Matthew Kampschmidt's 892 Weblog

One of the most popular websites to survive the dot com bubble burst was the online auction site eBay.com. It would seem that people like being able to buy and sell things directly with each other. One of the most important features of the site is it's reputation system which most people trust. Occasionally a news item appears on social news sites like Digg.com and Slashdot about an abuse of the ratings system but it's trusted overall. Chrysanthos Dellarocas, a professor at the Sloan School of Management at MIT, wrote a paper which discussed a few of the ways that such a ratings system can be abused. He identified 4 major types of abuse: ballot stuffing, or unfairly high ratings; bad-mouthing, or unfairly low ratings; negative discrimination, or when sellers provide good service to all but a few unliked customers; positive discrimination, or when sellers provide better service to preferred customers.

Chrysanthos identifies two major methods for combating these problems. Controlled anonymity would prevent ballot stuffing and bad-mouthing by keeping buyers from being able to provide ratings to people they knew. It would also likely prevent negative discrimination by keeping sellers from identifying buyer patterns. Unfortunately there are workarounds that people can use to get around anonymity so this solution sometimes fails. So instead the author recommends a "cluster" filtering solution which attempts to identify artificial inflation or deflation of ratings by targeting tight groups of ratings at abnormal distance from the average and eliminating them.

I think it would be extremely useful for eBay and other sites that rely on ratings to keep the user experience positive to implement a system like this. I think it would definitely improve the experience by removing the motivation to cheat the system.

Paper Cited: Immunizing Online Reputation Reporting Systems Against Unfair Ratings and Discriminatory Behavior [http://www.rhsmith.umd.edu/faculty/cdell/reputation.html]

One of the greatest dilemmas in online games such as World of Warcraft is expressed by the question, "Who do you trust?". Most of the time answering this question is a process of trial and error. The ideas expressed by Jensen et. al in their paper "Finding Others Online: Reputation Systems for Social Online Spaces" seem like they could be useful in this context. As the authors note, a 'positive' online interaction isn't just about avoiding 'bad' content, but about finding good partners. The paper puts forward a hypothesis that the the common sense idiom "we like those like us" will be a good predicator of good partners within a game. To this end they conduct some testing and find that, lo and behold, the common sense idiom is true!

While I'm hardly surprised by the outcomes of this study, I do feel that there was some important information demonstrated by it. Firstly, it showed that while similarity of interests is the main factor in determining if two people will play well together, another strong factor was how a person was rated by ones friends and if ones friends interact with another. Secondly it showed that while people didn't outright object to the privacy implications of such a system, they still declined to fill out a profile. I think the take-away lesson of this is that similarity ranking systems are useful, and that to get the most out of them we should not rely on self-ranking profiles, but perhaps on some other data collection method.

One of the most important factors in designing computer mediated communications systems is presence. Over the years designers have attempted to achieve a feeling of presence in their computer mediated communication systems through virtual reality simulators, video conferencing, emoticons, user avatars, and many other devices. A few Japanese engineers even developed a system to convey presence through smell. The paper "Designing Social Presence of Social Actors in Human Computer Interaction" posits that it is possible to increase a feeling of social presence in an application through modification of a computer generated voice alone. The authors of this paper conducted an experiment wherein they polled a subject group on their personalities using various personality tests, and then attempted to increase the feelings of social presence of a simple Amazon.com book description page by altering the qualities of a computer generated voice which read the book descriptions to users.

The study showed that several things increased the feeling of social presence: users reported higher levels of presence when the voices matched their own personalities, were extrovert type voices, or when the voices matched the content of the reading. All of this was achieved by modulating the volume, rate of speech, and basic tone of the computer generated voice. I found this idea fascinating and actually quite useful. Many product designers and innovators often go overboard when seeking methods for increasing their products appeal. This study shows that tweaking very simple qualities of a product can vastly change the feel of the product. I think that we need to keep that in mind and start to pay more attention to the details of our creations.

In a previous class I was given the opportunity to study context aware systems and found the idea intriguing. One device in particular, a context-aware music player, fascinated me. The idea of being able to have a device provide situationally appropriate musical backdrop based on location, weather, velocity, and a variety of other cues is very attractive. Similarly, being able to have a device assist me and my co-workers in maximizing our interaction based on context would be very useful. The concept of context mediated social awareness seems to have been latently present in many of our communication tools for some time. In particular, IM systems have provided this functionality in a very small way through manually updated status messages. Various scripts and plugins have been written to add more automation to this process by supplying information about music being played, physical location, or application being used. The AWARE platform clearly takes this a step further by integrating not only system resource usage to this status supply mechanism, but also extra information in the form of entity resources.

I think the most important aspects of this platform are minimal reliance on the user supplying context information, integration with other information systems, it's multi-platform approach, and it's respect for the privacy of the user. People often revolt against technology when it requires them to a) work harder, b) abandon old systems, c) use one and only one device, d) abandon their privacy. The AWARE platform seems to accommodate it's users in each of these categories.

In my search for a second article to review this week I found a commentary on enterprise knowledge management by Matthew Clapp. In this article Clapp makes an interesting observation: He states that all too often, corporate knowledge management teams design collaborative knowledge management tools with too much knowledge management, and not enough collaboration. This statement is backed by the observation that these systems often fail because the users get tired of the process of managing documents. If a collaborative knowledge management system requires 20 steps to share a document with another user, people will search for a quicker method that requires less work and allows for better collaboration.

I completely agree with this. In my experience some of the best collaboration tools have nothing to do with knowledge management. Some of the worst are all about knowledge management. The worst problem with some of these KM based systems are that they are often forced on corporate users via some enterprise policy about document archiving and management without any regard for what their actual impact will be. Clapp makes an excellent suggestion: companies should trial their collaboration and knowledge management with small groups who really need such a solution. This way they can ensure that the software is a good solution and fits with their organizational culture and workforce.

Source: Collaboration first, then Knowledge Management

As a recent addict of NewsFire and a long time user of Apple Mail, I really liked the ideas behind FeedMe. NewsFire is an excellent RSS feed aggregator for Mac OS X. It is a very powerful tool for viewing every bit of news I want without visiting dozens of websites. Apple Mail is Apple's email viewer (duh). Like FeedMe, it uses a näive Bayesian filtering system to eliminate spam. I also use a notifier application called Growl which works somewhat like the desktop component of the paper's software. It seems like FeedMe is an ideal combination of these three products along with the collaborative filtering smarts of NetFlix or Amazon.com.

As the sheer mass of digital information bombarding us grows we'll need new ways to help us manage it. This has been alluded to in previous weeks of this class, but has only been applied to the content we subscribe to, not notifications. I think this idea has great potential. While reading the authors description of how the glut of notifications is causing workers to lose track of work, I was nodding furiously. In every deskjob I've had I have had to click through at least a dozen notification boxes an hour. Having an application like FeedMe would help to keep me on track with my tasks.

While many of our article reviews are intended to explore the current state of the art in the field and discover new ideas, I felt that I should follow up last weeks review of "What is Web 2.0" with the history of the word and concept "Wiki". When first searching for information I decided to Google for the word "Wiki". As most technically savvy folk would expect, I was presented with quite a long list of entries. Fortunately, the very first entry seemed appropriate to me. The Wikipedia entry on Wiki. While this article was interesting, I actually followed a link at the bottom to a series of letters between a few people and Ward Cunningham, the coiner of the term "Wiki" as it relates to the software, as well as the inventor of the idea behind it.

The first letter on the page was from the etymologist for the American Heritage Dictionary, who was seeking information on the etymology of the word "wiki" which of course was given new meaning in 1995 by Mr. Cunningham. In his response to this etymologist, Cunningham noted that the word "wiki" meant quick in Hawaiian and that the first time he heard it was in reference to a "Wiki wiki bus" at the airport, which literally translated means "very quick bus". Cunningham found this word to be perfect for his new idea of a highly simplified networked community database. His idea was intended to allow for very quick editing of information. Thus he named his software WikiWikiWeb. In a response to another letter, this time from Oxford English Dictionary, he further clarifies his appreciation for the word as he notes that doubling a word in Hawaiian emphasizes its meaning, e.g. "wiki wiki" for "very quick". This seemed ideal to him because his software used doubling of characters for special meaning, such as a double carriage return for new paragraph, or a double single quote for italicized.

In his response to this editor from the Oxford English Dictionary Cunningham discusses the idea that he also named his concept Wiki because he didn't want a generic name such as "e-mail" applied to it. He felt that this unique concept should also have a strange and unique name. His particular interests lie in the use of language in computing and he noted that language used in computer applications, both in the final product and in source code, should be chosen systematically and should have meaning outside of a simple description. I think that this is well reflected in the best example of his work, Wikipedia, in that Wikipedia articles do more than describe something; Wikipedia articles reflect the people that write them and the times in which they are written.

Article review: Correspondence on the Etymology of Wiki

When our professor asked us to come up with ideas for our papers, my first thought was a review on Wiki technology and culture. It seems that people are beginning to rely on wiki's more and more these days, as evidenced by the rise in popularity of Wikipedia. I myself am hosting a wiki for another class right now, and it is being extensively used, much more so than our class blog. Thus this article certainly peaked my interest. Short and sweet, the general point of the article is that wiki's can be immensely useful to business organizations for knowledge management. This has of course been said before, but this article gave several examples of why naysayers concerns about wiki's are unfounded in the case of business applications.

When confronted with the idea that such an open KM system will have lower quality standards, the authors note that an active community will often spot mistakes just as well as a central information management team and that a business should have a relatively high trust level for contributors as they are specialists in their fields and were hired for their knowledge and abilities. The next argument against wiki's is a more potent one in light of many events on the Wikipedia. The case study company in the article felt that vandals would be a problem with using wiki's as KM solutions. This may be true of a large scale open community such as Wikipedia, but a corporation can require all wiki authors to sign and and could therefore verify who contributed to any such vandalism. As this would be a "career-limiting" move, few would attempt it. Also, the revision tracking capabilities of the wiki would make such actions pointless, as the wiki document could simply be rolled back to an earlier, un-vandalised version.

These arguments and others provide a potent reason for business' to go ahead and try wiki's, despite their misgivings. If it doesn't work out, then it won't be because of the standard problems with Wiki's in an open community.

I found two of my classmates posts very interesting today. The first was a post by David Maier about Open Source Collaboration. In it I noted that he talked about how an open source (or proprietary) control system is an inherently collaborative tool and that there is always a fear that you may alter or damage the work of another when you commit your work. It got me thinking about what we could do to alleviate that fear. Perhaps we could add some social aspects to the system, although that might encourage slacking off! The problem is an interesting one and while many companies and entities have tried to improve such code collaboration tools, I don't think anyone has tackled that social aspect.

The second post I found interesting was by Katie McDonald about a new project at MIT. This project's idea is to create a small device or "badge" that people would wear which could communicate with other badges to exchange information about their wearers. This sort of technology would allow a person to keep a running tally of the people they meet throughout the day which could potentially be incredibly useful. In my comment to her I agreed with her worries about privacy issues with such a device. On second thought however I wonder more how we might be able to leverage this data in a social context. Could we add this kind of data to Facebook profiles? Should we?

When we talk about social computing we often refer to components of what is now collectively called "Web 2.0". This term was coined by O'Reilly Media and refers to a revitalization and reinventing of web based technologies. To really understand social computing and Web 2.0 we must delve into what they are composed of and what lessons we can take away from them. In November 2005, Tim O'Reilly composed an article detailing his analysis of Web 2.0. This article compares many Web 1.0 services to their 2.0 counterparts and breaks down why they are different as well as what we can learn from them. For instance, O'Reilly compares:

"DoubleClick --> Google AdSense
Ofoto --> Flickr
Akamai --> BitTorrent
mp3.com --> Napster
Britannica Online --> Wikipedia"

Each of these changes in "leadership" was important in that they "leverage customer-self service and algorithmic data management to reach out to the entire web, to the edges and not just the center, to the long tail and not just the head" (O'Reilly, 2005). That right there is what defines social computing and Web 2.0. As O'Reilly says, "the service automatically gets better the more people use it". While computers have long been extremely useful they have only truly become a tool of the masses because of their ability to help people communicate and interact. The ideas inherent in social computing and Web 2.0, and their attendant tools, allow people to sift through enormous amounts of information and get at the little nuggets that they find important. I think that our goal now is to take this knowledge of what social computing is and use it to enhance that experience. We changed paradigms from application to service. Now we should change paradigms so that computing is about this Web 2.0 interaction of data and people instead of simple processing.

Article: What is Web 2.0