Reconciling Information-Seeking Behavior with Search User Interfaces for the Web by Daniel E. Rose/ Reviewed by J. Colannino

For a fellow who is an expert in search engines, Daniel E. Rose’s biography isn’t very findable on the web. At least not if you search with Yahoo, Alta Vista, or Google. This is interesting, because he has worked for two of the three firms and his stated expertise is to improve search results for users. Maybe his biography is buried somewhere in the 740 Google returns for “Daniel E. Rose,” or one of the 297 returns for Yahoo, or somewhere on Alta Vista. But of course, being an expert on search behavior he should know that no one really looks past the first page of returns. It didn’t matter whether I qualified his name with “biography” or “curricula vitae,” I just couldn’t find his bio, even on Wikipedia. He is referenced multiple times in a book about natural language processing, but its text was not searchable within Amazon.com’s search-inside-the-book tool. Google Book Search did allow for searching, but there he was listed in references only. However, that did lead me to a book that he did author: A Symbolic and Connectionist. But I could find nothing indexed regarding a biography or an “about the author” section.

Here is what I was able to glean about his career: according to this reference he worked for Apple. UC Berkeley listed him as Dr. Daniel E. Rose and gives video of one of his presentations here. The subject article lists him as working for Alta Vista. I found both the video presentation and this subject article a bit too commercial for my tastes; it looks more like a commercial for Alta Vista’s Prisma search engine then anything else. However, maybe there is something to Prisma, because using it was the only time I actually found something of a biography for Dr. Rose. It was for the Ordinology Consulting Group where he is a principal, and you can see his C.V. here. At this point I wanted to play the Hallelujah Chorus by Handel. Unfortunately, Prisma only gave me access to paid downloads. However, Google linked me to a free sample clip. If you click the link you can find it in an alphabetized list under “HANDEL ‘Hallelujah’ chorus from Messiah” where you can download the MP3. It would be nice to play right now and to loop continuously as you read on.

Go ahead… I’ll give you a moment.

I think all of these user-interface guys have it wrong. We need systems that will accommodate us not only when we are in our initial “webtarded” stage, but also after we become proficient. That means two interfaces – the cryptic command line Boolean string stuff for experts and the pretty picture point and click interface. Why can’t we have both? There is a paper I wrote that discusses this. It concerned computer operating systems specifically, but I think that a variant has relevance for search engines as well. You can find it here. Then you can tell me what you think about my idea.

Okay, you should probably shut-off the Hallelujah Chorus now.

Text vs. GUI – a False Dilemma/ Joseph Colannino

[Note: This is a reprint of a paper submitted in fulfillment of the requirement for a current issues paper for KM/LIS 5043 in July 2006. I have included it here to be used as reference material for another blog.]

The text-vs.-GUI dilemma is here to stay because it is irresolvable; user interfaces need both text and GUI. The text-vs.-GUI debate has been incessant since the GUI’s first major appearance with the advent of Apple’s Macintosh computer in 1984. A keyword search for text + GUI + interfaces resulted in more than 5 million hits on Google. A search under the coordinated phrase “user interface” indicated 201 million hits. Searching on the established phrase “human computer interaction” (a.k.a. HCI) gave nearly 30 million hits. Thus, the user interface continues as a mainstream topic worthy of academic and organizational interest. For example, W3C now hosts a user interface domain whose purpose is “to improve all user/computer communications on the Web (W3C UI),” and Gary Perlman maintains an extensive HCI bibliography at hcibib.org (HCI). But has not the issue of GUI vs. text been clearly decided?

A graphical user interface (GUI) now inhabits every personal computer, and text-based operating systems appear to have gone the way of the dinosaur. Alan Cooper summarized the general consensus more than ten years ago when he stated “The commonly accepted wisdom of the post-Macintosh era is that graphical user interfaces, or GUIs are better than character based user interfaces. This is a generally true statement…” (Cooper 1995). So where is the controversy?

Virtually all user interfaces contain a mixture of graphical and textual elements. However, the term GUI has come to mean an interface that uses a point and click structure as opposed to text entries on a command line. For example, the American Heritage online dictionary (Heritage) defines GUI as “an interface for issuing commands to a computer utilizing a pointing device, such as a mouse, that manipulates and activates graphical images on a monitor.” Yet, that most ubiquitous of graphically dominated venues, the worldwide web, continues to brood the text vs. GUI controversy. Witness Lesley Hensell’s 2003 article “Web Developer’s Dilemma: Text vs. GUI” (Hensell 2003). Her choice of wording inadvertently portends the subject thesis of this very paper; a dilemma, strictly speaking, is a choice between two unfavorable alternatives.

Hensell states that “with nearly a dozen large Web [sic] sites and complex e-commerce back-end systems, Encyclopaedia Britannica is one company that has forsaken GUI development tools in favor of text editors.” Nor is this an isolated instance. Hensell quotes Jay Cann, CTO of the web development firm Macquarium Intelligent Communications (Macquarium) to underscore the point: “We don't have any visual designers who just visually design – they’re all HTML coders.” But as reported by Hensell, Corel’s vice president of new ventures, Ian LeGrow, states that “HTML is definitely here to stay as the main vehicle for Web page layouts. But we're going to be seeing a pretty radical change from simple-to-edit, static HTML to more of an XML interface where [scalable vector graphics (SVG)] comes [sic] into play.” According to W3C’s SVG homepage (W3C SVG), scalable vector graphics “is a language for describing two-dimensional graphics and graphical applications in XML.” So what is the conclusion of these matters? Hensell summarizes in the last sentence of her article: “The balance of power may be shifting yet again, toward a new middle ground – and if it does, Web [sic] development tools no doubt will evolve to accommodate it.” The “new middle ground” is apparently the wedding of text and GUI capabilities in a single user interface.

Seeds of a dual interface have been around for a long time. As previously noted, all GUIs contain text, nearly all support text commands, and many contain more text than icons. Indeed, the mix is typically configurable by selecting which and how many icons one chooses to place on various toolbars (those horizontal strips of graphical elements near the top of the screen). To illustrate, consider the most ubiquitous of commands, the “save” function found in every application from the mundane (e.g., word processors and spreadsheets) to the esoteric (professional tools for specialty interests in the arts and sciences). Regardless of the particular application, GUI interfaces such as Microsoft Windows or Mac programs implement this in at least two different ways. One way comprises locating the disk icon and clicking it with the mouse. Another way is to type a command sequence – for example “alt+f s” in Windows. The latter sequence is completely executable without removing one’s hands from the keyboard. Indeed, it requires only one hand (the left) and is much faster than the point and click approach. Notwithstanding, many users, especially new users, will find the point and click sequence preferable because it does not require memorization of any command structure. The point is that differing users have differing user interface preferences.

Indeed, the GUI is not as obviously superior as Cooper’s truism may lead one to believe. For example, Mark Daoust illustrates in his online article “The Surprising Truth about Ugly Websites” (Daoust) that ugly websites often outsell pretty ones, or as he states succinctly: “Ugly sells.” Daoust notes that Plenty of Fish, a free online dating service (Plenty of fish) generates $10,000 per day in advertising revenue despite being “a very plain looking website.” Daoust believes that plain websites convey several important features: trust, functionality, and simplicity. He believes that glitz can sometimes be mistaken for slick and shady marketing while simpler websites exude a homegrown integrity. In support of functionality and simplicity, Daoust cites the Google search engine, The Drudge Report (Drudge), and Craigslist (Craig). In fact, the latter two websites are virtually all text. All are commercially very successful.

This author (Colannino) envisions a dual system where a command line at the toolbar level (similar in appearance to the subject line of an e-mail) gives the text equivalent of the user’s GUI selections as they are being made. Entry would be allowed in either way, and both methods would be fully supported and documented by online help with “show me” and textual instructions, and in offline documentation. GUI and text would be given no a priori bias. Both would always be visible and available. This differs from current Mac and Windows GUI interfaces that have incidental text and no command line. Once the user learned the text commands (gained fluency) there would be no need to interface with a GUI, except for unfamiliar actions. However, the GUI would always be available. The GUI would provide a consistent hierarchical interface; the command line would provide a concise flat interface. Apparently, this is not an original thought.

A Linux website (PCLinuxOS) offers the following advice in its online definition for “command line.” “‘Command Line’ refers to the entering of commands directly in a text mode using a terminal. For new users a graphical environment is easier to use but the command line is very powerful and definitely worth learning how to use as you become more experienced.”

Human language provides another direct though imperfect analogy: language students dream of foreign lands absent irregular verbs; intellectuals dream of concise and elegant expression of ideas and thoughts – the two are mutually incompatible. Neophytes require consistency. Experts desire economy and elegance. The war is between consistency and conciseness. Idioms and irregular verbs testify to conciseness and are the domain of the fluent. So it is in the land of computers; why should expert computer users be constrained to navigate the same hierarchy of menus in a GUI that novices must? Why should novices be required to memorize long lists of textual commands that serve only experts well? With a dual user interface (DUI) both graphical and textual command structures would be supported. Familiar commands would tend to be entered via the command line sequence while new commands and feature browsing could be done in a graphical environment. A convenient sequence, such as pressing both the “ctrl” and “/” keys simultaneously, would alternate between them.

The concept of a DUI is not well attested on the web. A Google search on the coordinated phrase “dual user interface” produced only 205 hits and the term was applied to mean different things. The following definitions had multiple attestations in the web literature; specifically, dual user interfaces have been defined as interfaces that

1. are cognizant of different user processing preferences or abilities (also called hypermedia) – for example, an interface serving both sighted and blind persons (Savidus and Stephanidis 1995),
2. produce two different scripts such as Arabic and English (Arabic Windows 95),
3. give different information based on viewing angle (Usernomics),
4. allow one to switch between graphical and text options (Flexus), and
5. present both graphics and text options simultaneously to the user (Silverlake).

The author uses DUI to mean the last option. Yet when the coordinated phrase “dual user interface” was qualified with the acronym “DUI,” it produced no hits. The author takes the wide variety of meanings and the absence of an established acronym to indicate a lack of consensus regarding the value of the dual user interface (definition 5) by the electronic community in general.

As the author has hoped to demonstrate, a user interface with both text and GUI elements would maximize the user experience and efficiency for both expert and newbie alike. The author would expect such an interface to have strong commercial possibilities and garner brand loyalty. Is this prescience or folly? Perhaps no degree of investigation is sufficient to answer now; at this juncture the best offering appears to be a proverb: time will tell.

REFERENCES

Arabic Windows 95, an operating system supporting both Arabic and English alphabets, http://support.microsoft.com/default.aspx?scid=kb%3Ben-me%3B498567 (accessed 27-July-06).

Cooper, Alan, About face: the essentials of user interface design. (New York: John Wiley and Sons, 1995), 41.

Craig, a regional classified advertising list, http://www.craigslist.org/about/cities.html (accessed 27-Jul-06).

Doust. The surprising truth about ugly websites, http://www.site-reference.com/articles/Website-Development/The-Surprising-Truth-About-Ugly-Websites.html (accessed 27-Jul-06).

Drudge, the Drudge Report, a news linking website, http://www.drudgereport.com/ (accessed 27-Jul-06).

Flexus, a website for COBOL programming tools, http://www.flexus.com/ (accessed 27-Jul-06).

HCI, a website compiling an extensive bibliography regarding the scientific field of human computer interaction, http://www.hcibib.org/ (accessed 27-Jul-06).

Hensell, Lesley, Web developer’s dilemma: text vs. GUI, E-Commerce Times, http://www.ecommercetimes.com/story/31596.html (accessed 27-Jul-06).

Heritage, the American Heritage Dictionary, 2004 ed. online, http://www.bartleby.com/61/26/G0302650.html (accessed 27-Jul-06).

Macquarium, a web design company, http://www.macquarium.com/ (accessed 27-Jul-06).

PCLinuxOS, a website devoted to the Linux operating system for PCs, http://www.pclinuxonline.com/wiki/DefinitionsTerminology (accessed 27-Jul-06).

Plenty of fish, a free online dating service, http://www.plentyoffish.com/ (accessed 27-Jul-06).\

Savidis, A., Stephanidis, C. Developing dual user interfaces for integrating blind and sighted users: the homer uims. Proceedings of the SIGCHI conference on Human factors in computing systems, (New York: ACM Press/Addison-Wesley Publishing Co., 1995), 106 – 113.

Silverlake, a software company producing a dual user interface according to definition 5, http://www.s-2a.com/smms_features.htm#3 (accessed 27-Jul-06).

Usernomics a software company producing a dual user interface according to definition 3, http://www.usernomics.com/news/2005/09/pcworldcom-sharp-shows-dual-view-lcd.html,

(accessed 27-Jul-06).

W3C SVC, the Worldwide Web Consortium, SVC website, http://www.w3.org/Graphics/SVG/

(accessed 27-Jul-06).

W3C UI, the Worldwide Web Consortium, user interface website, http://www.w3.org/UI/ (accessed 27-Jul-06).