Tuesday, January 15, 2008

The CACM is dead ! Long live the CACM

Readers of this blog will know that I love to HATE the CACM, the embodiment of all that is wrong in our IT-obsessed, SCIENCE-ignoring culture. Well, the main source of my angst is now gone ! vanished ! replaced by something that <shudder> resembles an actual scientific magazine. Here's the story:

Way back in 2005 (before Web 2.0 ! Prehistoric!), the ACM realized that something needed to be done about the CACM. As Peter Denning tells it,
When David Patterson was president of ACM, many researchers told him they thought CACM had never regained its vaunted glory of the 1970s. Patterson set up a committee to review the current model and propose ways to recharge its content and scope
This committee was chaired by Moshe Vardi David Patterson, and Moshe Vardi, who is now the editor-in-chief (EIC) of the new and improved CACM. He recounts a story that is familiar to all of us:
While the format envisioned in the early 1980s was that of a content-rich magazine, the reduced role of the editorial advisory board combined with a relatively small professional staff meant that most of the content came from submitted articles. Over the years those articles have evolved to be strongly slanted toward Management Information Systems. Over time, a significant segment of the ACM membership lost interest in the publication.
The committee was working off the model of Science, as a fast turnaround-high-prestige magazine that showcased work of broad interest, without necessarily being too technical in focus. The main conclusions of the committee were to leave the more "practice" sides of computing to Queue magazine, and focus on a more research and news structure. The new CACM has the following basic outline:
  • News: The news section will have a very distinct “voice,” covering research and practice in computing on a global scale.
  • Practice: CACM will offer coverage of cutting-edge, emerging technology issues.
  • Breakthrough Research: The goal is to bring research articles, covering the broad spectrum of computing research, back to CACM. This will be a combined effort of conference and program chairs (ACM as well as non-ACM) and the CACM editorial board to select the best research material coming out of conferences and share that information with the wide readership of CACM. Each selected research paper will be adapted to the broad CACM readership and will be preceded by a short “Technical Perspective,” giving readers an overview of the important points and significance of the research.
  • Refereed Articles: CACM will continue to publish, as it has since its early days, peer reviewed articles.
  • Opinions and Views: The final component in the content model of the new CACM is a section dedicated to opinions and views, typically of a nontechnical nature.
Of course, the middle two bullets are the new ones, and hopefully will drive CACM back to its roots as a technical magazine. What's interesting to me is that as Peter Denning tells it, almost this exact same restructuring was first proposed in 1983, and floundered after a visit to the offices of Science, where the ACM folks realized that a true research-driven magazine required far more technical manpower than the ACM of the time was willing to commit. Moshe Vardi indicates plans to expand the staff of CACM in order to make sure that the new incarnation can be effective in this mode.

Another important aspect of the new incarnation is a complete graphic design overhaul:
There is also a need for a complete redesign of CACM’s Web site, with the goal of creating a dynamic, rich Web site that is much more than an online store of the magazine’s content. The aim is to think of CACM as consisting of a print publication, a rich Web site, and email channel to readers.
And so we now have the 50th edition of CACM, in its new form. It's extremely unfair to judge a major revamp like this after one issue, but let's see how it looks anyway.
  • The design: It's pretty slick, with a web-based contents and navigation mechanism inside what is called a "Published Web format" designed by a company called Texterity. The feel is PDF-like, but on the web. There are all kinds of nice features for extracting interesting web links from pages, or from sections. Overall, quite snazzy (but then I'm not a professional graphic designer, so what do I know). I will add that given the general 30 year lag between technology in the "real world" and on the ACM sites, this is a ginormously humongous improvement. The interface does seem a little slow to respond to things like page turns and menu options; I'm using the actual PDF for quick reading and text cut-and-pasting.
  • General content: Given the 50 year anniversary, there are many retrospectives, as well as articles by each of the past CACM EICs. These tell a very interesting picture of the rise and fall of CACM, especially in the period when Peter Denning was in charge. He tells of how budget cuts and restructuring led to the decision to remove research content from it. There are perspectives from well known luminaries in the field, including Jon Bentley, Rodney Brooks, Jeanette Wing among others. There's a longer article by Gordon Bell on "a theory of the evolution of a computer"
  • Research content: There are two main "breakthrough research" articles. A first on MapReduce, the Jeffrey Dean/Sanjay Ghemawat algorithmic design paradigm that runs much of Google's internals, with an introduction by David Patterson. The second is on an old friend, locality sensitive hashing, (in a new incarnation by Alex Andoni and Piotr Indyk) and the introduction is by Bernard Chazelle, which basically means you all should IMMEDIATELY stop reading this and go read what he wrote, gems like, "Sharp measure concentrations and easy spectral predictions are the foodstuffs on which science feasts."

    Both of these are excellent choices for "broad interest" articles. MapReduce has revolutionized the way we think (or should think) about large-scale computing, and there's even been some attempt at an algorithmic model around it. Nearest neighbour search is one of the most basic operations in any kind of data analysis, and I know for a fact that many people actually use LSH for their data mining problems. More of this, please !
  • The opinion section: Not much to remark upon in this issue: definitely nothing that would satisfy a focus group attendee's wish for "blood on the pages". But again, time will tell.
For years, we've all been complaining about the CACM, and the new revamp definitely deserves our attention. It'll take some time before we can tell if the new version of CACM is truly better, but the signs are looking good.
Post a Comment

Disqus for The Geomblog