Dissecting Web 2.0 Examples: Chapter 3 - Web 2.0 Architectures

by James Governor, Duane Nickull, Dion Hinchcliffe
Web 2.0 Architectures book cover

This excerpt is from Web 2.0 Architectures. This fascinating book puts substance behind Web 2.0. Using several high-profile Web 2.0 companies as examples, authors Duane Nickull, Dion Hinchcliffe, and James Governor have distilled the core patterns of Web 2.0 coupled with an abstract model and reference architecture. The result is a base of knowledge that developers, business people, futurists, and entrepreneurs can understand and use as a source of ideas and inspiration.

buy button

“Web 1.0 was about connecting computers and making technology more efficient for computers. Web 2.0 is about connecting people and making technology efficient for people.”

--Dan Zambonini

So, what actually changed between the emergence of Web 1.0 and Web 2.0? In this chapter, we’ll compare Web 1.0 companies and technologies with Web 2.0 companies and technologies to begin developing the design patterns that distinguish them. We use the term “Web 1.0” to refer to the Web as it was understood in the period of around 1995–2000, though obviously it’s not a simple matter of dates. To help us get started, Figure 3.1, “Tim’s list of Web 1.0 versus Web 2.0 examples” shows again the list of Web 1.0 and Web 2.0 examples that Tim O’Reilly and others compiled during an initial brainstorming session to get a “feel” for what Web 2.0 was.

Figure 3.1. Tim’s list of Web 1.0 versus Web 2.0 examples

Tim’s list of Web 1.0 versus Web 2.0 examples


It’s important to note that some of the Web 1.0 companies included in Figure 3.1, “Tim’s list of Web 1.0 versus Web 2.0 examples” have evolved substantially since Tim made his original comparison. For the latest information, definitely visit each company’s website. Tim’s choosing them as examples actually speaks to their success in that earlier age.

Although several of the companies we use as examples in this chapter are large enterprises or corporations, the patterns themselves don’t apply only to enterprises. In fact, the value of patterns is that you can remove them from an enterprise context and reuse them in other applications. For example, Service-Oriented Architecture (SOA) is a pattern of exposing capabilities to potential end users. Whether the SOA pattern is used by online gamers to access the states of each other’s joysticks or by large enterprises to reach into their customer relationship management (CRM) systems and provide users of their websites with rich interactive experiences, the core pattern is the same when abstracted to a high enough level. In both cases, a service offers some functionality or capability that another entity consumes.

DoubleClick and Google AdSense

Before we compare these two companies, we must point out that DoubleClick has vastly enhanced its platform since it was formed; so much so, in fact, that Google acquired DoubleClick in 2007 to further broaden its media advertising ambitions.[32] Therefore, instead of specifically illustrating DoubleClick’s original ad model, we’ll illustrate the generic pattern of banner ad impression sales that many online advertising companies used in the late 1990s.

Applicable Web 2.0 Patterns

Watch for illustrations of the following patterns in this discussion:

  • Software as a Service (SaaS)

  • Mashup

  • Rich User Experience

  • Semantic Web Grounding

  • Asynchronous Particle Update

You can find more information on these patterns in Chapter 7, Specific Patterns of Web 2.0.

Advertising in Context

Banner ad placement originally operated on a simplistic model whereby advertisers purchased banner ads in lots (typically of 1,000 or more), and the banners were then placed on websites. The placement of these banner ads was often billed based solely on impressions, regardless of whether anyone actually clicked on the banners. This online advertising model clearly had room for improvement.

Initially, one of the main issues facing advertisers was the lack of any guarantee that the ads were effective; however, this problem was mitigated by the use of tracking software and new business models that charged based on the number of click-throughs. Another issue concerned the fact that some larger companies offering such services asked webmasters to place code in their sites and then served up ads whenever someone issued a request for a page containing that code. It was therefore quite possible that ads aimed at golfers, for example, might appear on fishing or other websites not concerned with golf. The placement pattern looked a lot like Figure 3.2, “Basic pattern of banner ad placement”.

Figure 3.2. Basic pattern of banner ad placement

Basic pattern of banner ad placement

In contrast, Google AdSense is a paid ad service that serves contextually specific ads on web pages and tracks the number of clicks on each ad by visitors to those pages. This form of ad delivery uses a simple yet effective pattern of contextual targeting. Rather than just advertising blindly, AdSense attempts to quantify the context of a user’s experience based on a keyword score within the web pages containing the ads. AdSense then cross-references the keywords with a list of potential target ads that might be of interest to the user of that web resource. As a result, visitors to a web page on golfing will typically see golf-related advertisements rather than completely random content. AdSense also lets web page owners filter out competitors’ ads. For example, a golf club manufacturer could block competing companies’ ads from being displayed on its website. This is a highly useful pattern for preventing competitors from targeting a website owner’s customers. Figure 3.3, “Contextual serving of ads based on user profile patterns” shows an example of this pattern.

Figure 3.3. Contextual serving of ads based on user profile patterns

Contextual serving of ads based on user profile patterns

Also attracting website owners to AdSense is the fact that ad revenues are split between Google and the website owner. Other banner ad companies also use this revenue model, but AdSense users have a better chance of increasing their revenues because the ads on their sites are contextually specialized for their audiences, so users are more likely to click on them. Given the fact that net revenue from a website must be calculated once the costs of hosting are subtracted, it makes more business sense to go with a contextual pattern such as that offered by AdSense than with a non-contextual pattern.

A Peek at the Future of Online Advertising

Serving contextually specific information based on a single site visit is only one aspect of how online advertising is changing. With the evolution of the Internet and some underlying technologies, the science of targeted advertising is reaching new heights. Dr. Usama Fayyad, chief data officer and senior vice president of Research & Strategic Data Solutions at Yahoo!, stated in the March 2007 issue of Business 2.0 magazine, “I know more about your intent than any 1,000 keywords you could type.”[33] He knows this because of his Yahoo! research into the click-stream consciousness of web users. Dr. Fayyad is an actual rocket scientist who worked at NASA’s Jet Propulsion Laboratory before moving to Yahoo! to manage the roughly 12 terabytes of user data—more than the entire content of the Library of Congress—that Yahoo! collects every day.

Yahoo! tracks user behavior with a multitude of technologies, including cookies, user account activity, bounce rates, and searches. The major search engine vendors have acquired the ability to build comprehensive user profiles based not just on contextual information from a single web page, but on many aspects of a user’s behavior. For instance, Yahoo! and Google have created services that consumers can use to help them build successful businesses and/or websites. Along those lines, Yahoo!’s acquisition of Overture let people target search terms based on available inventory. Overture’s tools can tell an advertiser how many people search for a specific term in a given month, as well as suggesting similar terms.

Another trend in Web 2.0 advertising is the move away from traditional graphic banner ads and toward text and video media. Bandwidth-light text has an advantage thanks in part to the increasing use of cell phones as people’s primary devices for connecting to the Internet. Jupiter Research reported a trend of growth in all three categories (text, graphical banners, and video), implying either that there are more advertisers or that advertisers are continuing to pull financial resources away from traditional media such as television, magazines, and newspapers.[34] This phenomenon must be somewhat scary to the incumbent media giants, especially when coupled with the recent history of small upstart Internet companies becoming the largest media sources within just a few years (YouTube and MySpace are good examples).

A third trend concerns the delivery of ever more targeted content. Engaging users in a context in which they’re open to an ad’s content requires walking a narrow line. Many bloggers make a few dollars a month on Google AdWords, but some deliver a more immersive experience, even using text ads within RSS feeds and carrying ads into their readers’ aggregators. This can be effective, but if consumers are bombarded with ads, the entire mechanism starts to void itself out, as the human mind starts to filter out too-frequent advertisements.

Web 2.0 hasn’t done much about the pressing question of how society will continue to react to ads that are often perceived as intrusive and unwelcome. We’re referring to one of the most hated words since the dawn of the Internet: spam.

Email spam is probably the most despised form of advertising on the Web. Despite numerous mechanisms (such as spam filters and legislation) to control it, spam is still rampant. Like spam, banner ads also permeate many corners of the Internet and are common on many web pages. Do banner ads drive people away or do they provide value? Users have expressed time and again that sites that are uncluttered with commercial messages are more attractive. Google was widely heralded as setting a new model for search engines with a simple, noncommercial interface, although web historians could point out that Alta Vista was just as clean in its lack of commercialism. Consumers and users flocked to Google.com when it launched: it provided information they wanted without bombarding them with advertising. Similar models have evolved from companies such as Flickr (discussed in the next section), although the old-world ways of commercial ads still permeate much of the Internet landscape (even to pervasive levels within newer presences such as YouTube and MySpace).

Some have even organized communities to fight advertising. The most notable is Adbusters, an organization based in Vancouver, Canada. Adbusters is a global network of artists, activists, writers, pranksters, students, educators, and entrepreneurs who want a new social activism movement piggybacked on the information age. Their goal is simple: anarchy. Their aim is to topple existing power structures and forge a major shift in the way we’ll live in the 21st century. In a similar vein, Canadian film producer Jill Sharpe released a documentary called Culture Jam,[35] a stab back at our mainstream media and advertising agencies. “Culture jamming” is a form of public activism that is generally in opposition to commercialism.[36]

Despite these challenges, Google AdSense delivers a service that many people use and serves ad content that can be mashed into most websites. Google has provided many other value ad services that make it easy for anyone to become an advertiser and get some value for her budget. Google’s stock price continues to rise as a reflection of its perceived strength and dominant position in the new advertising industry.

Ofoto and Flickr

Ofoto began life as an online photography service based in Berkeley, California. The service provided three basic features:

  • It let people upload JPEG images so that others could view them by simply visiting the website.

  • It let people create photo albums and share them online with friends.

  • It enabled users to purchase prints online. This feature was supposed to be the foundation for Ofoto’s business model, which was based on the premise that people would want traditional, printed photographs.

Ofoto later added a 35mm online film processing service and an online frame store, as well as some other services, but its core pattern still embraced a core model of static publishing. In May 2001, Eastman Kodak purchased Ofoto, and the Ofoto Web Service was rebranded in 2005 as the Kodak EasyShare Gallery.

Flickr is another photo-sharing platform, but it was built with the online community in mind, rather than the idea of selling prints. Flickr made it simple for people to tag or comment on each other’s images, and for developers to incorporate Flickr into their own applications. Flickr is properly a community platform and is justifiably seen as one of the exemplars of the Web 2.0 movement. The site’s design and even the dropped e in the company name are now firmly established in Web 2.0’s vernacular.

Applicable Web 2.0 Patterns

This comparison involves the following patterns:

  • Software as a Service (SaaS)

  • Participation-Collaboration

  • Mashup

  • Rich User Experience

  • The Synchronized Web

  • Collaborative Tagging

  • Declarative Living and Tag Gardening

  • Persistent Rights Management

You can find more information on these patterns in Chapter 7, Specific Patterns of Web 2.0.

Collaboration and Tagging

Flickr is often used as an information source for other Web 2.0 platforms or mechanisms. It offers simple application programming interfaces (APIs) for accessing its content, enabling third parties to present images in new contexts and to access and use Flickr’s services in their own mashups or other applications. Bloggers commonly use it as an online photo repository that they can easily connect to their own sites, but the APIs offer much more opportunity than that. Programmers can create applications that can perform almost any function available on the Flickr website. The list of possible operations is vast and covers most of the normal graphical user interface’s capabilities.


Flickr also lets developers choose which tools they want to use to access its services. It supports a REST-like interface, the XML Remote Procedure Call (XML-RPC), and SOAP (and responses in all three of those), plus JSON and PHP. For more, see http://www.flickr.com/services/api/.

Developers can easily repurpose Flickr’s core content in mashups, thanks to its open architecture and collaborative nature. A mashup combines information or computing resources from multiple services into a single new application. Often, in the resulting view two or more applications appear to be working together. A classic example of a mashup would be to overlay Google Maps with Craigslist housing/rental listings or listings of items for sale in the displayed region.

Flickr’s API and support for mashups are part of a larger goal: encouraging collaboration on the site, drawing in more users who can then make each others’ content more valuable. Flickr’s value lies partly in its large catalog of photos, but also in the metadata users provide to help themselves navigate that huge collection.

When owners originally upload their digital assets to Flickr, they can use keyword tags to categorize their work. In theory, they do this to make it easier to search for and locate digital photos. However, having users tag their photos themselves only starts to solve search problems. A single view of keywords won’t work reliably, because people think independently and are likely to assign different keywords to the same images. Allowing other people to provide their own tags builds a much richer and more useful indexing system, often called a folksonomy.

A folksonomy (as opposed to a top-down taxonomy) is built over time via contributions by multiple humans or agents interacting with a resource. Those humans or agents apply tags—natural-language words or phrases—that they feel accurately label what the resource represents. The tags are then available for others to view, sharing clues about the resource. The theory behind folksonomies is that because they include a large number of perspectives, the resulting set of tags will align with most people’s views of the resources in question.[37] The tags may even be in disparate languages, making them globally useful.

Consider an example. Say you upload a photo of an automobile and tag it as such. Even though a human would understand that someone looking for “automobile” might find photos tagged with “car” to be relevant, if the system used only a simple form of text matching someone searching for “vehicle,” “car,” or “transportation” might not find your image. This is because a comparison of the string of characters in “automobile” and a search string such as “car” won’t produce a positive match. By letting others add their own tags to resources, Flickr increases the number of tags for each photo and thereby increases the likelihood that searchers will find what they’re looking for. In addition to tagging your “automobile” photo with related words such as “vehicle,” “car,” or “transportation,” viewers might also use tags that are tangentially relevant (perhaps you thought the automobile was the core subject of the photo, but someone else might notice the nice “sunset” in the background and use that tag).

With this in mind, how would you tag the photo in Figure 3.4, “How would you tag this photo?”?

Figure 3.4. How would you tag this photo?

How would you tag this photo?

We might tag the photo in Figure 3.4, “How would you tag this photo?” with the following keywords: “mountain,” “bike,” “Duane,” “Nickull,” “1996,” “dual,” and “slalom.” With Flickr, others can tag the photo with additional meaningful keywords, such as “cycling,” “competition,” “race,” “bicycle,” and “off-road,” making subsequent searches more fruitful. Semantic tagging may require more thought, but as a general rule, the more minds there are adding more tags, the better the folksonomy will turn out.

Flickr has also built an interface that lets people visiting the site see the most popular tags. This is implemented as a tag cloud, an example of which appears in Figure 3.5, “Flickr tag cloud, from ”.

Figure 3.5. Flickr tag cloud, from http://www.flickr.com/photos/tags/

Flickr tag cloud, from

The tag cloud illustrates the value of a bidirectional visibility relationship between resources and tags. If you’re viewing a resource, you can find the tags with which the resource has been tagged. The more times a tag has been applied to a resource, the larger it appears in the tag cloud. You can also click on a tag to see what other assets are tagged with the same term.

Another advancement Flickr offers is the ability to categorize photos into sets, or groups of photos that fall under the same metadata categories or headings. Flickr’s sets represent a form of categorical metadata rather than a physical hierarchy. Sets can contain an infinite number of photos and may exist in the absence of any photos. Photos can exist independently of any sets; they don’t have to be members of a set yet can be members of any number of sets. These sets demonstrate capabilities far beyond those of traditional photo albums, given that each digital photo can belong to multiple sets, one set, or no sets. In the physical world, this can’t happen without making multiple copies of the photos.

Pages: 1, 2, 3

Next Pagearrow