Tata AIG General Insurance Company - Insurer Innovation Award 2024
Scientific computing 10 years from now - credit and reward
1. Scientific computing 10 years from
now – credit and reward
Daniel S. Katz
Senior Fellow
University of Chicago & Argonne National Laboratory
www.ci.anl.gov
www.ci.uchicago.edu
2. Scientific computing 10 years from now
Publishing Finishing
Modeling next
Working on dataset from paper on
generation
Swift 2.0 new mossy
fusion plant
simulation computing
• All my work is associated with me – I have a unique DOI (HOI?, not-federated?)
• When Swift 2.0 is done, it will state fractional credit assigned to
developers, libraries, and papers, and will get a DOI (e.g., my credit is 0.1)
• When dataset is published, it will state fractional credit assigned to people who created
and organized it, applications and tools that were used, papers, experiments, etc., and
it will get a DOI
• When paper is complete, it will state fractional credit ..., and will get a DOI
• When someone cites my work (tool, application, dataset, paper, etc.), fractional credit
will be assigned as defined in that work
• E.g., a new paper (A) assigns 0.1 credit to Swift 2.0. That 0.1 is assigned as defined in the Swift 2.0
release (each object knows its own credit assignment), and my credit is increased 0.01.
• If 100 new papers each assign 0.1 credit to paper A, my credit is increased 100*0.1*0.01 = 0.1
• If they also assign 0.05 credit to Swift 2.0, my credit increases another 100*0.05*0.1 = 0.5
• Time for a promotion? Let’s see how my work has been used, by looking at my credits
• Grid aspects: reliable database for objects and credit; tie to provenance systems
• Caveat: the details haven’t really been thought through enough – I also want the value
of an object to increase when more people use it
www.ci.anl.gov
Grid Computing: The Next Decade (2012-01-04) – d.katz@ieee.org
www.ci.uchicago.edu
3. Scientific computing 10 years from now
• Summary:
• Papers, datasets, and software are all digital products
that should be easy to “publish”
• These digital products should be easy to find and use
• All contributors to digital products should be
rewarded when those products are used
• This requires
• A supportive infrastructure (funds!)
• Supportive policies
• A supportive culture
(supportive = trustworthy, reliable, etc.)
www.ci.anl.gov
Grid Computing: The Next Decade (2012-01-04) – d.katz@ieee.org
www.ci.uchicago.edu