Creating Online Experiences in Broadband Environments
The term broadband is used in this paper to mean more bandwidthówhich means faster connections for Internet visitors. This differs from the definition found in Websterís Dictionary: "of, relating to, or being a communications network in which a frequency range is divided into multiple independent channels for simultaneous transmission of signals (as voice, data, or video)." If you took this out-dated definition and applied it to the Internet, it would be defined as a broadband environment regardless of the connection speed. Broadband can include multimedia (sound, video, images, etc.), but the speed of the connection is what differentiates it from dial-up Internet access. Cable modem or DSL (digital subscriber line) connections are the most common types of "consumer" broadband connections.
Cable modems and DSL connections range from 385 Kbps all the way up to 40 Mbps. These connections are roughly 10 to 1200 times faster than a 56K modem connection. In addition, cable modem and DSL connections are on all the timeóthere is no "dial-up." This a fundamental change in how users interact with the Web: for broadband users, the Web is not only faster but "always on."
According to the Yankee Group (http://www.yankeegroup.com/), a technology research company, residential high-speed Internet services will grow to 3.3 million subscribers this year. The number of subscribers is expected grow to 16.6 million by the year 2004.
Spurring the growth of broadband
connections is the trend toward more multimedia on the Web. According
to the report, "Beyond T1/E1 1999-2000, North American Residential Broadband
Access Market Opportunities," published by Communications Industry Researchers,
While the number of individuals with high-speed access is estimated to grow considerably, only a small percentage of the total U.S. Internet population is currently connecting with broadband technology. According to the Computer Industry Almanac (http://www.c-i-a.com/), there are 110 million Internet users in the United States at present. This means that less than 3% of Americans are connecting using cable modems or DSL connections from home, which, of course, doesn't include school visitors or people who surf at work with high-speed connections. Still, the number of visitors connecting with broadband is much smaller than those signing on with dial-up.
Interestingly, the CIR report also predicts residential broadband connections will top 31 million by 2003. This is significantly higher (and a year sooner) than the Yankee Group's estimation of just over 16 million by 2004. So while Internet researchers disagree on the numbersóthe increase in broadband subscribers will be significant by any estimation.
Webcasting is the first technology to take full advantage of broadband on a wide scale. Large commercial Webcasting sites have, in the last year, developed their own "broadband sections" trying to appeal directly to users of the technology. Webcasting giant Yahoo! Broadcast, for example, has a broadband category (http://www.broadcast.com/broadband/).
Webcasting is the perfect technology for the emergence of broadband. Itís completely scalable: You can create one program and deliver high-quality video to broadband users, a few frames-a-second video to 56K dial-up users, and even an audio stream to a user with an antiquated 14.4K modem. In creating these different streams, youíre not only reaching virtually every Internet user who wants to tune in, but, in creating the broadband stream, youíre providing a way for future users to see a relatively high-quality archive of the program. So while broadband users may be few in number nowóthe broadband stream may be the one most-accessed in the future as more and more users connect at higher speeds.
Much of my own experience comes from helping to create broadband streams for Webcasts held at the Exploratorium in San Francisco. In August 1999, the Exploratorium sent a crew to Turkey to film the solar eclipse and send back video via satellite. This signal was encoded and one of the streams was designated for broadband visitors (http://webevents.broadcast.com/exploratorium/solareclipse/). The broadband stream was encoded at 300 Kbps or roughly at 9 times the rate of an encoded stream for a 56K modem. With a higher data rate, there is higher quality. The video seems less jumpy (more frames per second), clearer (less "lossy" compression), and has a larger window size. Following the Eclipse Webcast, The Exploratorium conducted a Webcast for the 10th anniversary of the Loma Prieta Earthquake (http://www.exploratorium.com/faultline/remembering.html) and a program on the Science of Wine (http://webevents.broadcast.com/exploratorium/wine1199/). Both of these programs have broadband streams. The Loma Prieta Webcast has a T-1 stream which is about 100K per-second (which, it could be argued, is not a "true" broadband stream). The Science of Wine has a 300K stream. For almost all major Webcasts, the Exploratorium now routinely creates at a 100Kbpsóif not a 300Kbpsóstream.
The only drawback to creating high-band streams is cost. Most of the expense is the cost of bandwidth. One hundred people connecting to a 300K stream are going to use 9 times the bandwidth that 100 people connecting to a 56K stream will use. Creating the broadband stream itself is simpleóit just requires that such a stream is encoded. Which may require an additional computer to facilitate the encoding for a live broadcast.
Streaming video can also be delivered "on demand." Video clips can be integrated into Web sites, and different versions can be delivered to visitors with different types of connections. RealNetwork's (http://www.real.com) new "G2" format allows developers to encode video clips which are automatically "scaled" based on the speed of a visitor's connection. The RealPlayer, which is used to play RealVideo files on the visitor's computer, contains settings that alert the server to the type of connection. RealNetwork calls this "SureStream" technology.
Recently, I developed a site for the Exploratorium and NASA's Sun-Earth Connection Education Forum focusing on this year's Solar Maximum. SolarMax 2000 (http://www.solarmax2000.com/) contains video clips of interviews with solar scientists that utilize G2 technology. During the encoding process I specified which types of connections were to be supported. Each encoded file contains optimized "versions" for 28K, 56K, 80K, and 180K connections. Obviously, the quality of each version improves as the amount of data increases.
Since the SolarMax 2000 site is meant to reach a mass Internet audience, any broadband content needs to be scalable so as not to turn away potential visitors. The streaming video fulfilled this primary need, while providing broadband visitors with better quality video.
Due to the relatively small number of current broadband users, there are only a few examples of museum sites created specifically for that audience. In the commercial realm, Quokka.com (http://www.quokka.com) has developed a reputation for providing Web-based broadband content. Other sites such Intelís Web Outfitters (http://www.intel.com) also provides a few content sites geared toward visitors with high-speed connections (and faster processors). These sites and others are free to pursue what is at the moment a "niche" market. Museums for the most part, with a mission to serve the public at large, have so far not pursued the emerging broadband market.
Museums have created broadband sitesóbut have been careful to include low-bandwidth versions as well. A recent example is the Virtual Smithsonian (http://2k.si.edu/), whose high-bandwidth version utilizes a variety of interactive and multimedia technologies. The low-bandwidth version, not surprisingly, uses these technologies much more sparingly.
While creating two versions of the same site is one way to insure that a site can reach all visitors, there are some drawbacks. Some broadband visitors may still choose to visit the low-bandwidth version anyway. Many visitors with slower connections may feel that they are missing out by not being able to access the more media-rich, high-bandwidth versions. Also, some new users of the Internet might not know the difference between high-bandwidth and low-bandwidth and may not know which choice to make.
Another significant drawback is the development time required to produce these sites. Obviously, creating two versions requires additional time. Also, there are more time requirements and costs involved in updating and maintaining two sites.
Broadband sites themselves generally use more interactive technologies such as Java (a platform independent programming language) and/or Shockwave or Flash (interactive animation and multimedia authoring programs). To develop using these technologies, specific technical expertise is required. Hiring or contracting a Shockwave or Java programmer or a Flash animator can be very expensive. The cost of developing a high-bandwidth site can be much more expensive than creating a low-bandwidth version.
With all of these drawbacks, developing multiple versions of sites is sometimes the only way to adequately present the content. In 1998, I developed an online exhibit entitled, "A Memory Artist" This exhibit contains drawings and paintings created by artist Franco Magnani and photographs by Susan Schwartzenberg. The paintings and drawings are of a small town in Tuscany and are done entirely from memory. The exhibit compares these works of memory with actual photographs of the same scene. In order to compare these images, each one needed to be large enough so that viewers could adequately observe the details. Also, each image needed to appear on the same screen. Those unable to view the exhibit or unwilling to wait for the images to download were given the option to view the text article (which contained fewer images).
To a certain extent, the "Memory Artist" exhibit automatically detected the capabilities (although certainly not the bandwidth) of visitors, providing them with the "exhibit" that best suited them. More complex "auto-detect" scripts can tell if visitors have multimedia capabilities by detecting browser types and versions, as well as plug-ins (which expand the multimedia capabilities of browsers).
While multimedia capabilities can be automatically detected, the method is not foolproof and it cannot detect the speed of the visitor's connection. Such a development would certainly simplify the creation of multiple versions of sites. As the gap between broadband Web users and dial-up users continues to grow, a method for detecting the speed of a visitor's connection becomes more desirable.
It's certainly possible such a development could come to pass in the near future. Perhaps, in fact, browsers will be able to communicate with servers in the way the RealPlayer doesóby letting them know what type of connection they have.
Considering the problems that exist in developing multiple versions of sites, perhaps the best method for designing sites is to create sites that are flexible and contain scalable media elements. These elements may scale automatically (like some streaming media) or the user can select different versions of media files. In other words, the structure and some of the contents of sites are available to everyoneóbut some of the individual media files may require or work best with broadband connections.
Providing the visitor with choices for high-bandwidth or low-bandwidth media (or other components) means the overall structure of the site can be inclusive (and one version), while video, audio, images, and even interactive components can be scaled. These components would be one "click away" from the general structure of the site. Another benefit of this approach is the amount of flexibility it provides. Low-bandwidth visitors can easily browse the overall structures keying in areas of particular interestóif they are willing to spend the time downloading a high-resolution image or a large video file. Or they may view a "low-res" version, and then decide to view the higher quality one.
As we've seen, streaming video (both Webcasting and "on-demand") provide a scalable solution that doesn't require the visitor to have to choose between versions and, importantly, doesn't require extra work for the developer. QuickTime video (generally a "non-streaming" high-quality digital video standard) and images cannot be "automatically" scaled. Different versions of these files would need to be created. Links to these files would be offered as choices to visitors.
Thumbnail images linking to high-resolution images have been used extensively on the Web for many years. The visitor can make the choice to view the larger image; generally the size of the image is listed, letting the users know what they are getting into. Broadening this to include the option of viewing even larger images with more detail is an obvious extension. In the museum world, visitors could view super-high-resolution artwork or scientific illustrations.
This technique can also be extended to audio or video. Streaming audio is already "scalable," but other audio file types and QuickTime video are not. It's easy to provide the user choices between high-quality or low-quality versions of the same clip. Creating additional versions of these files does require extra work and extra file management and storage. However, creating and editing the clip is usually the most time-consuming aspect of audio or video production. Saving different versions of the clip is easy and having a high-quality version provides more "value" for the time invested in creating and editing.
Sometimes "non-streaming" audio or video can also be used as a high-quality alternative to streaming media. One example of this technique is "Life Along the Faultine," a Web site which I helped design for the Exploratorium. The article on Loma Prieta (http://www.exploratorium.edu/faultline/loma_prieta) uses RealVideo and QuickTime to present news footage from the 1989 San Francisco earthquake. The RealVideo is scaled so that broadband visitors see better quality video than dial-up users do. The QuickTime versions allow broadband users to view even better quality video (low-bandwidth visitors can still download the clips, but it takes some time). All Web visitors, regardless of connection type, can still read the article and view the imagesómany of which are "clickable" to higher quality versions. The structure of the site is as flexible as possible, providing the user with options for interacting with video and images.
As I mentioned, creating different versions of media files does require more work on the part of the developer (with the exception of G2 RealVideo files). But compared to the amount of time involved in creating two versions of a site, the costs are minimal. Having these "broadband extensions" in place provides more options to current visitors and ensures that future ones will be able to view or interact with higher quality files. It should be mentioned that variations of this model are found in many places on the Web. In many ways, it's a common-sense approach. You build a structure that is completely accessible and then provide choices for visitors, offering content that takes advantage of new technologies only when it's the best way to present the material.
The model I've presented doesn't allow for "immersive environments." These are sites that create "virtual spaces" on the Web. These types of environments are created using interactive technology such as Java or, more often, Shockwave for navigation and the "building" of the space. Such tools can provide compelling navigation systems and interesting "spaces" for exploration, but the sites lack flexibility for low-bandwidth and "low-tech" visitors. Even if some of the contents of the sites could be "usable" for visitors, they need to meet the technical requirements of the navigational structure (the Shockwave plug-in and broadband connection) for visitors to view or interact with the contents.
Developing these immersive environments can require a considerable amount of resources. Space is important, but with limited museum resources it makes sense to focus on the exhibits or other content. Good Web design (just HTML and images), may not provide the visitor with as powerful an experience as an immersive environment, but good design can create "space" as well, and in most cases it is more functional, and it is always more accessible.
As the number of potential users with broadband connections increases, museums need to find compelling ways to reach them. As I've mentioned, some commercial sites already cater to these visitors, and certainly more sites will follow. If museums ignore broadband visitors, they risk losing many of them to more compelling experiences elsewhere on the Web.
At the same time, focusing too much energy on broadband visitors can be dangerous. With limited resources, museums need to focus on creating resources that are flexible and inclusive. The Web by its very nature is an "exclusive" environment. According to "Falling Through the Net: Defining the Digital Divide," a report by the "National Telecommunications and Information Administration" (http://www.ntia.doc.gov/ntiahome/fttn99/), more than half of the U.S. population, particularly poor urban and rural individuals, do not have access to the Internet. Obviously, creating additional barriers such as bandwidth restrictions or the use of emerging technologies further reduces your reach.
There is a real tension between creating compelling resources that take advantage of the latest technologies that the Web offers and reaching the public-at-large. Museums should be places where experimentation and innovation take placeóbut they also need to reach their audiences. In addition, the realities of limited resources and few online revenue models for museums can't be ignored.
The challenge for developers is to balance these competing needs and not be carried away by all the hype surrounding the broadband revolution. Common sense, compelling design, and thoughtful uses of new technologies can provide rich experiences for online visitorsóregardless of their connection speed.