From A2K Wiki

Contents 1 Dan Burk 2 Mark Gerstein 3 Arti Rai 4 Paul Uhlir 5 Questions & Responses 6 Description 7 Scientific Knowledge Resources 7.1 Web Resources 7.2 Work & Research by Speakers 7.2.1 Dan Burk

Dan Burk

Dan Burk Oppenheimer, Wolff & Donnelly Professor of Law University of Minnesota Law School

Presentation

Databases – Access to information in databases: The power to name is the power to control. How does law and how do social institutions affect this power?

Issues to consider:

Database Convergence Informatics Layers – Written about by Yochai Benkler. True of databases as well. Focuses primarily on the logical layer in the context of databases. “Metadata”

Standard setting – There are efficiencies and benefits to having the logical layer converge on a standard. There will be positive network effects as a result. One can see this in the information and databases in genomics, proteomics, etc. Has lead to attempts to create uniform nomenclature systems so that you can relate data from genomics to proteomics databases.

IPR Effects - Who gets to decide what the nomenclature is? TM comes to mind when you think of naming and IP. Copyright can also affect databases Star pagination dispute (Westlaw) Patent Law With the expansion of business method patents, systems of nomenclature could be patentable. Off the shelf memory and storage devices when infused with data also become patentable as a new object.

Data Standards

Where do we see this play out? HUPO – Standardization movement was developed and resulted in a competing standard dispute. Which should prevail?

Customized Data Cultural questions come into play. What are you excluding? What will you not see because you aren’t looking for it? What is the relationship of identifiable groups to this information?

One example: Indigenous Australian Peoples Attempt to put anthropological data into a database. Difficult to represent in a westernized database. Rather than looking at the database as a container, they tried to think of the database as a performance metaphor. Tried to create a “virtual garma.”

Reveals that databases can lock us into a particular cultural setting.

Solutions?

Architectural Neutrality? Is there a way to create information nomenclature systems that are friendly to data from other cultures? Problem here is that you lose convergence at the logic layer.

Open Source Ontologies Genetic Ontology Project Avoids Proprietary lock-in

Mark Gerstein

Mark Gerstein - Albert L Williams Associate Professor of Biomedical Informatics, Molecular Biophysics & Biochemistry and Computer Science, Yale University

Relevant links: Gerstein Homepage, Lecture ppt, Movie-1, Movie-2, A2K articles

Practical Issues relating to A2sciK

As an example, Human Genome Analysis – Paradigm for modern database oriented science

Background: HGP: Database Science

--Signature large-scale science project Many Scientists contributing “facts” to a distributed collection of DBs Mathematical analysis and annotation

--Fusion of Information and Life

-Social Framework Cooperative – International teams with an altruistic spirit; however it’s also competitive and people need/want credit for their work.

--Rapid growth in DBs in science. Creation and usage of databases increasing rapidly, as well as revenue from such DBs.

How does DB science work?

Distributed efforts among information servers administered by groups with very different rules, protocols. So for now, if you want to research a particular gene, you have to do a distributed search. Major hubs connect minor, boutique databases. Hub databases have power because their links to boutique databases give them visibility and prominence.

Aspects of distributed annotation --

--Elaborate synchronization between many sites to provide annotation. Updates occur extremely rapidly; requires intimate coupling of nodes. Need information connectivity to do proper science.

--How does this fit into academic publishing? Journal articles function as “annotation”. Powerful mechanism for doing distributed annotation. Can also use computers to read the literature. Can have them do text mining so that scientists can stay up to date with data.

--Statistical data integration – In data mining scientists want to put information together and synthesize it. Involves combining in tot disparate, heterogeneous information sources and computing statistical union. Requires that you download an entire DB and synthesize it.

What are the social impediments to database interoperation?

Impediment #1 - Vast resource devotion to protecting database information.

What are the practical implications?

Patches Make building intricate systems for interoperation difficult Impedes broad dissemination of ideas between labs

Impediment #2 – Clashing Cultures

Pay for use academic publishing vs open-source genomics Different traditions in academic publishing vs DB world Free text initiatives often come down to a question of resources.

Impediment #3 – Absence of social framework for protecting data

Researchers are unclear on the copyright law or the social norms. What about quoting annotations? What about credit? This is a disincentive to data integration. Scientists are impeding integration because they want to protect their data. Limits on bulk downloads and global analysis (passwords, IP filtering)

Arti Rai

Arti Rai Professor of Law - Duke Law School

Social architecture questions in collaborative innovation –

Biomedical Research

Not typically collaborative in the past – was a small lab/firm model. R01 – single principle investigator model.

Also a fair amount of secrecy is the tradition in biomedical research.

What about today?

In the academic sector it may be now worse because investigators are involved in commercialization of their research. Those actors in the social ecosystem that tend to deny access are motivated by desire for glory and involved in commercialization.

Why is this a problem? Secrecy prevents criticism, aggregation of knowledge.

Countertrends – HGP - Came from a micro-culture (worm genomics community) that was much more open. NIH is very interested in collaborative innovation. AR is doing empirical research: see her website. http://www.law.duke.edu/fac/rai/

Rai has conducted research in three main areas of scientific investigation:

Software (typically bioinformatics software), Databases and Wet Lab investigation.

What are the incentives? Traditionally publication. All scientists are looking for the first-author publication. So why collaborate?

Software - Bioinformatics

Many of the licenses are not GPL-style. (significant minority) Legal requirement to contribute improvements has less of a role because they are smaller communities. Many open source projects operate from the academic sector. The motivations here are additionally to get publications out of improvements, which is a difficult task. This creates an incentives problem for such incremental improvements. In the commercial sector you get reputation benefits from incremental improvements, but this does not appear to be the case in the academic sector. Perhaps public funding is required for this sort of work.

Databases

Human Genome Project: The Human Genome Project is perhaps not the best model for open access database development. This was in fact a very top-down project. In terms of disclosure, the head of the project was very involved on a continual basis. In addition, normative pressure for quick release of data was high. Pressures for publication were also put aside. NIH gave them monopoly position in return for getting sequence data out quickly into the public domain.

HGP may not be replicable. In the alternative, the Protein Data Bank may be a more useful day-to-day approach that has worked reasonably well for getting data dissemination. However the release of data takes place upon publication. NIH pressure has still been extremely important, as has pressure by scientific journals requiring data deposition after publication.

Wet lab work

WL is not going well in terms of getting information out. Concerns in the NIH funding community and the pharma-firm community (pharma business model depends on academic wet lab data). Could collaboration take place in this context? Perhaps, but one problem is a lack of data modularity – open source collaboration requires modularity of information. Proponents of patents say patents provide modularity through upstream patenting. However it is not clear this is actually working very well. We have upstream patents, but it is still not producing what we would like.

Information work going on here within public funding agencies and firms on pre-patent collaboration in a concerted fashion. E.g. Pre-patent collaboration on biomarkers. Also in CEO roundtable on cancer. Should we revisit the ways we think about ownership in patent law?

Paul Uhlir

Paul Uhlir Director of International Scientific and Technical Information Programs, National Academies – Creating Global Information Commons for Science

Presentation

Where innovation outpaces the ability of humans to respond, we must respond with new ways to manage and overcome entrenched opponents to change by those whose business models depend on the current paradigm.

What are the characteristics and benefits of print and digitally networked paradigms? Print Fixed, static Rigid Physical Local linear

Global Digital Networks Automated approaches to information extraction, processing and organization Zero marginal distribution cost Unlimited contents and media Transformative, interactive

There are new forms of open, distributed, collaborative research and information production on the internet. Open source software Distributed grid computing Virtual labs or collaboratories Community-based peer review (Journal of Atmospheric Chemistry and Physics) Collaborative research websites, blogs and portals (NASA Clickworkers, Wikipedia)

There are also new forms of open dissemination online Open data centers and archives (GenBank) Federated open data networks (World Data Centers) Virtual observatories (Digital Earth) Open access journals (PLOS) Open institutional repositories for that institution’s scholarly works (Indian Institute for Science) Open institutional repositories for publications in a specific subject area (PubMedCentral) Free university curricula online (MIT OpenCourseWare) Discipline-based commons (conservation commons)

The defining elements: information made available openly and freely. Often uses Creative Commons Licensing.

There is also open availability of publicly-funded scientific data online. It facilitates transfer of information from North to South. Promotes capacity building in developing countries.

GICSI – (Global Information Commons for Science Initiative) Goal is to create an online “open access knowledge environment.” Collaborative effort among many world organizations.

Questions & Responses

Was the flow of information prior to the internet really so poor? Perhaps we should not dismiss the global flow of information over the past few hundred years.

How will we filter bogus information in a common information pool? Some of it is masqueraded quite well.

Principle is that openness will be the default rule. There may be reasons to not make something open, but they should be justified. Regime may be context and disciple specific.

1. 2 – Question to Professor Rai

What purpose does secrecy serve beyond self-interest of agents. (commercialization and glory) Are their social benefits to secretive environments? What does secrecy promote that we should address if we are to promote a more open environment?

Rai response: Secrecy could be presented as a collective action problem. Is this necessarily bad?

Other reasons to be secretive other than self-interest. Scientists want to release high quality data. Glory may have advantages in that the competition forces creativity. We may want to maintain this glory incentive.

1. 3 – How can we overcome professors’ reluctance for more open sharing?

How might universities use their leverage to induce professors to release data in an open format? Perhaps a few of the following policies would be a good place to start:

Altering the % of revenues that follow to profesoors from patented technologies. Modifying tenure review policies, i.e. prestige accorded to articles published in certain outlets.

Rai: w/r/t tenure: Publication in prestigious articles is key. Can we introduce alternative measures? Could it be that data is downloaded by X people and was used in Y # of publications?

Gerstein: One of the largest barriers to collaborative databases is the structure of university evaluation of faculty.

Burk: Who are the gatekeepers in the scientific community? Journals, societies, universities. These advances are causing the roles of these traditional gatekeepers to break down. There is a tension because they were funded by proprietary and closed formats. As they are now unable to charge, they are less able to perform their traditional functions. How do we fund gatekeepers and maintain marginal cost close to zero? How do we now review and assure quality of data?

• Stevan Harnad: Apart from research-data self-archiving, let us not forget research article self-archiving. The way to induce researchers to self-archive is to mandate it. Recall also that -- unlike with open-access data and open-source software -- open-access article code (i.e., the text) is read-only: Its informational content can be used (with proper attribution), but its formal text may not be re-used (except for quotation and citation).[1]
  • On the Deep Disanalogy Between Text and Software and Between Text and Data, Insofar as Free/Open Access is Concerned
  • Apercus of WOS Meeting: Making Ends Meet in the Creative Commons

Description

The concept of an information society is predicated in part on the idea that scientific knowledge can and will be shared easily between those who can make use of it. Exchange between previously isolated research groups was, in fact, one of the building blocks of the internet. The scientific endeavors of the 21st century are fundamentally different from the single-laboratory work of the past; some of the most exciting research is interdisciplinary, multinational, and data-intensive. From multibillion dollar nuclear colliders to thorough study of the genetic code, information transfer and collaboration has become a necessary part of the research landscape. As such projects become the norm rather than the exception, a concerted effort must be undertaken to facilitate the flow of scientific knowledge both during and after the research process.

Science today is caught in a conundrum: Some system must be in place for efficient data sharing, but the imposition of such a system implies control of the basic elements of science by those parties who create and institute it. While scientific knowledge builds on past knowledge and therefore requires sharing of information, the proper policy and legal tools to support such sharing are not obvious. For example, databases, including those of scientific knowledge, are protectable in the EU but not the US and such legal differences may make sharing impracticable or impossible. Will the legal concepts of patent and trademark be useful in solving these problems, or should scientific knowledge be governed in a different way? How should data be treated in an intellectual property system? Is technological protection superior to legal protection?

This panel will explore some of the policy problems in modern scientific knowledge sharing, potential solutions to those problems, and normative issues that arise from the sharing of data and results.

Scientific Knowledge Resources

Web Resources

• American Scientist Open Access Forum
• Berlin Declaration on Open Access to Knowledge in the Sciences and Humanities + Berlin 3 Policy Recommendation
• Bibliography of Findings on the Open Access Impact Advantage
• Budapest Open Access Initiative
• Consumer Project on Technology, Access to Knowledge Page
• Consumer Project on Technology, Treaty on Access to Knowledge
• European Commission "Study on the Economic and Technical Evolution of the Scientific Publication Markets in Europe"
• Model National Open Access Policies Stronger + Weaker
• Open Access Archivangelism (blog)
• Open Access Briefing Paper (JISC)
• Open Access News (blog)
• Open Access Overview (Peter Suber)
• Open Access Timeline (Peter Suber)
• Open Archives Initiative
• RCUK Open Access Policy Recommendation
• Ressources langue francaise: INRA (H. Bosc) + INIST
• ROAR (Registry of Open Access Repositories)
• ROARMAP (Registry of Open Access Repository Material Archiving Policies)
• ROMEO Journal/Publisher Self-Archiving Policy Directory
• Science Commons
• Self-Archiving FAQ
• "Subversive Proposal" (1994)
• UK House of Commons, Science and Technology -- Tenth Report
• Wellcome Trust Open Access Policy Summary
• "What Is Open Access?"

Work & Research by Speakers

Dan Burk

• D Burk. Cyberlaw and the Norms of Science, 1999 B.C. Int. Prop. & Tech. F. (1999)
• D Burk. Lex Genetica: The Law and Ethics of Programming Biological Code, 4 Ethics and Info. Tech. 109 (2002)
• M Lemley and D Burk Biotechnology’s Uncertainty Principle, 54 Case W. Res. L. Rev. 691 (2004)