Computing Now

a new code of ethics?

2009-07-29T16:38:51Z

While we are on the topic of ethics, let us consider whether technology that is obtuse and difficult to understand is an issue for software engineering ethics. Should there be a new code of ethics designed to ensure users benefit from the full realm of capabilities afforded by the technology they have available to them? And if so, how would that code be applied, and what would be its scope? Should this code require that devices do what users want them to in a simple and straightforward way — abstracting away the underlying and often siloed technologies on which they are based? Should it mandate transparency to the inner operations of a device or software to facilitate troubleshooting and provide clues to the causes of unexpected or unwanted behavior? Should it let users see how and when information collected about them is being used?

Part of the problem is that too much information is hidden from users. It was early in my studies of computer science, during the late ’80s, that I first encountered the term "information hiding." I believed this term to be an unfortunate choice of words. I found it condescending — rightly or wrongly — and associated it with unpleasant interchanges I had when seeking answers about software or hardware behavior. It was implied that understanding the complexity of underlying systems was not important for me, that I needed to be protected or shielded from it, and that understanding was beyond my grasp.

After all, this was at a time when it was generally accepted that VCRs all across the world were endlessly blinking 12:00 because their owners were unable to set their clocks. It was okay that technology was out of the reach of the consumers it was being sold to. Friends in other fields of study would proudly proclaim that they were "technologically" inept. That was also 20 years ago.

Unfortunately, people today still rely on the help of their "technology-savvy" friends with the most mundane and common operations—for instance, supporting Wi-Fi connectivity with different devices, managing email access, backing up, zipping and unzipping files, or for help with major problems that turn out to be well-known device-specific issues. Personally, I have yet to find an OS or device that makes all of this easy, and I still encounter inexcusable incompatibility and settings issues with mainstream devices and software. But it is worse for those users who do not understand how the underlying technology works and are not comfortable navigating through hidden settings, for fear of creating irreparable damage. I’m surprised that they are astounded to learn that a simple search on terms describing their problems brings up posts and user forums with help and fixes—though the help offered is still slightly out of their reach to apply comfortably.

Separately, another part of the problem is that users are not presented with simple intuitive interfaces that belie the complexity of the systems on which the services they are using are built. Too often the architectures and engineering restrictions of the underlying systems percolate up to the user unnecessarily, complicating even the most rudimentary operations. A software engineering code of ethics could set standards for abstraction so that users become both aware and capable of seamlessly moving across networks, devices, clients, media, and modalities.

So, when it comes to privacy, could this new code of ethics require that feedback to privacy settings be embedded in each object, making it clear what is and is not being shared, to what degree, and to whom? For example, a user should be able to see what their social network page looks like to different classes of users, what buying history and economic information is being shared with their vendors, and what actions and information caused an ad to pop up.

Maybe we need a code of ethics to force the issue to render systems more transparent, provide the right levels of abstraction to increase ease of use, and make it clear when personal information is driving a system’s behavior

Free/Open Source Project Hosting and Software Engineering Projects

2009-03-17T03:57:36Z

Every 2-3 semesters, I teach the Free and Open Source Software Practicum at my university, among other capstone and upper-division advanced project courses in software development. When I first offered this course several years ago at my university (initially bearing the magnanimous title of "Extreme Linux" but really being more generally focused on FOSS development), I thought to myself, "There must be someone else doing this already." After all, on the internet, you have the ability to test just about any idea with a few carefully chosen keywords. When I approached Google Code's project lead, Brian Fitzpatrick, about what I was doing with Google Code in my class, he wrote this blog posting. (You can also grab my article in CISE from this link.) If there is anyone who knows what people were doing with Google Code (and FOSS hosting), he would be among those "in the know". So, I figured, I must be onto something.

Just about everyone knows about Google Search and GMail but I think Google Code is one of the most compelling Google "products" in its own right, and it is highly useful for introducing many ideas of software engineering. Much like other Google offerings, it is free to anyone and sports a "less is more" design. Better yet, every project gets a significant amount of storage (1GB when I last checked, and increasing), which is more than most departments can provide on their servers for anything but a handful of student projects. And to my amazement, there is no advertising.) There are numerous features, including source code hosting (via Subversion), issue tracking, project documentation (through wikis and downloads), effort tracking (through an activity meter), and ad hoc collaboration. Although I am proficient in setting up most of these technologies on our own servers (and was in the business of doing so for years), the ability to do the same as Google Code in a robust, hosted solution that is well integrated is something to behold. And should any of my projects become wildly popular, I think Google is better able to handle the large number of downloads than my own institution (no slight intended but Google has the economics of scale working in their favor). With Google Code, my students can self-organize their projects, provided they agree to release any developed code under an accepted FOSS licensing agreement.

Although my course is about FOSS development and trying to emulate the process that real-world FOSS projects use, I find that the addition of project hosting sites like Google Code also allows students to achieve a "practicum" in software engineering as well and think that it would behoove many SE project-focused courses to do so. Many ideas of software engineering and the development "process", for example, are actually brought to life through the FOSS development model and Google Code in particular. It is well established in SE circles that review is a critical component in defect prevention. Google Code provides the ability for the development team itself to comment on anything on the project's site itself. The source code browser allows one to comment on code. This is an invaluable tool for doing code reviews and catching bugs before they're reported. The availability of a customizable issue tracker, again, is something that allows one to do defect tracking, not just in terms of code but the entire "process" itself, which can be adapted by the development team for its needs. Last but not least, the project can use wiki pages to set up a complete project web site without spending an inordinate amount of time on trying to make the most beautiful looking HTML, which is largely a waste of time when the focus of the project should be on, ahem, the product and ensuring there is appropriate documentation (especially in an academic setting). Of course, Google also offers Google Apps for Education (another free service) which can be used for making even slicker looking collaboration sites--public or private. (I won't go there now.)

While GC and FOSS project hosting might not be appropriate for all software engineering and capstone/project focused courses, especially in the unlikely event where your student project truly needs to be proprietary, it strikes me that at the very least it ought to be an option that is presented to students. I am convinced that student projects in many CS and SE programs tend to be part of a "culture of impermanence". In an time where the job market is tightening (even for our graduates), we need to think about how every student can have a portfolio. Employers often ask, "Have you ever worked in a team setting?" or "Can you walk through some code or software you developed as part of your studies?" In most cases, my students can now answer in the affirmative to both questions, among others. In addition to student projects, I use Google Code for a number of my own projects, and I can track "interest" in my various works and focus my efforts accordingly. In the short time I have been doing this, I have seen the number of inquiries about my research projects increase dramatically. While I cannot guarantee results for you or your students, I think you'll find Google Code to be well worth your time, and it's actually a pleasure to use.

In the interests of disclosure, you're not limited to Google Code. There are others. (e.g. SourceForge). Google Code, however, is one of the few that allows anyone to join (without special approvals from their staff) and doesn't bury its users in advertising and has a clean, intutive interface, and they seem to be committed to its evolution. Better yet, it seems like Google is trying to give something back to a community that has helped it to be successful. I suggest taking it for a spin and welcome your comments.

Folllow Me (Really)

2009-02-02T14:05:31Z

I must confess that I spent an inordinate amount of my childhood memorizing lyrics to popular (and not-so-popular) music songs and listening to AM radio, which I still think is vastly underrated. I used to pride myself on knowing the lyrics and being able to pick out just about any popular song on the piano, which I credit with helping me to cross the boundaries of familiar "classical" pieces (which I was able to learn more conventionally, by reading the music) and learning to pick out music by ear, which I think is what makes music a bit more wholesome and spontaneous. I'm still learning!

Consider these lyrics from the late John Denver (1974):

Follow me where I go what I do and who I know
Make it part of you to be a part of me
Follow me up and down all the way and all around
Take my hand and say you'll follow me

Of course, readers who have been following me (a growing number, thank you) since I started posting on Computing Now already know that I have a knack for moseying into a topic. Over the past few weeks, I have been finding myself more an more intrigued with the growing phenomenon of social networking. While many in my circles dismiss it as a fad--and I might initially have been one of them--I am finding the possibility that it might be useful to be more than a bit tantalizing, owing largely to my interest in distributed systems, which I think can help us to understand the truth (think "consensus algorithms"). Since joining LinkedIn, Facebook, and (ehem) Twitter, more and more people--past and present--have been coming out of the woodwork to find (or follow, if you will) me. As a friend (and colleague) of mine often joke, "You'll have to fight them off."

For my Stateside readers, I am sure you don't need to be reminded of the many great folk songs written by John Denver, especially if you are now in "middle age", which used to be the term to describe people in their 40s, so I am now officially part of the club in a technical sense. Most American children of the 1960's/70's is at least familiar with his work. And yes, he really did change his last name to Denver for the sake of his recording career. As far as I know, though, he never lived in West Virginia. Anyway....

I copied the refrain above, because it says a lot about what I have experienced with social networking thus far. It allows me to recapture a bit of my past, which for various personal reasons was spent living in many different towns, suburbs, and cities. There were many special people I left behind, and through social networking and this great big thing called the internet, I can at least let others know what became of me and find out what became of them, allowing me at least for a brief instant to reconstruct what I think my piecewise-continuous life is about. In this always on era, it can be really difficult at times to pause, and the clock keeps on ticking nevertheless. Although social networks can generate a lot of noise, I do believe there is a signal that ultimately emerges.

So forgive me for waxing spiritual or philosophical but during these difficult times for most world economies, I think there is a real opportunity to come together and take comfort in our interconnected nature. I see a huge amount of potential in the IEEE Computer Society and think social networking can, indeed, enhance it. I want to do whatever I can through my efforts in Computing Now and CISE Magazine to learn more about our readers and what we should be doing. It's my growing perception that today's readers already "get it" and would be willing to post their thoughts publicly. And I think they should. Toward this end, I believe an experimental approach is in order. I will continue to welcome your thoughts and comments on my postings but sometimes wonder whether the notion of a blog and comments is already passe. Something tells me that social networks have the potential to stay, because it is at least one way we can remember our past and not forget our future (to bastardize the words of Steven Hawking).

I am "gkthiruvathukal" on Twitter and am calling upon you to "follow me".

If it weren't for IEEE Distributed Systems Online, you wouldn't be reading this

2009-01-07T05:23:34Z

Computing Now has learned a great deal from several previous online efforts in the IEEE Computer Society, the most influential of which was IEEE Distributed Systems Online magazine. DS Online closed in December after nine years of publication. Jean Bacon, the magazine's inaugural editor in chief, has written a retrospective of the magazine, discussing DS Online's evolution and the characteristics that set it apart as an online publication (including archiving and topic areas).

Having been on DS Online's editorial board all these years, I can attest that our experience with DS Online influenced Computing Now's development in three major ways. We knew what level of IT support would be required to maintain an online community, and we understood the importance of financial sustainability (two key goals for CN). We had also gained some experience in representing multiple magazines (CN by definition represents all IEEE Computer Society magazines). Because Computing Now isn't a magazine, we don't archive original content under our own name; instead, CN relies on individual magazines for archiving.

Had it not been for the heroic efforts of DS Online's editorial board and Jean's gentle but effective leadership, you wouldn’t be reading these words. Please join me in commemorating DS Online's nine years of publication, and in thanking Jean and her board for making such an impact on the IEEE Computer Society.

Technical writing considered useful.

2008-11-25T15:55:51Z

Without a doubt, I must be showing my "middle age" just a bit with the title. Although I missed out on the fun of programming with punch cards, I still remember being part of the wave of CS students that was introduced to Dijkstra's classic paper, which was written a year or so after I was born.

We're definitely in an age now, I believe, where the goto statement might be in the process of becoming one of the least of our problems. Today's students are exposed to the virtues of formal methods, design patterns, and some such. Of course, it's still possible (scratch that...easier) to write bad software, and we almost don't get through a day without discovering some new malady, be it a bug or security/privacy issue. The optimist in me, of course, clings to the hope that we are making strides forward as a discipline to introduce a culture of reliable computing, even if there is still a lot of work to do.

To get to back to my topic, however, as readers of this blog (a growing number it seems, thank you) of this blog now, I try not to shy away from the hard issues. I try to practice what I preach, especially when it comes to curricular matters. In the past few years, I pushed our department and my colleagues to support a technical writing course that would be targeted to all of our majors in particular but could also be taken by other students in the university either as part of a minor or (more recently) a writing-intensive option. I've believed for a long time that the vast majority of software professionals lack the basic skills when it comes to technical writing. Collectively, this group is part of a "problem" that I call the "software comprehension problem" (for lack of a better term). Sure, technical writing is "just" writing. It's a lot like writing an essay or a scientific paper, right? Well, not quite. Writing technical documentation is something programmers "just don't do". It's often contracted out to others in many cases who just don't do programming, so the programmers end up having to dictate how the program works anyway, thereby distracting them from actual programming. These leads to an impedance mismatch of sorts, where people who just don't do each other's kind of work, ehem, work collaboratively to produce a document that the users just can't read. In short, I felt like we needed to do our part to make sure our graduates can write proper documentation and were they to end up on development teams that contract the work out, they'll at least know how proper technical documentation should be written.

So as always, I'm interested to hear from readers. As a discipline, should we be offering classes like this to our students? Should we make specialized courses like this available to others? Should they get writing credit for it? How should these kinds of initiatives be integrated into the curricular guidelines (e.g. those from ACM and IEEE)? Should we just make writing an integral part of the overall CS/CE experience? To all of these questions, of course, my colleagues and I answered a collective affirmative. We've been pleased with the results, and I think many of those employing our students are thanking us for it, even if it created some initial headaches for us as a department.

Enhancing Wiki Participation

2008-10-27T01:58:21Z

In a recent committee meeting at my university, the committee chair commented that he wished that more people on the committee would contribute content to our committee wiki, and even suggested that he thought most of us were not interested in the committee. Needless to say, I came to the conclusion that this is a topic worth blogging about, given that my use of wikis predates their being crazy, sexy, and cool.

By now, just about everyone reading this column knows that the wiki concept is behind Wikipedia, which effectively allows anyone to freely create, edit, or modify content(subject to some oversight, of course). Of course, fewer people might be aware that the concept was originally intended to support software engineering and the collaborative nature of requirements gathering and analysis.

The same technology behind Wikipedia has become more fashionable in the industry and among academics to enhance collaboration. Unfortunately, as is so often the case with web technologies, great ideas like wikis and portals fall into disuse (often within the same year). The common case I have witnessed is that the people behind the decision to adopt a wiki often become the only maintainers and--in many cases--leave it to the wiki itself to perform the actual task of managing. Until there's a significant breakthrough in artificial intelligence, humans will still need to manage other humans, especially if the end goal is to get something done at the end of the day.

I'm almost certain that I'm not alone in realizing that wikis aren't always used effectively but have a number of ideas that I think can help to enhance participation and make wikis an integral part of doing a commitee's (or any group's) business.

Consider having one person at each meeting take minutes, which can be composed immediately as a wiki page. I see many groups (including the one I'm in) taking meeting minutes in a word processing program and posting them online later. (It kind of defeats the purpose of a wiki, which renders HTML pages that do not require one to download, more often than not favoring the Word enabled user).
Consider prioritizing agenda items based on wiki participation, where folks who want to discuss a topic during valuable meeting time will get their items listed on toward the top of the agenda (when most participants are awake) with the maximum airtime. In-person meetings are often more effective when people take time to prepare meaningful content (and gather meaningful input) in advance.
Minimize paper by doing committee-specific work in the wiki itself and in real-time, if possible. In one of our meetings, I noticed that we were using hand-written comments (on printed forms) to evaluate a vendor solution--something that could have easily been done on a single wiki page, where you'd also have the advantage of seeing the previous comments (before making your own). Wikis can help to generate meaningful consensus and synthesis, not to mention eliminating the painstaking process of transcribing comments from multiple sources. Hand writing is bad, but having multiple documents or forms that contain duplicate information is even harder to synthesize.
Make sure all members of your group understand wiki technology and its core abstractions. Believe it or not, most people don't. Although some wikis are easy to use (Wikipedia being one of them), not all have been designed with ease-of-use in mind. And the reality is that as easy as web-based editing is, some people still take comfort in the good ole word processor--even in the technology field, the word processor is like comfort food. It may be years before it is displaced, which probably explains the prevalence of web based word processors, such as Google Documents, Zoho, and Microsoft Office Life.
Make sure everyone subscribes to wiki site changes. Believe it or not, e-mail remains the most used application in the internet era, especially in academic circles. I know more than a handful of people whose work has degenerated to r/w access to e-mail. When you get a notification that the site has changed, your first instinct might be to ignore it. But as wikis grow in use, eventually someone posts on a topic you truly care about. Not long afterwards, you're notified. And then, you post changes to that page or comments. Wikis are intended to be dynamic systems, and e-mail represents the "push technology" that springs us into action.
Finally, emphasize at every step of the way that wikis are not web sites but could be someday. In many failed wikis, I see folks who spend way too much time thinking about having the pages and navigation organized perfectly, not to mention wordsmithing things to death. What we want is to encourage the topics, content, and feedback thereupon to come in. Then we can make periodic adjustments. It is often necessary (while we're on this topic) for a few folks to play a lead (editorial) role in rebalancing the wiki from time to time to make sure the group doesn't completely lose focus. But an emphasis on perfection from the beginning almost inevitably ends in non-participation. Pardon the bad business cliche: Better is often the enemy of the good, which holds true especially in the wiki world. (No wonder Wikipedia is so popular. It mirrors our imperfection, so we can live with it!)

I hope these tips will help you to organize your own efforts. I see human productvity as the great unresolved challenge in our field and one that cannot be left entirely to proprietary solutions (like office "productivity" software). Wikis are helping some folks to be more productive, yet there are many who continued to be baflled by this technology. As I see it, this owes largely to a preponderance of companies and individuals who don't place a premium on writing. I like writing, so I find wikis tremendously empowering. Those who don't like writing might not share my sentiments. So I would urge all companies and organizations to consider not only adopting wiki technology but also embracing the importance of written communication. It's easier said than done but, I think will lead to a virtuous upward spiral in terms of productivity and team building.

d-clutter

2008-10-01T17:31:02Z

I've been teaching about distributed systems for a number of years now. When explaining distributed systems in my non-geek circles, I'm sometimes at a loss for words for why the subject is truly important, especially when I see how most of the ideas find limited use in the real world, even now. And for those keeping score, distributed systems as an area has a track record that leaves a bit to be desired, yet it's of vital importance in today's software world. Remember distributed operating systems? Fault-tolerant systems? The list is a long one, and my colleague, Dejan, in Computing Now posted a bit about this topic earlier in Grids faded in the Clouds.

As we all know, today's infatuation is cloud computing. It will solve all of the world's problems, or one would hope. Of course, it is precisely this kind of hubris that characterizes the technology field these past few decades. Yet one has to ponder the notion of cloud computing deeply and ask, "Can it truly deliver as a complete desktop replacement?" I think the answer is, perhaps, but there is trouble ahead.

So let's take a look at a real world problem and how cloud computing addresses it. It is a problem I have code named, d-clutter (tm). Ok, I don't hold the trademark yet but think I have a case brewing for one. Distributed Clutter is a problem that affects every one of us. I took a quick inventory the other day and realized that my computing life is, ehem, a classic case of self-created clutter. Well, this is almost true, my life has been set up in such a way that I'm increasingly nomadic. I have two desktop computers at the two different campuses where I teach and do reserach. I have three laptop computers (you've read about my EEE PC, the travel wonder); I also have a Mac 17" laptop (which more or less acts as my home desktop, given it's form factor), a Tablet PC, and a Linux desktop at home. I tend to use all of these computers at one point or another during the week. To add to the mix of desktop machines, I have distributed clutter piling up on "cloud" sites across the net, which also don't interoperate with one another. What's a computer geek to do?

Here's the problem. I tend to do development, research, and writing on these different computers. At least for the writing side, I am more often than not using a web-based word processor. The cloud does a nice job for me here--at least when the word processor is actually available. (Yesterday, there was a hiccup that prevented me from uploading my document. A nuisance, to boot.) But the cloud fails to help me when it comes to managing the local resources of my various computers? For that matter, what does it do to help me find private content that I keep in the clouds themselves? For example, I can install/use desktop search to find something on any given computer (using Apple Spotlight, Google Desktop, Beagle, among others) but when it comes to searching across computers, foggetaboutit (a word I picked up from some TV commercial recently and probably have transcribed incorrectly).

And this is where I see a huge disconnect between what we teach/research in CS and reality. Cloud computing may well be using some ideas of distributed systems behind the scenes, but it fundamentally presents a centralized view to the outside world, not at all unlike mainframe applications of the "past". The good ole client/server paradigm is alive and well, just with new window dressing. But distributed systems remains largely underutilized for solving the most basic problems of distributed data management and synchronization (the right way to back up, I contend), let alone for conducting a private search across devices that I own. The cloud could have a serious role in all of this, I believe, by acting as a key bookkeeping service for all of our objects of affection. Even if those behind cloud computing really don't want to store all of our data for us in its various incarnations, at least make it possible for us to find it.

The core ideas of distributed systems center around various forms of transparency. The desktop and cloud computing experience simply are not there yet, at least when it comes to transparent access to resources and data. It is true that this problem could be eventually solved by having a desktop that is always delivered to you from the clouds (remember thin client? the network is the computer?) and your own private virtual server. But this seems a long way off, especially when the internet providers are capping and shaping internet bandwidth. Furthermore, the basic notion of "always on" is at best a pipe dream (or pipes dream, as I often jest) so the mere notion of cloud computing itself, while catching on, will not achieve its full potential for years or decades to come.

I'd be interested to hear from you out there. (Don't be shy!) For those of you who maintain 3 or more computers for your home/work/life, can you explain to me how you keep your data coherent and consistent? I haven't played my hand entirely for how I'm doing it but hope that I can learn something from you that I might not be doing already.

Computational Thinking and Core Curriculum

2008-09-15T20:07:43Z

"Computational Thinking" represents "a universally applicable attitude and skill set everyone, not just computer scientists, would be eager to learn and use as described by Jeanette Wing (in CACM, Viewpoint, Vol. 49, No. 3). Like many computer scientists, while reading her viewpoint, I felt like she was saying what many (ok, all) of us are thinking during a time where computer science enrollments are still near historic lows. And for those of us who have been down in the trenches trying to remedy the situation, we have been hard at work to introduce new ideas within our respective institutions.

I won't rehash all of Jeanette's excellent article. Instead, I hope to commence a discussion that I hope could eventually become a topic or set of topics within Computing Now: how to engage the world in computational thinking. In today's posting, I am focusing on curricular matters. I pose the rhetorical question: How do we teach computational thinking? It's a discussion and debate worth having.

For the past few years at my university, Loyola University Chicago, I have been working on how to foster interdisciplinary and multidisciplinary research. My experience has convinced me that it's easier said than done and will probably require more than merely thinking computationally. It requires us to think from alpha to omega, that is from the beginning of the academic process (the secondary schools) to graduate studies, on how to engage students, other faculty, and the community in general. Toward meeting at least a part of this goal (getting students early) I introduced a new course on the History of Computing, aimed at the vast number of students who enter college and might never have taken a computing course. (After all, it's not required in most of our secondary school system, remaining a topic of niche interest.) My initial thought in introducing this course was to acquaint our majors with the pioneering ideas of computing, which do go back thousands of years, because I noticed time and time again that the vast majority of computer science students (and even many professional computer scientists) really don't know terribly much about computing history--or worse, history itself--and the important impact it's had on actual history! So I was developing the course concepts and, suddenly, an opportunity presented itself: core curriculum renewal.

At my university, similar to many universities, there is a good amount of rethinking on "what should be required" of all students. In the case of Loyola, the core renewal process focused on areas of knowledge/learning as opposed to what departments are teaching the courses. This important separation of concerns enabled the possibility that departments such as computer science could propose courses in certain knowledge areas, subject to intellectual merit of the course being proposed. I decided to focus on historical knowledge and, seeing that IEEE was publishing the Annals of the History of Computing and (now) Computing Then, I gained confidence that there must at least be some merit to the idea. There were also other related courses I discovered on the net (Kernighan's "Computing in our World" at Princeton and Bergin's History of Computing class at American University), so it was not without precedent at least to get such courses approved in other universities. What might be without precedent is to get the course included as a recognized core historical knowledge course and to develop a course that meets the rigid requirements of our core curriculum. (The course has proper texts, films, lectures, discussion assignments, and strong writing/presentation components.)

To my amazement and great delight, the course was approved for students to elect as one of their two required Historical Knowledge core courses in 2004, meaning that students can learn a good amount about history in general from a computing perspective. Since then, hundreds of students have taken the course, and many students have approached me to become majors. (Our enrollments haven't changed dramatically but are improving. We'll be happy to teach students about computing using a more humane approach.) I think it's an important part of introducing computational thinking, especially to incoming students. All too often, I feel that the first contact most students have with computer science starts with heavy-duty mathematics and programming courses, which is definitely important but sometimes fails to present a meaningful context. And a bad first experience with CS, based on my discussions with numerous students, is likely to be the last experience.

We don't need no stinkin' requirements (or do we?)

2008-09-01T15:17:00Z

I've been working on a small business venture in recent months, in which we are partnered with another company to do new (joint) product development. I've been in the middle of a heated debate with my colleague and our partner about whether we need requirements. The debate remains unresolved, oddly enough. Although a believer in agile software development approaches, I did happen to be around during what I term the "methodology era", wherein companies (especially in the Fortune 500) embraced--at least on paper--the notion of a software life cycle that comprises distinct stages: requirements, design, code, test, etc. Of course, the software industry knows now what it knew then: The waterfall approach is inherently sequential and in most cases resulted in software that took years to complete, after which many development projects were either no longer needed or developed by more "agile" competitors. It is no wonder that the notion of agility became the basis for an approach, wherein software is conceived and developed in a more iterative fashion, or waves (as I prefer to think of it).

But I do find myself wondering, "Is there something missing in the world of agile methods?" I see a good number of so-called agile development projects that seem to eschew the notion of good documentation, both in terms of requirements and user docs. It strikes me that having a requirements document that undergoes some degree of peer review can be helpful to many of these projects, especially where there is a complex or less-understood business domain. Even for well-understood domains, it can help to remember what has been done right and make sure it is implemented in version 0.1. So (as an example only) if you are building the world's next greatest web based word processor using agile methods--incidentally, there are at least 3-4 of them, some not so good--it might be time well spent to think about what makes a word processor a word processor and ensure key functional areas are addressed. I might be out of my mind but think that a software requirements specification (SRS) remains one of the most important "templates" for outlining this sort of thinking.

So this brings me to my last point. The IEEE has guidelines/recommended practices for writing software requirements. It's called the IEEE 830 standard, which appears to have been last revised in 1998 (1993 before that date). So I find myself wondering whether this standard is still applicable to the industry as a whole, or whether at some level, there is an implicit acknowledgment (for an interesting industry perspective, see this article by Conrad Weisert) that it's just too heavyweight for anything but the largest projects, perhaps focused on projects that will not be developed in-house. It just so happens that 1998 overlaps in time with the "Web 1.0 era" when, I believe, many were starting to question the traditional approaches to software development.

In conclusion, I think it would behoove IEEE to refresh the 830 standard to reflect the highly dynamic nature of agile software development methods. There is no doubt that requirements are important, especially when working with other companies and customers. It might also help me to end the heated debate of whether I (or we) need to do requirements. After all, if IEEE hasn't updated it's requirements for specifying requirements since 1998, it is no wonder that some question their importance.

What have you done for your "home page" lately?

2008-08-17T21:25:19Z

I sometimes wonder about academics and researchers in the computing and engineering fields, which has produced a great wonder that we now take for granted known as the Internet. Yes, even a good number of English-language publications find the Internet so matter of fact now that it's no longer capitalized. (In fact, internet is in the dictionary, which would explain this phenomenon.) But I find myself not in awe of this great invention becoming passe but rather how many academics in our fields--who really need to have a public presence and be visible in the field--are unable to create, let alone maintain, proper web pages. In fact, many personal sites I see out there seem to have been written once, much like the way CD-R technology works.

As we all know, scholarship and impact are matters of ongoing debate. In our field, however, there is a growing belief that journals and conference publications, while remaining important, are not the only things that are important to have actual impact. In the computing field, software and other artifacts (often unrefereed but nevertheless referenced) are paramount in importance. Your web page is one of the best ways to make these and other materials available (including your conference/journal papers, subject of course to the copyright restrictions) and, strangely enough, citable. In short, you owe it to yourself to put together a site that is well organized and focuses on your professional life. After all, not everything belongs on the internet, which is why people visit Vegas. (You know, what happens in Vegas stays in Vegas. What happens on the Internet might be in your search engine's cache!)

At the risk of getting myself into some serious trouble, I contend that most folks lack the expertise to maintain their own pages in HTML. Even those who do (like me, having written a book on web development) really should not be wasting time on coming up with some whiz-bang theme and spending countless hours tweaking HTML and scripts that generate it. In most cases, a web site with a straightforward and consistent appearance will do. Your goal is to convey essential information. (I would argue this is true of most web sites but especially applies to personal web sites.) Your students will also thank you for it later.

More often than not, I see badly organized pages that clearly don't take advantage of a content management system or an appropriate hosted solution. There are a number of routes you can go, depending on whether you have the resources to support it (or IT, as in information technology) locally. There are numerous open source content management systems out there: Plone, Drupal, and Joomla! are all excellent and easy to install. There are also a number of hosted solutions. The one I use to maintain my own web site, http://gkt.etl.luc.edu, is a relatively new service offered by Google, and it is free for academics. (You'll need to get your institution's IT department or your own department to register, so everyone in your department can use it. You can also set up a personal account that comes with limited storage, but I would avoid this option.) You can maintain your web pages, navigation, and do basic customization of the appearance without knowing much of anything when it comes to HTML and CSS. You'll also notice by looking at my URL that the site appears to be hosted at my institution but in fact is hosted virtually at Google. While there is the possibility that I'd eventually return to a department-hosted solution (we do actually support Plone locally), there is much to like about having Google host it for me. For one thing, my site can take advantage of Google's search capability. (For example, try "parallel" or "XML" and you'll find one of my papers in PDF!) In addition, there is something about knowing that my site is likely to have little or no downtime, given that Google stores multiple copies of everything in their data centers all over the world (distributed systems are a best practice, and most computing and IT organizations cannot compete with Google in this regard).

In any event, my point is the following. Whether you go with a commercially-hosted or self-hosted option is not important. The point is that there is no point in hand-editing HTML pages (and CSS) anymore, unless you are doing double duty as a graphic designer (or perhaps are doing research in computer art/music or graphics, where a fancy site might matter). You need to have a "less is more" presence that allows you to present a meaningful professional image and is easy to keep up-to-date. That's what led me down the present path. I'd be interested to hear your thoughts!

Give me (free) wi-fi, or give me...

2008-08-05T15:42:11Z

Please forgive me for invoking (spoiling) the words of Patrick Henry, one of the great thinkers and proponents of the American Revolution, who was no stranger to taking government officials to task on various and sundry matters, such as bureaucratic inefficiencies and corruption. Anyway, I don't understand politicians. I'm a computer scientist/hacker, and like many in the field, I cling to the hope that through technology and software, the world could be made better. So on a recent business trip to Philadelphia--the same trip where I was accompanied by The Incredible Shrinking Access Device (my Asus EEE PC 900 series computer), I was a bit dismayed to see yet another public project that appeared to have gone awry: "free" public wi-fi in Philadelphia.

Now before anyone (or everyone) living in Philadelphia starts flaming me, I'd like to point out that I already live in a city where a similar project was to be rolled out (not free but at extremely low cost) that appears to have gone nowhere as well. So for those of you in the City of Brotherly Love, fear not, I still love thee.

So what is it that got me so irked about Philadelphia's public wi-fi initiative? (suspense-building sound inserted here) Well, at least for a few hours, I had the unfortunate pleasure of being part of the public: I was in a public setting known as an A-I-R-P-O-R-T as opposed to one of Philadelphia's hip districts, enjoying the hot new dining scene there. If there is one place where people--including citizens of Philadelphia, mind you--tend to be where they might actually need free wi-fi, it's the airport. Anyway, I roamed to about 5 different locations in the terminal (terminal C) and found no evidence of free wi-fi anywhere. I did actually find some evidence of a paid option being available but, alas, I could not even connect to this option--an option that was becoming more palatable with each passing hours as my flight into the Windy City was being delayed by stormy/windy weather in the area.

So I decided to do a little more digging and found the sad story at this blog, and the following paragraph is all you need to read:

It has been well publicized that EarthLink recently announced its intention to sell its Wi-Fi networks and exit the municipal wireless business. Wireless Philadelphia and the City of Philadelphia continue to work together to explore options for the network's future. In the meantime, we are committed to our core mission of serving Digital Inclusion customers with internet access, hardware, technical support and training.

Hmm, what does this mean?

I think it means a couple of things. First and foremost, EarthLink realized that this is a losing proposition for which they were never going to make a dime. And that little I do know about government projects, this looks like another classic case of an award being made to the lowest bidder, only in this case the bidder did not realize the hidden costs. Second, it's another example of a feel-good project designed to appeal to constituents but--when it comes down to it--the city probably had not clear plan in place to finance. Free wi-fi ain't free. You need something called infrastructure. Expenses for even a simple deployment can rack up quickly, to which any IT department in a corporation or university will readily attest.

Because of the enormous costs, it's really important to identify a suitable pilot project and proof-of-concept. The City claims to have done this but by all accounts did not do the right pilot project. When it comes to wi-fi deployments, you really need to choose carefully. I would start with a large public outdoor space where the public actually gather. Cities like Philadelphia and Chicago have many such places. I'd also consider a large public (mostly) indoor space, e.g. an airport. An airport has one major advantage. There is an ample supply of people waiting around for long periods of time who will put significant demands on the network--a stress test, if you will. Coverage issues are non-trivial and replete with numerous dead zones. (I'm almost certain that I found at least one of them. But 5?)

I'll conclude by submitting that we, the technology professionals, must increasingly broaden our interests to include business and politics. We believe in feel-good projects but are sometimes unwilling to guide others appropriately to understand the risks and hidden costs. I'm convinced that the project in Philadelphia is just one of many that hasn't worked right or delivered the goods. Like many technologists, however, I cling to the hope that the day of ubiquitous affordable (or free) public wi-fi will come. I'm not quite ready to fork out US$40-50/month for a wireless broadband connection that I am likely to use infrequently.

So I apologize, Patrick Henry, but I am sure you will agree that I owe an even bigger apology to one of your contemporaries when I say that I have "not yet begun to...surf."

The Incredible Shrinking Access Device

2008-07-11T21:07:40Z

So this posting marks my debut on Computing Now: IEEE's exciting new foray into the Internet. It is a great pleasure to write this posting from an altitude of about 31,000 feet (about 10,000 meters), where I am unable to connect to the web (yet) to post it immediately. For as far as we've come and as high as we've gone, I'm mystified by the fact that a couple of rovers can blog (ok, it's more like photoblogging) from Mars while much closer to earth, my device is silenced.

Nevertheless, I'm on one of the few business trips I have taken since starting to raise a family a few years ago, and I am writing this posting on my new Asus EEE PC 900 (one of a growing number of ultraportable computers). For the first time ever, I have a computer that I actually enjoy carrying onboard the aircraft with plenty of armroom. It features an 8.9" display and ways just under 1 kilogram at around 2.1 pounds. With no moving parts (solid state drive, SSD, or cooling fan) it is arguably the beginning of a new era in computing. While not the only product of its kind, and probably one inspired by the One Laptop Per Child project (OLPC), this commercial offering comes with Xandros Linux preinstalled, which is fine for those wanting to have a device for basic authoring and web access--almost all I need, especially when I'm on business trips. Of course, being the eternal geek, I decided to install the Ubuntu EEE 8.04 distribution, which is the same as Ubuntu 8.04 LTS desktop edition but with all the needed drivers bundled and optimized for the smaller EEE display (smaller windows, smaller fonts, and fewer services started automatically at boot time). It's the perfect setup for my ultraportable computer and all the more remarkable when considering that you can actually run a full development environment (for Java and Python in my case), not that I am going to do serious programming with this device.

Heralded by a Sony laptop executive as a "race to the bottom" because the price point (models ranging from my top end model at $549 down to $299) and performance are a bargain in the laptop market, the new line of ultraportables from Asus, HP, and Everex (just to name a few) is something to behold. (The MacBook Air is a lovely ultraportable but comes at a cost of 3x-5x the EEE based on your desired configuration.) As I am sitting here in coach, I am amused that both I and my laptop--neither being all that tall-- can work comfortably together. Given the limited space on most aircraft nowadays, this incredible device is one of the few laptops that can actually sit on the retractable tray tables, leaving room for coffee and a snack.

Back to the future—and the subject of my blog—I see the this and similar competing offerings as the future of mobile computing or (perhaps more accurately) the way mobile computing should be. Most of us need portable computing as an access device but don't want to feel deprived. Having struggled with "large" devices that provide way more than I need at one extreme and "handheld/mobile/wireless" devices at the other extreme that leave me wanting a lot more, ultraportable computers offer the best of mobility and sufficient power to do meaningful work. It has been said by many in the field (including myself at times) that Moore's Law is "dead". While likely true at the top end, Moore's Law is no longer about transistors but remains important nevertheless in our quest to do things faster, cheaper, and better. It's about how much functionality we can pack into a given space (per dollar) and understanding the trade-offs. You've gotta be excited about the future when you can get a laptop like this that is vastly more powerful, capable, and reliable than the offerings of just a short time ago.

Pardon the obscure movie reference: "The future is so bright, I'm going to need to get me some shades."

PhDs and being an External Examiner

2008-06-26T16:10:10Z

In the past eight months I have been the external examiner for eight PhDs. Reading these PhD reports has been a very interesting experience, but equally it has taken quite a lot of effort, and to an extent been rather tedious. Having gone through this rewarding and to certain extent painful procedure, I believe I have learnt a lot about the process and have formed views too.

One factor that is evident is the need for universities to “churn” out PhDs. In the UK this is partially due to the Research Assessment Exercise (RAE), which uses the number of passed PhDs as a metric. Also, the limited three-year PhD funding in the UK, and other countries, means that unless a supervisor can find additional funds, the student will be unfunded and probably not complete. The need to finish a PhD within a three-year timeframe often means that the quality of the PhD is not always great, as there seems to be a head-long rush to complete work. This affects the quality of the PhD work itself, which can be weak due to the limited time to complete it.

One critical area that is often not carried out well is the comprehensive review of the state-of-the-art of the particular area being studied, and the resulting critical review of the existing work, which helps justify and motivate the thesis that the PhD student will pursue. To me, this review and critical evaluation is very important, and without it, often the resulting research work suffers from limited understanding of the area being studied. Another area in the PhD report that is also often weak is the final chapter that summaries, concludes and discusses future work. Again, this is an important chapter, because it gives the PhD student a chance to thoroughly discuss the work undertaken, as well as provide reflections and views on the research work completed. Finally, I believe that probably the most important factor for any PhD student is to have a supervisor that they meet regularly, are able discuss matters related to their research and who thoroughly reads their PhD report before it gets submitted to the review process.

No doubt every former PhD student, supervisor and external examiner will have views on these processes too, but I believe strongly that we must try and limit the attempts to “water down” the PhD process, as it should remain a hallmark of first class research.

Log in to comment!

Computing Then: A Case Study for (In)Directly Charging for Value

2008-06-05T16:15:01Z

Traditionally products or services are sold for the value they bring to the customers. In the recent past, the model has changed. HP earns more money on supplies for printers, than on printers themselves; Google earns money on advertisements, not on the search engines which are free to use; plethora of social networks are free of charge for its users, the revenue stream comes either from special membership (LinkedIn) or from advertisements (FaceBook, MySpace). The source of revenue is not directly related to the value customers get from the products and services. When we were conceiving Computing Now, we had a similar dilemma.

A few previous IEEE CS online publication attempts failed because they were not financially sustainable. Each IEEE publication costs money to produce it (staff, editing, printing, editorial board meetings, etc.) and it earns money through subscriptions (individual, corporate, library, DL, etc.). Intuitively, online publications have reduced costs (no physical printing, although content design costs remain), but it is also somewhat harder to justify subscriptions: typically a lot of content comes for free. We too chose to offer CN for free and to finance its costs from advertisements. CN features advertisements on the main page, newsletter, and through Webinars. All profit from CN goes back into making it better for its readers. What remained the same is a requirement to bring the value to the customers, in this case the visitors to CN pages.

The core value for IEEE CS and consequently for CN is the content from Digital Library and how we can bring it to the readers in a value-added fashion. In this second month of Computing Now, we bring an idea that came from many sources – Computing Then – a historical perspective on technology innovation. As Jeffrey Yost, Editor in Chief, IEEE Annals of the History of Computing states: Computing Then is designed to take a step back—to contemplate, explore, celebrate, analyze, and learn from the past. CN will continue to explore new ways how to bring the value to its readers, e.g. we plan to introduce standards and education departments in the forthcoming months. We welcome you to make suggestions, either by logging to Web account and posting comment or by sending an email to computing_now@computer.org.

Grids faded in the Clouds

2008-05-23T23:03:32Z

I remember many popular technologies that came and went by without significant adoption. Massively parallel processors (MPPs) were promising in nineties as well as distributed operating systems, such as Amoeba, Mach and Sprite. In the new century, Grid computing offered a new vision, it attracted investments, research, and even limited deployment but it has started fading as of recently. The most popular technology today is Cloud computing. A number of companies offer their vision and approach to cloud computing, such as Amazon, Dell, Google and IBM, HP, and Sun. Will cloud computing result in a broad adoption in community or is it yet another transient fashion?

There are many examples of substance v. fashion in the past: TCP/IP v. OSI protocols; Linux/UNIX v. microkernels. There were also technology wars, such as VHS v. Beta, or more recently Blue Ray v. HD DVD. What is driving technology adoption? Is it a superior technical aspect, or a good timing, price, robustness? There are numerous books on this topic, such as Innovator’s Dilemma or Crossing the Chasm, to name a few. There are opinions that architecture wins the technology war, or that it is the platform that matters.

Here is a dilemma: is it worth jumping on a bandwagon of new technologies? I recall distributed shared memory attracting research and then later papers on this topic being “banned” from publications, Distributed Hash Tables (DHTs) are another more recent example. What are hard research problems and what are fashionable avenues to attract funding or increase publication records? What are technologies that will change the way how we do computing or even living tomorrow v. those that will only remain in dictionary. For example, what are the interfaces that cloud platform will offer to users, will it be indeed be possible to pervasively use clouds for computing needs? What are the higher level services offered, how much can we trust platforms to store our data or even that they will be there when we need them.

Publish and Perish

2008-05-12T23:25:39Z

A lot of my friends have stopped publishing journal papers because they consider it a write-only medium. They publish a journal paper and almost nobody reads it thereafter. Instead, they publish at conferences with immediate community feedback.

So, I started asking myself, what if we could enable every reader to comment on IEEE CS articles, possibly even on every article stored in the Digital Library (as suggested by the editorial board of IEEE Computer Graphics & Applications magazine)? Could we move the Digital Library closer to the Wikipedia model without losing the academic rigor of traditional reviews? In addition to having reviews, we'd be getting a wealth of new ideas, critiques, and even a way to fight against plagiarism, by letting readers comment on all published articles. Not an easy task, but the opportunities are mind-boggling.

I’m envisioning even more opportunities for Computing Now: hosting useful services for its community, bringing in the best projects from student contests and the best courses for educating engineers, and gathering wisdom from researchers and practitioners across the world. Can we convert a mostly static operation into a vibrant community, where we all share our technical, academic, and practical knowledge and experience for the good of all—a Wikipedia for computer science services?

When Computing Now takes off, I can see it becoming a Facebook, LinkedIn, or MySpace equivalent for the IEEE CS. Let's use the benefits of long tail by opening up authoring and feedback to our community. To accomplish this, we need to identify what value means to our users.

For example, I can imagine professors creating lectures by mashing up papers from the Digital Library, annotating them, and assigning homework using a Computing Now online service. Similarly, could we enrich computer science professionals by taking them through an appropriate-level course on Ruby on Rails, packaged with books, papers, tools, and a chat room to answer questions?

We’re starting with modest goals, but we have high ambitions, the highest one being to serve you as a professional, student, and fellow computer scientist. Log in and post your comments on this blog or email us at computing_now@computer.org.

Computing Now: Catching Up or Taking Off?

2008-05-01T20:34:36Z

I have coauthored a number of papers, books, and patents, chaired symposia, and conducted many interviews. However, nothing has excited me as much as the opportunities raised by Computing Now. When I applied for the editor position for Computing Now (it had no name, shape, or vision at that time), my colleagues were surprised with my interest, given that it wasn’t a typical IEEE magazine editor in chief role. So, what got me so excited?

Subscriptions to most magazines, transactions, and journals—published not only by the IEEE, but also by the ACM and other organizations—have gradually declined. USENIX gave up on its Computer Systems journal a long time ago and retained only a newsletter and some events. Transitioning to online versions have been experimented with and pushed hard for, but without significant success so far.

The IEEE Computer Society has tried various models, including IEEE Distributed Systems Online, but was not entirely satisfied despite some successes. Lack of either IT support behind it or financial sustainability, or both, have indicated that we need something new. In other words, this was a challenging and hard problem—exactly what motivates me. :-) But that wouldn't have been enough.

I saw a tremendous opportunity to finally catch up with the rest of the community that is living and breathing Web 2.0, social networking, long tail, and global knowledge. Could we finally convert the one-way, broadcast model of IEEE CS publications into two-way communication between the authors and the community?

For our beta release, we have set some modest, yet exciting goals, leveraging efforts across all IEEE CS magazines and bringing together articles, book reviews, and departments in a unified fashion. Much as IT is consolidated in industry, we’re consolidating resources and efforts from all IEEE CS magazines into one space for our readers.

In this first issue, we went through the IEEE CS’s recent publications to identify hot topics and interesting papers. We chose to highlight articles on computing games from Computer and IEEE Computer Graphics & Applications magazines. We spotlight the latest articles from other magazines on hot topics such as green computing, robotics, and social networking. First the first time, we’re also consolidating book reviews from different magazines (write one for us).

We want to hear from you, and we will listen carefully. What can we offer you? Which articles do you want to see more of? What features? Read our About page and tell us what else would you like. For example, would you like to see a package of best papers in security? An interview with a professor or a CTO of company Y? Would you like to get discounts on membership or book purchases? Log in and post comments on this blog or email us at computing_now@computer.org.

We see this as a journey together: IEEE CS staff, volunteers, and, foremost, our readership. Let’s make Computing Now a success for you.