility of Results” Organic
Process Research and Development) to drug discovery (Asher Mullard, “Reliability
of New Drug Target Claims Called Into Question”
Nature Reviews Drug Development) to psychology (Meyer and Chabris, “Why Psychologists’ Food Fight Matters” Slate), scientists are discovering difficulties in replicating
published results.

Various groups have tried to uncover why results are unreliable or what characteristics make studies less reproducible (see John Ioannidis’s “Why Most Published Research Findings Are False,” PLoS, for example). Still others look for ways to incentivize practices that promote accuracy in scientific publishing (see Nosek, Spies, and Motyl, “Scientific Utopia II: Restructuring Incentives and Practices to Promote Truth Over Publishability” Perspectives on Psychological Science). In all of these, the underlying theme is the need for transparency surrounding the research process – in order to learn more about what makes research reproducible, we must know more about how the research was conducted
and how the analyses were performed. Data, code, and materials sharing can shed light on research design and analysis decisions that lead to reproducibility. Enabling and incentivizing these practices is the goal of The Open Science Framework, a free, open source web application built by the Center for Open Science.

The right tools for the job

The
Open Science Framework (OSF) helps researchers manage their research workflow and enables data and materials sharing both with collaborators and with the public. The philosophy behind the OSF is to meet researchers where they are, while providing an easy means for opening up their research if it’s desired or the time is right. Any project hosted on the OSF is private to collaborators by default, but making the materials open to the public is accomplished with a simple click of a button.

Here, the project page for the Reproducibility Project: Cancer Biology demonstrates the many features of the Open Science Framework (OSF). Managing contributors, uploading files, keeping track of progress and providing context on a wiki, and accessing view and download statistics are all available through the project page.

Features of the OSF facilitate transparency and good scientific practice
with minimal burden on the researcher. The OSF logs all actions by contributors and maintains full version control. Every time a new version of a file is uploaded to the OSF, the previous versions are
maintained so that a user can always go back to an old revision. The OSF performs logging and maintains version control without the researcher ever having to think about it – no added steps to the workflow, no extra record-keeping to deal with.

The OSF integrates with other services (e.g., GitHub, Dataverse, and Dropbox)
so that researchers continue to use the tools that are practical, helpful, and a part of the workflow, but gain value from the other features the OSF offers. An added benefit is in seeing materials from
a variety of services next to each other – code on GitHub and files on Dropbox or AmazonS3 appear next to each other on the OSF – streamlining research and analysis processes and improving workflows.

Each project, file, and user on the OSF has a persistent URL, making content citable. The project in this screenshot can be found at https://osf.io/tvyxz.

Other features of the OSF incentivize researchers to open up their data and materials. Each project, file, and user is given a globally unique identifier – making all materials citable and ensuring
researchers get credit for their work. Once materials are publicly available, the authors can access statistics detailing the number of views and downloads of their materials, as well as geographic
information about viewers. Additionally, the OSF applies the idea of “forks,” commonly used in open source software development, to scientific research. A user can create a fork of another project, to
indicate that the new work builds on the forked project or was inspired by the forked project. A fork serves as a functional citation; as the network of forks grows, the interconnectedness of a body of research becomes apparent.

Openness and transparency about the scientific process informs the development of best practices for reproducible research. The OSF seeks both to enable that transparency, by taking care of “behind
the scenes” logging and versioning without added burden on the researcher – and to improve overall efficiency for researchers and their daily workflows. By providing tools for researchers to
easily adopt more open practices, the Center for Open Science and the OSF seek to improve openness, transparency, and – ultimately – reproducibility in scientific research.