Note from the Editor

From Hemant K. Bhargava and Ramayya Krishnan:

In the mid-1990s, soon after commercial use of the internet began, and TCP/IP standards became widespread, we began discussing how the web could be leveraged for computational work, specifically OR/MS models. We recognized that the limitations of machine-specific computation—the need for models, data, and solvers to all be on the same machine or network, and be built specifically to work on them—worked against more widespread use of OR/MS models. Web technologies offered the potential to break this gridlock. The initial web had some, although limited, provision for distributed computation (versus mere display of HTML content). Subsequently, numerous advancements for distributed web-based computation had occurred within a few years, and we felt that the larger OR/MS community would benefit from knowledge of the vast array of new technologies—tools, languages, standards, programming methods—that could transform how OR/MS work was done and practiced. This is why we set about writing this article. Other catalysts for the paper included Harvey Greenberg and Art Geoffrion who had mentored and championed our work, and our experience within the DecisionNet project which aimed to lay out an electronic marketplace in which decision models, data, and other objects could be offered or used over a distributed computational network.

We believe we did a thorough job of exploring the range of relevant computational solutions, illustrating them, and organizing them all into a meaningful framework. There was, of course, a lot we did not know then and wish now that we did! For instance, we did not pay much attention to the market and economic aspects of leveraging the web for OR/MS computations. Similarly, while there are many brilliant examples of OR/MS deployments on the web, those interested in such deployments are often throttled by the lack of a reliable and enduring “public” infrastructure (e.g., an INFORMS server) for hosting, maintaining, and managing the use of data, models, and solvers. Still, many of the ideas presented in the article have survived in the intervening 20 years, and have become even more relevant in today's era where data and computation have become ubiquitous. Today, one doesn't think twice about the limitations of machine-specific computation when engaging with applications which deploy OR/MS methods: web-standard APIs and other such tools allow us to interact with and execute highly complex models and solvers with the swipe of a finger on our smartphones! Thanks to the work of Tim Berners-Lee and his intellectual descendants, OR/MS has lots more to gain as computation moves from the confines of local computing devices to anytime, anywhere, on anything.

I hope you enjoyed these insightful commentaries and agree that we need to do more to recognize the impactful papers and authors of our IJOC heritage. This is a small first step in doing that. Look for other steps to be enacted soon in a similar vein.

My best wishes,