From the Web to Walden

Peg Syverson
December 20, 2006

I conclude that there are two ways of constructing a software design: One way is to make it so simple that there are obviously no deficiencies, and the other way is make it so complicated that there are no obvious deficiencies.

C.A.R. Hoare, The Emperor's Old Clothes, Turing Award lecture, 1980

It is often asserted that evolutionary change proceeds in the direction of greater complexity: the more complex the organism, the more highly evolved it is. It is possible, though debatable, that this holds true for biology. It is less clear whether it holds true in the domains of social structures, theories, or technology. Even in biological ecosystems, surely there have been extinctions that directly resulted from organisms evolving into greater complexity and specialization only to encounter life conditions that required simplicity and flexibility. Viruses are much simpler than humans, yet they are capable of species-level mutation at blindingly faster speeds. Who is more adaptive? Which species is “more evolved?”

In the evolution of both pedagogical methods and technologies, it seems clearer and clearer to me that the simple-minded belief that more complexity = higher levels of evolution cannot be supported. In fact, the lessons of research and observation both for teaching and learning and for the usability of new technologies demonstrate quite convincingly that real evolution often entails a reduction in complexity and a shift toward greater simplicity. It seems paradoxical, but consider the immense leap forward in technology afforded by the graphical user interface, which greatly simplified the demands on human cognitive resources required by complex command-line applications.

In teaching and learning, many complex schemas and frameworks and programs have evolved, only to perish under the weight of their own administration. Inevitably, the return to simplicity, to the most basic and authentic relationships between teachers and students proves the most effective method, and this is most apparent in the most highly technologized educational environments. Despite enormous, expensive, and intelligent efforts, the vast majority of technological applications for education will never enjoy widespread adoption. A very small number will be widely adopted by institutions and as widely despised and minimally used by frustrated and resentful teachers. Why is this?

Years ago my syllabus for a course, once I had thought through how the course would be structured, took only an hour or two to type and print out, and perhaps 15 minutes to make class sets. In the early days of the web, I could post a class syllabus in an afternoon of careful work. Now, by the time I design and build a class web site with cascading style sheets, an email list, a class blog, an archive of resources and contacts, and links to other sources of information, it is likely to blow a week.

The Learning Record project as it has evolved followed a conventional technological trajectory until very recently, with each iteration resulting in greater complexity and more features, more convergence, built-in data collection, multimedia archiving and greater scale. The Learning Record began as an 8-page form developed by elementary school teachers and educational researchers in inner city London schools. It grew into a full-blown large-scale alternative to standardized testing for high-stakes assessment.

When I arrived at the University of Texas the web was just emerging, so it was a simple matter to provide the Learning Record form in a text format for downloading from the web, for the students in my classes. As other instructors began using it and asking questions about it, as my students raised their questions, I began to provide more information on the web site to address their questions. Another level of complexity evolved when John Slatin and I wrote a succcessful DARPA proposal, at the program director’s request, for creating a version of the Learning Record that could be used to evaluate learning in online text-based MOOs and MUDs.

As the web developed the capability to connect with databases, the next obvious evolution for the Learning Record was to create a database-to-web application. This required a great deal of programming and scripting, but ultimately resulted in a successful and award-winning application which was used by over 7,000 students and several hundred teachers at 14 institutions. But it was complex. There were passwords, complicated scripting and server needs, system administration, and ongoing issues with file uploading and downloading. Databases require tending, database users require tending, and everyone involved with the project was busy fielding inquiries, fixing password issues, solving technical problems, and also using the application in their own teaching. It was a grand project on the Open Source model, developed with the best principles of collaborative programming, user-centered design, and the most advanced scripting languages and database architectures. We were reasonably well-funded for this development work. But the complexity of the scripting and database together required administrative support, servers, system administration, programming fixes and constant tweaking whenever new security patches were installed on servers, when server migration happened, when scripting languages were upgraded and did not support earlier functions.

It is inevitably easier to secure funding for technological innovation and new development than for maintenance. Faced with unacceptable performance issues, we reluctantly took the Learning Record Online down. We referred users to the original Word document as a stopgap measure.

I thought about the next possible evolution of the Learning Record, and reflected on the ongoing net discussions in my field about the ethical questions around aggregating and archiving student work in “content management” systems and eportfolios. It seemed to me that these issues, as well as the problems with the web to database version of the LR, could be solved by developing a standalone version of the Learning Record in FileMaker Pro. Each student would have his or her own copy; it would be portable, and passed back and forth between the student and the teacher as work was submitted. All of the student’s work could be contained in this little database that did not require any server architecture or arcane scripting language.

This flew in the face of conventional administrative wisdom, and I was advised more than once that “everything” was evolving toward large, integrated monolithic courseware and content-management systems delivered over the web. I was urged to develop the Learning Record in this direction, find a suitable institutional home where it could be fully supported by the kindly IT folks who are administering such packages, and even to integrate it into a more established courseware application that was systemically supported, such as Blackboard. My instincts and my experience told me to run as fast as I could in the other direction.

I worked on this FileMaker Pro version for over a year, with a growing sense of urgency based on the enthusiastic response wherever it was presented. It reached the point where I had created very sophisticated structures to simplify the experience for users: ultimately there were about 69 layouts and 81 scripts. The layouts were appealing and attractive. It was truly a thing of beauty. And still there was the documentation for users to be created. User-testing for students and teachers with disabilities. Promotion and marketing and distribution, as well as consulting with faculty about its use.

Suddenly, this all seemed somehow archaic and vastly more complex than it needed to be, for a simple form that organizes information to document student learning in a class. The tool itself required more attention than the work it was intended to showcase.

I took a radical turn and returned to the simplest technologically-supported representation of the Learning Record’s functions: I created a new form in Microsoft Word, using some features that improve its stability and usability. Specifically, I implemented form fields so that at any time in the future, researchers could set up their own databases and export data from Learning Records (with appropriate permissions, of course); I implemented hyperlinking for navigation and to make connections between the Learning Record and the samples of student work; and I implemented form protection so that the fundamental structure and instructions of the Learning Record cannot be altered. The time required for this completely new, ground-up development was about a day and a half, in the midst of normal working days filled with meetings, appointments, and other work. Since then, I’ve even dropped the form fields as too cumbersome.

Radical simplicity wins. Now anyone can play. No net connection is required to use it, and students and teachers can transfer Learning Record files by any means they wish: email, flash drive, CD, servers, Blackboard, print on paper —any file transfer method will work. It will run just about anywhere on just about any hardware. It is fully compatible with other word processing applications, including Google Docs and open source alternatives such as OpenOffice and its derivatives. It is a nimble 120K in contrast to the FileMaker Pro version, which tipped the scales at 9.8 MB even without any student data in it, so the LR can now be easily sent as an email attachment. It is of course completely accessible for users with disabilities. It is easily printed out and easily read or scanned.

It’s not flashy to take this route, it is not sexy or showy. It is simplicity itself; and in this way, the Learning Record can have the widest possible distribution, utility, and life span, with the least administrative and technological overhead. No database servers, no system administrators, no security concerns, no extra ethical issues, no extensive documentation, no password glitches, no funding needs, no arcane, ever-shifting programming languages, no complex upgrade paths. Just teachers and students working in the simplest possible environment to best enable them to collect and see, together, the evidence of student performance and achievement. I think of this as the evolution of evolution: from unneeded, unwanted, and unsupportable complexity toward utter simplicity and utility.

Or, as Roy Bhaskar, the British social scientist and philosopher put it:

From the critical realist perspective...emancipation depends on the transformation of structures, not the amelioration of states of affairs. Indeed, in present and foreseeable circumstances, the transformation of structures may be a practically necessary condition for more humane states of affairs. But this transformation does not involve a magic transformation into a realm free of determination, as imagined by both utopian and so-called 'scientific' socialists. Rather, it consists in the move or transition from unneeded, unwanted and oppressive to needed, wanted and empowering sources of determination.(Reclaiming Reality)

Ultimately, the Learning Record model of organizing and presenting evidence of learning is technology-independent. The principles can be represented in any format. There is no reason it cannot be made available for PDA, cellphones, podcasting, or any other medium. In that sense, it is an open architecture for a process consistent across all disciplines, all media, all subject matter, all levels of ability, and all teaching methods. This adaptability, simplicity, and flexibility gives the Learning Record, in my view, the necessary fitness for survival in an increasingly complex technological environment. More importantly, it gives the Learning Record the simple appeal that invites more participation and willingness to play on the part of faculty and students.