1. The proposed copyright licensing model (e.g. CC licenses) may not be a suitable mechanism to control or restrict use, as the type of data published through GBIF falls in a legal ‘grey area’. These data may not in fact be eligible for copyright at all, at least in some jurisdictions. Deciding this point may also be complicated by the types of content offered, particularly when video, image or other multimedia content is included. GBIF were advised not to contribute to misunderstanding by promoting licenses which may not be legally actionable.
2. There is a strong desire by some (about one third of the respondents) to attempt to restrict access for commercial use. However, as some discussed in their responses, it is very difficult to agree or define what constitutes commercial use.
3. Many of the concerns expressed relate to the need for better attribution. It is suggested that, rather than asserting a copyright license, GBIF should focus with high urgency on improving citation and tracking data use.

In response to this, GBIF has issued a second consultation call, with a set of proposed changes which are in my opinion a tremendous step forward¹. These include adopting CC0 for all data², educating the community, drafting community norms, allowing publishers to at least express the wish that data should not used for commercial gain, and building tools to facilitate citation, attribution, and data use tracking.

If you want to support GBIF in taking these necessary steps towards open biodiversity data, please reply to the consultation, even (or especially) if you are not part of the GBIF community. The deadline for replying is June 14.

In my previous post, I highlighted the legal issues showing 13,297 American bullfrog records downloaded from GBIF on a map. 96% of those records had no or a non-standard data license, making data use legally cumbersome.

But how much of this applies to all 417+ million occurrence records in GBIF? How challenging is GBIF’s 2014 mission to provide a machine readable, standard license for all datasets? Fellow Datafable¹ member Bart Aelterman and I tried to figure out.

Methodology

We used the GBIF registry API to obtain the metadata for all 11,000+ GBIF registered datasets and in particular the rights field, which is where data publishers can provide the license under which the dataset is published. We then created a unique list of all licenses used, which we annotated with parameters such as use allowed and attribution required. This information was joined back with the dataset information to get an idea of the distribution of certain types of licenses over all datasets and occurrence records. We also documented the guidelines we used for annotating these licenses.

In total we analyzed 11,974 datasets², representing 415,927,654 occurrences. The first thing we noticed is that only 10% of those datasets (26% of the occurrences) have a license. This is problematic (see further), but it had the welcome side effect that we “only” had to annotate 432 different licenses.

All code and data³ for this project are available on GitHub. #openresearch #ftw

Results

Overview of the licenses used

Standard licenses

Ignoring for a moment that CC0 is the only sensible license for data, a standard license (Creative Commons or Open Data Commons) is at least standardized and easy to understand. Only 1.4% of all datasets however (2% of all occurrences) are published with a standard license.

Data dedicated to the public domain under CC0 represents an even smaller percentage: 0.9% of all datasets (0.5% of all occurrences). The silver lining is that most data publishers who choose a standard license, choose CC0 (105 datasets).

Interpreting the other licenses

All other data are provided with no or a non-standard license, with a percentage similar to the bullfrog sample (98% vs 96% of the occurrences). These data are in a legal gray zone: it’s a mixture of legalese, norms, restrictions, agreements, or in most cases no information at all. It is up to every data user to figure out the details.

We tried to lift some of that burden by interpreting all these licenses, extracting some characteristics, but it should be clear that this is an attempt⁴ that should only be used with caution. The results are presented in the charts below. You can click the legends to toggle parts of the chart.

Datasets

Occurrences

Conclusion

Our analysis of the licenses of all 11.000+ GBIF registered datasets shows a bleak picture. Very few GBIF registered datasets can be easily and legally used, let alone without restrictions. This is mainly due to data being published with no or a non-standard license.

Fixing this is crucial, and GBIF’s 2014 mission to provide a machine readable, standard license to all datasets is a step in the good direction. We hope our analysis (which can be run again) and guidelines already help with:

The Secretariat would review existing metadata provisionally to assign⁵ each current data set to one of these categories and would then communicate with data publishers to confirm the assignment. [source]

More importantly, this mission should be used as an opportunity to make the rights field mandatory, require CC0, and shift the discussion about ethical data use (including attribution) to norms rather than ill-suited legal tools.

That doesn’t mean however that it is actually easy to legally use the data, something GBIF is aware of. As a test, I downloaded all 13,297 georeferenced American bullfrog records and would like to visualize and share these on a map using CartoDB. Technically, this would only take me a few minutes, but to make sure I’m not violating any restrictions, I need to take a closer look at the fine print.

65 data licenses

By downloading the data from GBIF, I agree with the data use agreement, which states among other things:

• Users must comply with additional terms and conditions of use set by the Data Publisher. Where these exist they will be available through the metadata associated with the data.

Indeed, GBIF includes metadata for each dataset included in my download, and a file with all the rights as supplied by the data publishers (rights.txt). Since my download aggregates records from 65 data publishers, I have to read and understand 65 license statements before I can use the data¹.

Standard data licenses

Four datasets are provided with a standard data license (e.g. Creative Commons) and thus easy to understand:

Interpreting the other licenses

I am entering unknown legal territory by interpreting the non-standard licenses, but since I would like to create an occurrence map with more than 4% of the data, I’ll try anyway.

24 datasets don’t supply rights, so I could either interpret this as: 1) I’m free to use these data under the general GBIF data use agreement, or 2) I don’t want to risk violating any applicable copyright or database rights, so I won’t use these data.

I interpreted² other statements as Non-commercial use (1 dataset, 22 records), Non-commercial use with attribution (5 datasets, 3,190 records), Public domain (1 dataset, 1 record) or All rights reserved (1 dataset: I’m looking at you, Royal Belgian Institute for Natural Sciences, 1 record). For 3 datasets (373 records) it was unclear to me what the license allowed.

The bulk of the data however (26 datasets, 6,894 records) have a license of the form (emphasis mine):

[Institution A] data records may be used by individual researchers or research groups, but they may not be repackaged, resold, or redistributed in any form without the express written consent of a curatorial staff member of [Institution A]. If any of these records are used in an analysis or report, the provenance of the original data must be acknowledged and [Institution A] notified. [Institution A] and its staff are not responsible for damages, injury or loss due to the use of these data.

… or something along the same lines, which I interpreted as Some use with attribution, no redistribution.

Overview of the licenses used

These results are quite bleak: only 4 data publishers (4% of the data) publish their bullfrog occurrences as open data and only 8 (28% of the data) publish their data compatible with practices or expectations in GBIF today. This doesn’t even address if the chosen license makes actual sense for data (see my previous blog post).

What maps can I show you?

Public records

This map shows all 544 records dedicated to the public domain (4%). To comply with the GBIF data use agreement, I must publicly acknowledge the data publishers whose biodiversity data I have used here. I am happy to highlight such open datasets:

• Herpetology Collection - Royal Ontario Museum
• Colección de anfibios - Museo de Herpetología de la Universidad de Antioquia
• Ministerio de Medio Ambiente, y Medio Rural y Marino. Dirección General de Medio Natural y Política Forestal. Inventario Nacional de Biodiversidad 2007, Anfibios, though it would be useful if a standard license was used instead of Público.

Public and non-commercial use records

Since my blog is not ad-supported, I can also include records with a non-commercial use restriction, which adds up to 3,756 records (28%). Beware if you want to repost this image.

The included data publishers are now:

Museum of Vertebrate Zoology, Ministry of Agriculture, Food and Environment, Cornell Lab of Ornithology, Bird Studies Canada, Ohio State University Insect Collection, National Museum of Natural History, Smithsonian Institution, Royal Ontario Museum & Universidad de Antioquia

All records

I cannot show you a map³ including the other 72% of the records, either because the license is incompatible with the other data (such as share alike), but mainly because I first need to contact 52 institutions to either get some clarification of their license or to receive a written consent that I can actually repackage the data as a map. Frankly, that seems quite a hassle for a simple map. And even if I did this, you would have to do the same if you want to use, repackage or redistribute that map as well.

Conclusion

I just used a small sample of aggregated data⁴, but I hope it demonstrates the unnecessary legal burden that is put on users of the data (note that I haven’t even started assessing the data quality or fit for my use). This will either result in less use of the data, or - and I think this is what happens most often - users ignoring the fine print. In both cases, it shows what I and others have written before:

A legal license is not the correct tool to enforce or communicate expected data use.

Biodiversity data should be dedicated to the public domain. Its ethical use should be communicated via community norms.

I faced the same question for the data shared through the Canadensys network, and I’m still immensely proud that most of the Canadensys participants released their data into the public domain under a CC0 waiver and rely on community norms, not legal instruments, to express how the data should be used. This makes the data truly usable.

This is the best and probably only valid approach GBIF should take, as much more eloquently expressed then I ever could by Jonathan Rees, Karen Cranston, Hilmar Lapp and Todd Vision in their response to GBIF on this issue. Canadensys benefitted from being a new, unified and small network in taking this approach, but that doesn’t mean GBIF cannot take such a “more radical” approach (as GBIF puts it).

This is why I would suggest to:

1. Draft community norms for data use and publication. The Canadensys norms could be a starting point (and are actually available on GitHub for that purpose). This would be immediately beneficial to data publishers that want to rely on GBIF community approved norms with CC0, instead of using the Canadensys norms (e.g. this dataset from the INBO).
2. Educate the GBIF community about licenses and open data. It’s only recently that widely-recognized best practices (such as CC0 for data or a definition of open data) are emerging from numerous domains, so it’s no wonder that many data publishers don’t care, don’t know or have misconceptions. Heck, I didn’t know or care about this until I educated myself 2 years ago.
3. Allow new datasets to only be published under CC0 and the community norms (this is the approach Dryad has taken). This ensures that at least new data are truly usable. The data publishers currently have to agree with the data sharing agreement before publishing, so the mechanism is already there to notify them.
4. Communicate the truly open stance the GBIF community is taking with data publishers of existing datasets, instead of provisionally applying a license that might be based on a misconception. I don’t know though if there should be a hard deadline to this migration period, what options/restrictions to offer to data publishers that do not want to move to open data, and if GBIF can legally apply CC0 to datasets after numerous attempts to contact the data publisher.
5. Promote standards and technologies that enable the effective tracking of data use (from Rees et al. above). GBIF is already working on this and should continue to do so.
6. And on the specific question regarding supporting restrictions on commercial use (option 1 in the document): I would find it disheartening if effort and resources are put into creating and supporting an infrastructure that would allow the use of an ill-defined and non-open data license.

I think most of all that this is an excellent opportunity for the GBIF community to send a strong message that truly open data is the way to go. I know it will help me to convince data publishers to publish their data as open biodiversity data.

As an exercise, I searched for open access (OA) research articles published after 2008 and related to "GBIF" AND "data quality".

Search tools

I tried four search tools:

I found it surprising that not all of these offer the option to filter on open access only. I am curious to know if this is mostly because of technical or other limitations.

Finding the license

For three articles, I dug a little deeper to find the license. Two of these use and clearly indicate Creative Commons Attribution (coincidentally these two were not found by ScienceDirect):

Hill A.W., Guralnick R., Flemons P., Beaman R., Wieczorek J., Ranipeta A., Chavan V., & Remsen D. (2009). Location, location, location: utilizing pipelines and services to more effectively georeference the world’s biodiversity data. BMC Bioinformatics, 10(Suppl 14), S3. doi: 10.1186/1471-2105-10-S14-S3

Belbin L., Daly J., Hirsch T., Hobern D., & La Salle J. (2013). A specialist’s audit of aggregated occurrence records: An ‘aggregator’s’ perspective. Zookeys (305), 67-76. doi: 10.3897/zookeys.305.5438

The other one:

Costello M.J., Michener W.K., Gahegan M., Zhang Z., & Bourne P.E. (2013). Biodiversity data should be published, cited, and peer reviewed. Trends in Ecology & Evolution, 28(8), 454-461. doi: 10.1016/j.tree.2013.05.002

… (which was not found by PubMed Central and DOAJ) can be downloaded as pdf (= OA), but if you actually want to do something with the content, you have to indicate in a very detailed manner what you want to (re)use the article for. Very few use cases seem to be free. In other words: yuck!

I am quite familiar with the Creative Commons licenses, but now I realize what a mess you get if those aren’t applied. This OA article also demonstrates that the term “open access” doesn’t tell you that much: there’s still a whole range of how open something really is.

In my previous post, I explained what open access is. But we can go beyond the simple question of “Is it open access?” and evaluate how open a resource actually is.

PLOS, SPARC and OASPA developed a 2-page guide to do just that. I will use it to assess the openness of my first paper (co-written with Luc Brouillet) that was published two weeks ago in PhytoKeys, a Pensoft journal. Not as a vanity project, but to figure out how well I - as an advocate for open - am doing by submitting my paper to PhytoKeys as the journal.

Desmet P, Brouillet L (2013) Database of Vascular Plants of Canada (VASCAN): a community contributed taxonomic checklist of all vascular plants of Canada, Saint Pierre and Miquelon, and Greenland. PhytoKeys 25: 55–67. doi: 10.3897/phytokeys.25.3100 GBIF key: 3f8a1297-3259-4700-91fc-acc4170b27ce

Reader rights

This journal provides immediate open access to its content on the principle that making research freely available to the public supports a greater global exchange of knowledge. (source)

No fees, no embargoes, better science. 5/5

Reuse rights

The article and any associated published material is distributed under the Creative Commons Attribution License 3.0 (CC-BY). (source)

Reusing and remixing galore. 5/5

Copyrights

Copyright on any article is retained by the author(s). […] (source)

And the transfer of copyrights between the authors and publisher is regulated through the CC-BY license (which applies to anyone), but some of these rights are explicitly spelled out in the policy. Nice! 5/5

Author posting rights

[…] Authors are thus encouraged to post the pdf files of published papers on their homepages or elsewhere to expedite distribution. […] (source)

This encouragement is not mentioned in the editorial policies, but on the journal’s homepage. It is not 100% clear if pre-prints (before peer-review) can be posted as well, but the fact that:

The work described has not been published before (except in the form of an abstract or as part of a published lecture, review or thesis) […] (source)

… and a SHERPA/RoMEO search indicating PhytoKeys as a blue journal, with a “pre-print archiving status unclear”, leads me to conclude that this is not allowed. 4/5

Automatic posting

Archived in PubMedCentral and CLOCKSS. (source)

The journal is indeed archived in PubMedCentral, but the latest issue (25) is not available yet. I’m assuming this will be done within 6 months. 4/5

Machine readability

This criterion is more verbose:

Article full text, metadata, citations, & data, including supplementary data, provided in community machine-readable standard formats through a community standard API or protocol. (source)

… so let’s break it up in parts:

The article (including full text, metadata, citations) is available as XML. These XML files are also deposited on a GitHub repository and seem to be marked up in a community standard (see the xml header info) and provided through a standard API or protocol (GitHub and http). I’m not sure the links to included figures work, as these don’t start with http://.

In addition, I posted the full article (including figures) as an editable, Markdown formatted text file on GitHub.

The data, which are hosted on the Canadensys Repository and indexed by GBIF are available as a Darwin Core archive, which is a community standard, and are provided through a standard protocol (http). There are no supplementary data. 5/5

Conclusion

28/30, which is awesome!

Congrats to the folks at Pensoft on providing clear policies for PhytoKeys, although the author posting rights and automatic posting could be clarified. Please let me know if I misinterpreted something.

Open access is free, immediate, and online availability of peer-reviewed research articles and other scientific works. It has several advantages over traditional publication practices and will hopefully replace those in the long term.

Old model

The invention of the printing press in the 15th century suddenly allowed works to be created, copied and distributed to a wide audience, rather than being painstakingly copied by hand. Scientific societies were founded, which used this tool to facilitate the publication of scientific works (peer-reviewing, type-setting, distributing). To recoup the (rather high) costs, articles were published in subscription based journals.

New model

This business model has worked for a long time (and is still the dominant model in use today), but since the ubiquity of the internet, the cost of publishing works has decreased dramatically, so other publication models are possible.

Numerous scientific works are also the result of scientific research that is funded with public money, so there is a moral duty to make those results available fast and free to anyone.

And publicly available scientific works increase the efficiency of research, not only research in general, but also the research of the author of the article, by:

• Reaching a wider (and different) audience
• Publishing research results faster (including negative results)
• Increasing engagement and collaboration
• Increasing transparency (which allows reproducibility and plagiarism detection)
• Allowing alt metrics systems to be built around scientific works

The importance of open access is illustrated in this video by Nick Shockey, Jonathan Eisen and Jorge Cham (of the excellent PhD Comics).

Licenses

From a legal perspective, the open access movement has been made possible by the creation of licenses - such as the Creative Commons licenses - that allow copyright holders to loosen some of the rights they hold over their work. The CC-BY license for example, which I use for this blog post, allows anyone to legally share and remix the work, as long as they credit the original source.

Flavours of open access

As with any paradigm shift, the ultimate goal of open access won’t be reached immediately, so there are different flavours of open access:

• Green open access: the journal might no be open access, but the authors can self-archive their article under an open access license (here’s a tutorial).
• Gold open access: the journal itself is open access (see for example the Directory on open access journals).
  • Gratis open access: The article is available online at no-cost.
  • Libre open access: The article is available online at no-cost and there are additional usage rights (yeay!)

Learning

I’m enjoying the courses on Codecademy, but since I have a tendency to abandon them before I reach the end, here are my learning resolutions:

• Finish the Python course
• Finish the JavaScript course
• Finish the jQuery course
• Create 3 projects
• Follow 3 API courses

And as a bonus, try R on Code School.

Creating

Learning is fine, but doing is better:

• Create a data repository on GitHub and release it to the public domain.
• Create a one-page website visualizing data, using the CartoDB API.
• Transfer my website from WordPress to GitHub, using Pelican.
• Write at least 5 posts.

And as a bonus, create a theme for my website.

Contributing

• Contribute at least once a month to someone else’s repository or documentation. (Must. keep. contributions summary. active.)

And as a bonus, help an old lady cross the street and/or nurture a wounded fantasy animal into a powerful pet ally.