(Updated from the October 2010 version.)
The Unidata Policy Committee during its April 2010 and October 2011 meetings raised issues about significant use of Unidata services and software without acknowledgment, including:
Since the 2010 meeting, the situation has improved. For example, the original version of this document referenced the NOAA GEO-IDE Unified Access Framework site as an example that described several data access tools but didn't credit Unidata as the source of the software. Now the site prominently displays Unidata logos for THREDDS, the Unidata IDV, and Unidata Tools, as well as logos for third-party tools such as NCO and ncBrowse that depend on Unidata software libraries.
However, a search for the text string "unidata" on that site will come up empty. The Unidata logo appears with other logos in a NASCAR-style "Interoperability" box, but text represented as pixels in a logo is not indexed for searches. Hence it is difficult to determine, from the text on that web page, the extent of Unidata's involvement in creating the software or in developing and maintaining the infrastructure on which the other software depends. Without Unidata, much of the software on that page either would not exist or the data that makes the software useful would not be in an accessible form.
Another example is the McIDAS-V homepage, which provides information about the next generation open-source McIDAS visualization and analysis software, primarily based on Unidata's IDV software (which in turn is based on SSEC's VisAD software). There is no mention of Unidata or appearance of a Unidata logo on the McIDAS-V home page or the associated About McIDAS-V page.
These examples are not intended as criticism of NOAA's GEO-IDE UAF project, the McIDAS-V project, or the developers of those very useful web sites, but rather to provide typical examples of the context in which Unidata's software often appears without explicit, indexable exposure of the organization. NOAA's GEO-IDE UAF and SSEC's McIDAS-V are also trying to establish their own branding for greater recognition and exposure.
How large a problem is the lack of exposure or branding of Unidata? The situation could be viewed as analogous to Unidata making use of Java libraries in software such as TDS, netCDF-Java, and the IDV, but not acknowledging the organization responsible for maintaining Java.
It could be argued that organizations and developers who have chosen to create widely useful open-source software have implicitly also chosen utility as a higher priority than credit. This is especially true for middleware, such as TDS and netCDF, because the users of products that make use of such infrastructure are typically only aware of its existence when it stops functioning as intended.
On the other hand, it seems reasonable that NSF, the sponsoring organization that has provided the funding and support for Unidata to develop widely used software and services, deserves credit and recognition for their sponsorship. Ultimately people should be able to determine the value of funding NSF, if spin-offs from that funding result in leveraging the value of research and development for use not only in education and research, but also in the infrastructure used by other developers in products like McIDAS-V and NOAA's GEO-IDE, and in commercial products such as ESRI ArcInfo, MATLAB, and IDL. Crediting Unidata provides a way to determine where the funding came from for development of products that depend on Unidata's efforts.
But how practical is requiring citation of Unidata? According to Google, there are over 1,200,000 web sites that mention netCDF. Would trying to get those web sites to cite Unidata be a worthwhile use of resources?
Even if it were practical, how effective would it be to require that Unidata's name be part of the name of Unidata software, such as "Unidata THREDDS" or "Unidata netCDF", so that users of the software could at least trace its ultimate funding? Unfortunately, it turns out that the mere mention of "Unidata" doesn't uniquely identify the NSF-funded program. According to Google, there are currently about 1 million web sites that mention "Unidata", but only 12 of the first 100 hits refer to UCAR's Unidata. The rest refer to various other commercial entities, products, and projects that also use a "Unidata" name. UCAR's Unidata has no trademark rights to the name.
How practical would it be to just require acknowledgment of UCAR Unidata in future papers, web sites, and software that depend on NSF-funded Unidata efforts? It turns out that such a requirement in a software license disqualifies that license from being labeled "Open Source", according to current definitions of that term by both the Open Source Initiative and the Free Software Foundation. Such a restriction was included in the original Berkeley Software Distribution license but was later removed, primarily because acknowledging every contributing organization in aggregate works was deemed impractical, and because it adds restrictions beyond those already imposed by accepted open source licenses. For more on this issue, see the discussion in the Wikipedia article on the BSD license.
We recommend that Unidata request but not require acknowledgment of both UCAR Unidata and NSF (or other funding source as appropriate) in mention of Unidata products and services. We also recommend changes Unidata can implement to web pages, documentation, and other communications to better convey the development and continued support of Unidata software and services.
Some examples of such changes include: