Thursday, July 8, 2010

ICIS Outstanding Paper Award

In the evening of Wednesday 28 January 1998, R.R. Donnelley's A. John Michaelis moderated a memorable panel discussion on electronic publishing at the EI Color Hardcopy Conference in the San Jose Hilton. Up to that heated discussion, the buzzwords were electronic printing and electronic publishing. After that day's visionary papers, Ralph Agazi of UC Davis led the charge in the panel discussion that we should stop talking about "electronic," because historically that term has been used when the steps in a mechanical workflow were changed to take advantage of the latest electronic technologies, but the workflow itself remained unchanged.

Inspired by the papers and the opening statements of the panel speakers, the audience demanded new buzzwords to characterize the paradigm shift from evolutionary process improvement to revolutionary new workflows. The label "digital" stuck.

In the past dozen years, a lot has been said and written about digital publishing, but less on digital printing. The reason is simple: the paradigm shift is taking much longer to complete in printing than in publishing. The print workflow has been perfected over centuries and is very efficient, with pipelining, buffering, and scheduling.

Electronic printing workflow

The camel's back was broken only when content creation tools evolved sufficiently that skilled workers were replaced by amateurs, because as Marshall McLuhan phrased it, instead of saving work, labor-saving devices permit everybody to do their own work. While FrameMaker died and InDesign+Illustrator got replaced by Word+PowerPoint, Web 2.0 brought the concept of a print service provider and the run length for a digital press job went from 500 to 1.

Digital printing workflow

While to a print job a print plant looks like a pipeline, to a print plant a stream of print jobs is a collection of processing steps that must be performed in parallel while avoiding idle times on resources, requiring new scheduling paradigms.

In such a complex heterogeneous system, the challenge is to find emergent model behavior, i.e., behavior caused by the interaction between characteristics of different models and not intended or foreseen by the model. The classical example is priority inversion between threads in an operating system, which causes deadlock.

Emergent behavior is discovered through simulation — the execution of models. HP's research in this area was presented at the recent 31th International Congress on Imaging Science in Beijing. The paper, Operations Modeling and Analysis of Commercial Print Services by Jun Zeng, I-Jong Lin, and Tzipi Netz, has received the Outstanding Paper Award. The paper is not online, but copies can be requested from Jun Zeng.