StumbleUpon's Self Critique: Investigating the New Media Practice of Collaborative-Filtering

My Creative Poster Explaining the Project

The purpose of this blog is to present findings on a critical investigation of the New Media practice of Collaborative-Filtering, also known as Social Bookmarking/Tagging. This practice falls under the broader Social Media category of Collaboration/Authority Building.

Arguably, the most prevalent and popular web service to engage in collaborative-filtering is StumbleUpon (SU), a website that allows users to surf the internet based on their personal preferences. SU can also be downloaded as a toolbar, compatible with most internet browsers. SU directs users to various websites, which they can then choose to "like", "dislike" and write reviews on. Users can click the "Stumble" button on the toolbar to move on, discovering the next site. SU offers users the most modern, digitized equivalent to the process of television channel surfing.

The process of collaborative-filtering SU engages in, combines users' opinions with the machine-learning of preferences. It relies on the use of expert systems which are "computer systems designed to operate at the level of a human expert" (Green, par.6). This develops a community of like-minded surfers that engage in community-based surfing. Users ratings and recommendations heighten the community value of websites. Websites that receive positive reviews and "likes" get passed along and recommended to others. This creates a very social and hierarchical community of stumblers who value each others' opinions and share the same types of interests.

I had a number of questions running through my head about how SU generates quality content when I began my investigation. I knew the search engine was excellent at providing users with a bevy of fun and interesting content, but I wanted to know if it was also capable of gathering and presenting quality information. Do human opinions combined with expert systems create stronger networks of data than say, a general search engine like Google? The difference between the two is that SU recognizes, tracks and adapts to user's preferences over time.

Some of my other questions:

• What do these type of sites accomplish for the user? how do they impact the user's experience of gaining knowledge about a specific topic?
• What quality of research can one expect to find?
• What are the primary sources one is directed to via StumbleUpon? ie. blog pages, personal websites, organization's websites, video content, visual content
• Who/ what processes control the distribution of content? what are the power relations behind such sites?
• Does the content vary based on different internet browsers?
• How original and current, versus recycled is the content? Will a source ever come up more than once?
• How closely does StumbleUpon really adhere to your interests?

So the task was set out to test StumbleUpon's own ability to critique itself and its New Media techniques. I was interested in using the web service to mine data that was specifically categorized as "StumbleUpon". I started fresh, creating a new account. When it came to the point where I was meant to check my preferences (there's around 500 topics to choose from), I was very particular and chose only the topic of "StumbleUpon". That way I could be sure I was sending a clear message to the search engine that I was only looking for content that had to do with SU itself.

I started stumbling and tracked, analyzed and recorded each web address SU directed me to. I stumbled 50 times on the internet browser Firefox and 50 times on the Safari browser. I made detailed notes of my impressions and recorded whether the content presented a positive, negative, or neutral portrayal of SU, I also accounted for the content that was irrelevant to my topic.

I came across an array of materials, mainly web blogs, websites and visual images. These materials generated both criticisms and praise of the social bookmarking webservice. I thought I'd start off with discussing the visual images SU generated about itself. They present an interesting story of visual rhetoric based on the StumbleUpon culture.

Here is a selection of the positive visual portrayals presented by StumbleUpon about its own platform. 

 

 

 

 

These images cast StumbleUpon in a prolific light, celebrating it for its supposed achievements in online social traffic, popularity and intelligence. They communicate the message that SU has had a huge impact on online culture. In particular, the image of Albert Einstein generating the equation for SU is significant. It suggests that SU is the ultimate in online social networking, as other services such as orkut, ma.gnolia and del.icio.us are terms in the equation adding up to StumbleUpon.

Here are some of the images that stir up controversy about StumbleUpon.

As I stumbled I found that many visual images came up. This imagery certainly does make a strong comment on StumbleUpon culture and internet surfing in general. I found that these visual images were some of the strongest critiques of SU I came across. They have the power to say a lot in a straightforward way. These images helped guide my observation that people see SU as very harmful. SU is presented as an internet addiction and huge time waster. In these images internet surfing on the whole is cast as very negative, relegating people to the status of "nerds" and "hermits" who stay up all night, never leaving their screens. They also suggest that much of SU's content is trivial. These images offer important cultural sentiments about the usefulness of SU. 

Now I'll discuss the textual sources that SU offered. There was many positive "PR" portrayals of SU, via web blogs and tech websites but there was also some key sources that provided critical perspectives, assessing and questioning SU's usefulness.

Many of the results SU fed me that treated the web service in a positive light were independent from the actual SU organization. Often these were forms of internet marketing – “good press” for SU. In other words, these reviews came from an organization or company. One such example was a blog page that presented the following graphic, (at right, also in better detail above), which came from an internet marketing company called BlueGlass. This was actually the very first page SU offered me in my Firefox exploration. This well designed graphic boasts of the great numbers behind SU’s social media network. According to this, SU is second behind Facebook for global usage and ranks first “among social media traffic sources in the US”.

 

Another category of positive reviews was bloggers focused on enlightening readers on useful SU features people may not be aware of, such as the “search feature” on the tool bar. I came across titles such as “The Most Useful StumbleUpon Feature You May Not Have Known” and "StumbleUpon for Dummies". Not sure if I’m missing something here… I wouldn’t consider myself a Stumble “Pro” but don’t most people using the internet understand that a search box lets you search specific terms? Many of these types of positive reviews focused on reiterating basic SU concepts. Perhaps at this point SU thinks I’m not experienced enough yet and may need tips. 

I quickly noticed the pattern of SU directing me to various pages on its own site. Often the purpose of these pages was to encourage me to add more preferences to my interests and search for new content. Then it started to recommend specific topics to me. I guess SU was trying to build up my profile of interests, to encourage me to do more stumbling.

Screen shot while stumbling: SU suggesting I choose more topic of interests

SU didn’t offer a wide array of critical portrayals of its own web service, but a few quality ones came up. A humorous student newspaper article from the University of South Carolina stressed how "randomly skimming internet pages leads to hours of aimless, uncontrollable surfing and wasted time" (Solley). Later in the article the author asserts, "Stumble has a way of taking hold of its users and not letting go. Not every site that you are directed to is interesting — some information is boring and absolutely pointless — but I think that keeps you obsessed. You continually stumble through random knowledge hoping to find that one perfect Web site — the Mecca of the Internet." Although he raises valid concerns, the perspective is described through the vehicle of humour, thus downplaying the seriousness of the critique. It's interesting that SU often presents information through humorous ways. What does this say about how SU views the preferences of its audience? It seems like its always trying to offer the user something less serious and more fun, which may be why its gotten its wrap as a web service for aimless entertainment.

Something on the side of positive critical analysis for SU came up. A quick blurb from the website, Read Write Web pointed out the fact that although SU is in its own right is a social media behemoth, it’s marketing is a great deal less in your face than other sites such as Twitter. Certainly says something positive about how Stumble has gone about presenting themselves.

Another area of interest - who are Stumble’s competitors? While stumbling I was presented with an article titled “Massive Censorship Of Digg Uncovered” about some bad press for Digg (another similar social media website that relies on user rankings). It’s interesting that this came up because it casts the community of Digg as manipulative and engaging in censorship. Later I was presented with another recent article that contrasted the decline of Digg’s traffic with the rise of SU’s. Interestingly nothing ever came up that addressed the sentiments of SU’s own community of members. 

During my exploration on Firefox I found that random content didn't come up as often as I expected. However I found while stumbling on Safari there was a great deal more random content. This makes me question the process SU uses to generate content. I only was interested in "SU" itself yet I got these types of websites. I’ll admit, sometimes you embrace aimless internet entertainment. As Richard MacManus of Read, Write Web suggests, “the most popular content on StumbleUpon tends to be easily digestable and entertaining. Lists, bizarre things, scientific discoveries, animals, humor, images, and so on” (par.6). The downside of stumble is that, depending on your preferences (or perhaps maybe not), it can offer you too much of a good thing - a never ending sea of time wasting content and four hours you’ll never get back... 

Stumble also promoted other major internet sites and services by directing me to Wikipedia, Flickr, Firefox, and Time Magazine. Stumble has the power to promote both itself and these other internet identities in a powerful way. 

Screen shot while stumbling: SU promoting itself and Firefox through the new Firefox add-on

Although most of the critiques SU offered me were not very critical of SU's usefulness, a few key sources did articulate a critical stance. In particular this article “How to Become a Top StumbleUpon User (or Why You Really Shouldn’t Bother)”, highlighted that many people use Stumble to generate traffic for their own sites. The author’s main point was that “content determines StumbleUpon visitor traffic, not any special power within your SU account”. This points to a main issue of SU: some people only care about using it to build a profile, while others stumble simply for fun. The author suggests “don’t worry about being a top stumbler/power user and stop trying decipher the SU code. Focus more on learning, using and appreciating how StumbleUpon can and will work on a content level.” From reading this article I realized just how large and hierarchical the SU community is. However it seems as though quality of content is really the main driver of attention paid to sources by other uses, not the status of who recommends it.

One last area of interest before I draw some final conclusions. At the start of this blog I mentioned the larger theme of transitioning from channel surfing to web surfing. One of the last sites SU directed me to was the article, "Murda He Wrote: StumbleUpon May Save America from Decline to Illiteracy" by Mark MacMurdo. It captures this idea in a interesting way. The author comments on the rise of American rates of television watching. He then goes on to suggest that SU is a much more effective version of time wasting than TV surfing. Here are some key quotes:

• “Instead of taking breaks from work to watch television, I've found myself Stumbling in short breaks”

• “Even though I substitute watching television with StumbleUpon, it's not a fair trade. Stumbling isn't just another form of entertainment — it makes me a better person.”

• “There is something to be said about engaging in "active" entertainment over just having sounds and images thrown at you."

 

This article touches on several main tensions that are of significance as people transition from spending their time "flipping channels" through outlets of traditional forms of media, to "flipping channels" in New Media environments. The author asserts that Stumbling is more beneficial than channel surfing because its activity is more active, requiring you to read webpages rather than watch television stations. So is flipping through Stumble and other related social bookmarking websites more educational than television surfing? Through the process of "tagging", there's no doubt online content is more personalized. I think it can lead to some useful sources, but as I learned, even when you have specific preferences, SU still gives you a lot of irrelevant content. So perhaps, as author Clay Shirky asserts, the overall process of tagging is valuable and will become better with time. But the content of SU seems to be quality "entertainment" over quality/useful/academic information. In other words, SU's user recommendations system is very effective at generating sources that many people enjoy, but that's just it - the content is more directed towards entertainment than serious literature. However we can see the potential value that collaborative-filtering has for organizing databases based on human opinions rather than generic search terms.

CRITICISMS

•       Alot of people describe stumble as an “addiction” and “time waster”, thus connoting that the action of web surfing through the vehicle of SU is more detrimental than useful

•       SU does indeed offer a lot of “random” time wasting content. in my experience it was completely irrelevant and distracting to my topic. Yet maybe this “random” content is just something you have to accept if using SU. Perhaps there is no way to avoid it based on their algorithms.

•      I found SU itself had a heavy presence in my stumbling (this could have been due to the specificity of my preferences). But SU was constantly trying to get me to widen my preferences and check out other parts of their services.

•      Overall my stumbling exploration did result in some critical opinions, but most of the negative portrayals were visual images that were also meant to be humorous.

•      Thus in conclusion, SU critiqued itself in a fairly gentle way, poking fun at the cultural sentiments surrounding its usefulness. For the most part it offered positive feedback.

•      There were a few key sources that reiterated the negative stereotypes of SU, but also addressed more serious questions such as:

• Is web surfing a more efficient time wasting process? Can one enjoy this process more and learn more using collaborative-filtering web services versus television channel surfing?

• What does it say that SU is 2^nd only to Facebook globally in terms of social media traffic? Can we define social media and social bookmarking processes as in a completely new category of cultural activity?

• Does SU truly have a unique/efficient/ effective system of collaborative filtering? How do we differentiate between general search engines such as Google and what SU is doing?

• What’s the future for collaborative-filtering technology? Clay Shirky makes an important point: “It's all dependent on human context. This is what we're starting to see with del.icio.us, with Flickr, with systems that are allowing for and aggregating tags. The signal benefit of these systems is that they don't recreate the structured, hierarchical categorization so often forced onto us by our physical systems. Instead, we're dealing with a significant break -- by letting users tag URLs and then aggregating those tags, we're going to be able to build alternate organizational systems, systems that, like the Web itself, do a better job of letting individuals create value for one another, often without realizing it.”  

SU is an excellent web service for mining your personal interests, following the recommendations of others with similar interests and offering you casual sources of information. Its “good press” and traffic statistics prove that SU holds an important place in the New Media landscape. It has surpassed other social networking/ collaborative-filtering web services in popularity, which points out it is an effective and enjoyable New Media practice. That being said, I believe that currently SU’s popularity rests in its escapism – it allows people “down time” on the internet, separate from their more goal oriented internet activities.

 

Through completing this project I gained deeper understanding of the power of combining human opinions with machine learning. As Clay Shirky suggests, there is great search power behind the process of “tagging”, “liking” and “recommending” content to other like-minded people because it places value back in the human individual and their meaning making processes. We’ve learned in our class about Computational Media Aesthetics, and that one of the challenges facing the effectiveness of search engines is semantics – or the ability to understand human-created meanings. Clearly SU is a forerunner in collaborative-filtering technology. Its sheer popularity and massive community prove that it is definitely effective at generating meaningful content. Imagine SU creating a web service focused on generating quality academic content. This could redefine how we conduct online research, shifting away from general search engines to human recommended sources.

It is interesting to muse about the potentiality of human meanings + machine learning algorithms and strategies, and how this may impact the New Media landscape in the years to come. Group consensus and amateur production on the internet is a tremendously powerful thing. As professor of Communications Henry Jenkins suggests, “the Web has made this layer of amateur production more visible again, providing an infrastructure where amateurs can share what they created with each other: this ability to share media has helped to motivate media production, resulting in an explosion of grassroots expression”(par. 9).

We can see that the collaborative-filtering processes involved in online web surfing are a great deal more complex, useful and fulfilling to the user than television channel surfing. There is essentially an endless option of paths and routes to explore online as opposed to the linear navigation of television channel surfing.

To leave you with some final words by Henry Jenkins about the greater access to cultural production, created by New Media technologies, “the result has been the diversification of cultural production. Each new technology spawns a range of different uses, inspires a diversity of aesthetic responses, as it gets taken up and deployed by different communities of users. Such transformations broaden the means of self and collective expression” (par. 4).

StumbleUpon has contributed to both self and collective expression in innovative ways. It will be interesting to see how SU evolves, develops and adapts to a changing New Media landscape in the future.