Monday, November 29, 2004

A good question

Just where are the backchannel logs, anyway?

More on backchannels

Richard gives his thoughts about backchannels. He participated on the CSCW panel. Liz and I have commented there.

Monday, November 22, 2004

The ideal aggregator

Jack Vinson has been ruminating about the perfect aggregator. I like his blog threading points:
  • Thread of related blog posts. This is my current killer feature. I want to see the trail of a post as I am reading: Who else has commented or linked to the post I'm actively reading. Who has linked to that post? Go beyond trackback to find anything in my aggregator that references the post. Maybe this could show additional trackbacks to this article (outside of my current feeds).
  • Even cooler would be some gizmo that showed the conversations even outside the feeds I aggregate, so I could possibly see an originating article and all its "children." This is one of those holy grails that Lilia has discussed for seeing a blog conversation. It also would involve tight integration with a blog search service, such as Technorati.
As for me, I'm still partial to Bloglines. While it doesn't give you the thread, it tells you who has commented on the post you are reading. Offline reading isn't much of an issue for me. I'm normally connected. Jack might want to look at FeedDemon, which uses the Bloglines API to synch with Bloglines, which should give him the best of both worlds.

I'm still looking for a Bloglines-equivalent aggregrator that I can host internally, providing access to internal company feeds. Maybe the FeedDemon/Bloglines combo will provide part of the solution with Bloglines just ignoring the internal feeds. But I'd really like the whole solution.

Saturday, November 20, 2004

up a lazy river

so, i've been playing around with picasa and hello this morning

easy as pie

once predominantly text, communications over the internet are now rapidly giving way to much richer auditory and visual formats. broadcast radio and television (as we know them today) are doomed: the concentration of ownership and consequent baseness of its content, the passivity, the isochronality .. say goodbye

more on this soon

Friday, November 19, 2004

RSS =>> Data Visualization

10x10 makes it hard to ignore even the inherently dull

perhaps though, it could be put to some better uses,
instead of focusing on what the news oligarchs say:

various market transactions, collecting, studies in theory vs practice

what if instead of JPEGs, it keyed off bit torrents in a semantic web?

Thursday, November 18, 2004

A must read

Adam Bosworth posted his talk at the International Conference on Service Oriented Computing. You need to read the whole thing. Now. Really.

The first thing that came to mind when I read it was this essay by Dan Bricklin. Keep it simple and slopply and the power of dumb networks. Different layers of the OSI stack, but the message is the same.

Wednesday, November 17, 2004

Empire gone, encyclopedia following

I emailed the quoted bit below yesterday to some of my co-workers in response to one of them sending me the article in question. I should have blogged it. Tim Bray and David Weinberger did.

Hmmm. This great scholar mocks an entire work by extrapolating from one article. Now that's the sort of impressive research and better than "sophomoric" writing I expect from the Encyclopædia Britannica. So, unlike the esteemed Britannica editor, I will actually point you to more than one reference from which you may draw your conclusions:
Wikipedia tops Britannica on 5 of 6 articles, but Britannica is more consistent. Lots of good comments.
The more political an article may be, the iffier the content
A Wikipedia contributor writes about Wikipedia and accuracy (Okay, I just scanned this one, but I'm still ahead of the Britannica guy.)

Friday, November 12, 2004

Action on Atom

I feel the need to join the chorus praising Tim Bray's effort to kick the Atom spec to the next step in the IETF standards process. From Tim's blog post:
Ever since then, I’ve been convinced that standards organizations shouldn’t try to invent technology. (The W3C, which is jam-packed with super-smart people, has produced some horrible, damaging standards when they’ve tried to get too inventive.) The right role for a standards body is to wait till the implementors have deployed things and worked out the hard bits, then write down the consensus on what works and what doesn’t.

More CSCW bloggers

I had some good conversations with Jack Vinson at CSCW. He's posted links to conference bloggers here and here, so check them out for some different perspectives, details I missed, etc. I have some more to add about the conference, but that will have to wait a bit while I catch up on other things.

Wednesday, November 10, 2004

CSCW - Lawrence Lessig

I was really looking forward to this talk -- Hacking the Law to Rebuild a Free Culture. Those who had to leave early were not happy at all about missing it.

First, we had the obligatory wrap up and look ahead by the conference directors. There were 452 attendees from six continents. Not bad. Notes were new this year, but posters weren't. Being a newbie, I'd have guessed the opposite. The upcoming conference locations on similar topics look pretty enticing: Paris , Sanibel Island, and Banff.

Well, the presentation was worth the wait. Great use of visuals, obviously well practiced, and really seamless use of the remote. Here are some points that struck me: The writer and reader together make culture. Copyright is good and essential and just. Historically rational. Commercial remix is regulated. Uncommercial remix is free. Discussed what Hitchcock had to do to make Rear Window. Technology has now reduced the economic barriers to essentially nothing. Obtaining rights is enormously expensive, orders of magnitude more so than creating derivative works. He showed a hysterical remix of the Peanuts partying to "Hey Ya!" Wow! This one really made his point about digital creativity. This is powerful new multimedia speech that goes beyond text because it has more impact.

The same technology enables creative power and uncreative reuse. The "war" against uncreative destroys creative. Today, it is illegal. In future, it will be "impossible" unless you are a geek. The current version of copyright doesn't fit for technology today. It is too lawyer centric. There is no way to look up copyright holders; you are required to hire detectives to figure it out for you. Rejecting the law is stupid. He rejects the concept of rejecting the law. The law can change and has changed in the past to deal with technology advances fairly, as with the advent of player pianos. The mechanical reproduction fees that Congress mandated are the basis of the recording industry today. A modern equivalent would be a predetermined remix fee of some sort that would keep the lawyers out of the mix.

One other reasonable example of an attempt at change is called PDEA. After 50 years, you'd have to pay $1 to continue holding the copyright on your works. 85-90% of content would be free, because it wouldn't be worth even one dollar to the copyright holders to retain their rights. Congress, unfortunately, bought the argument that this would hurt poor copyright owners!

The political problem is IP McCarthyism. Lessig's answer is Creative Commons. No lawyers are needed because rights are built into the architecture. Some rights reserved, not all or none. Avoiding the extremes will stop the McCarthyism.

DRM can allow copyright owner to opt out of restrictions such as fair use. Technology can enable remix in new ways. Remix has always been going on. We don't just want fair use, but free use. All fair use means is that you can hire a lawyer when you are sued.

His most important audience is creators (creative artists in whatever media). Some musicians and authors are using the Creative Commons license today -- David Byrne, Lessig himself (Cory Doctorow is another example that Lessig didn't mention). Why? To drive more attention to your work and gain more influence and money, or as part of an explicit recognition that all creativity is a remix, building on the past. Ultimately, parents must get it. The Creative Commons license comes in three basic forms: machine readable, human understandable, and legally enforceable.

CSCW - Wednesday afternoon

First up was a CMU study of cross-cultural impacts in collaboration. The differences they looked at were High or low context, task or relationship focuses, and individualist or collectivist. They used Americans (including Canadians) and mainlaind Chinese for the study, since they were the most diametrically opposed. The participants communicated via F2F or IM. The study had several hypotheses. Here are a few of them:
  • more common ground yields more efficient conversations (true for Americans, but not for Chinese).
  • lean media will have a negative effect with participants from high context cultures (false - the Chinese were more efficient in IM)
  • same culture pairs will do better (false - they all did poorly - no significant difference)
The Chinese pairs spent time on relationship building, and discussed items in greater depth than the American or mixed teams.

Next, an MIT Media Lab team discussed their effort to influence group participation by showing them a shared display. Problems include groupthink, group polarization, and poor information sharing -- where folks talk about what everyone knows, not the unique information each one possesses. These can be addressed by considering minority viewpoints, alternative approaches, and increasing the breadth of discussion. Technology can potentially help: if you are aware of an imbalance, you can correct it.

All participants wore a mike with volume dectection only. The shared display graphed participation with over/under participation indicators. The primary hypothesis was that users would adapt their participation levels based on the feedback. What the found was that the over participators scaled back, but the underparticipators did not change their behavior. Indeed, they often refused to believe that they had underparticipated. You can find more info here.

An Australian team from CSIRO presented their study of remote online meetings between farmers and researchers dealing with drought and water salinity issues that affected the farmers' crops. They used NetMeeting over dialup connections and had many technical problems, but the farmers expressed high satisfaction anyway. The farmers changed their farm management, and the researchers got validation and feedback for their simulation model. My moral: any connection is better than none if it is delivering something really valuable.

CSCW - Wednesday morning - second session

Another case where there are two sessions I want to be in. The first is a panel discussion of digital backchannels in shared physical spaces. The other is a paper session on interruptions. Oh, well. I guess I can just read the papers.

This was a very good session, and the most active backchannel I've seen. No big surprise there. The IRC backchannel was projected on the screen at some points, which made part of the primary channel. As I was an active participant in the IRC chat, it was nearly impossible to take notes.

Here are some points the panel and audience made. I'm not even going to try to attribute them individually, even the ones that I'm pretty sure are my own. :-) I view it as a collective effort. So here goes, in one huge paragraph:

There is always a backchannel, be it whispering, note passing, eye rolling, fidgeting, or whatever. Physical presence doesn't mean mental attention. The audience could be counting ceiling tiles, doodling, reading the proceedings, or whatever. When laptops and wireless enter the equation, it is simply the means changing. Having an official backchannel provides the audience the opportunity to more productively deal with confusion, boredom, and the like by doing something that is more on topic than emailing or doodling. The perceived ephemerality of the channel leads to highly informal interaction, but actions such as projecting the backchannel on the big screen and logging the traffic changes the behavior of the channel, although the former has a much greater effect than the latter. People tend to forget they are being logged, much like email or IM. The claim was outside participants reduce focus, but we had a few outside participants during this session and it wasn't apparent at all to those in the physical space. While the focus was on physical spaces, there was a discussion on the backchannel (which included panel members) about the need for a backchannel on conference calls for everything from bio break notifications to building consensus.

CSCW - Wednesday morning - first session

This session is on distributed collaboration. Yea!!!

Pacific Northwest National Lab developed a collaboratory for the biological sciences. A collaboratory is a "center without walls." They held participatory design sessions with biologists and bioinformaticists. They were interested in supporting many different data contexts. They developed a web-based portal using Apache, Jakarta, Jetspeed, the University of Michigan's CHEF framework, scientific annotation middleware, workflow, etc. The portal has a shared data repository, user configurable metadata with temples, data translators and graphical viewers featuring automatic translation and linkage, an electronic laboratory notebook, and live access to external data sources with dynamic notification of new data. They are looking to move from project based to community based collaboration in their future development.

Sun was next with Meeting Central, software that they developed to overcome the problems that people had with meeting software they were having. Those problems were primarily audio, behavior, and technology. Their design approach was to eliminate the need for difficult to remember behaviors, make everything audible or visible, and provide a backchannel. They had a very impressive demo. They used Java on top of Collaboration Server and VoIP Server.

A team from Michigan studied partially distributed teams, which is very relevant for us. Would the colocated parties form an in-group? Is there an advantage to being colocated? They hypothesized that the colocated folks would form an in-group and outperform the isolates in their experiment. While the colocates did form an in-group, so did the isolates. Surprisingly, the top performers were the isolates, although that group also had the lowest performers. Location is a strong determiner of who you collaborate with. In groups may form in partially distributed teams. Isolates (telecommuters) also have some advantages. Future directions for research include travel, social interventions, and technological interventions such as IM. Also are developing a NetLogo model with software agents to do things that can't be done in the lab.

Tuesday afternoon - first session

I spent this session going back and forth between rooms, so I could catch the security paper for Guy.

First up -- the evolution of KM in a large organization. The area studied was technical engagement help for novel business services. There were two types of design employed: reflective design and local adaptation. The upshot: design for adaptability, because things change.

The security paper was very interesting. Nine interviews with nine ID experts recruited thru a local Snort user group. The group had many different job titles and came from organizations of various sizes. IDS requires a lot of different skills which are hard to obtain. Most got those skills by "playing around," typically on their home LAN. All described this as fun. Situated knowledge: the most important thing is knowing your own network. Transferring expertise to different networks environments is problematic. What is normal varies greatly from network to network.There was evidence of both organizational collaboration and community collaboration. Organizational collaboration typically happened within the network security team and coordinating efforts across the business. Community collaboration occurred via mailing lists and informal lines of communication. There is interdepence between analysts ant the larger community for new signatures and support. IDS is about people and expertise, not tools. This is a ripe domain for expertise development and collaborative tools.

I ended up staying in the room for Focus on the Agenda. You get a better meeting if everyone contributes to the agenda. Since so many meetings are boring and annoying, what goes on in a meeting that makes it boring or annoying? The authors developed the DEEPAND framework: Describe, Explain Evaluate, Predict, Alternatives, Negotiate, Decide. The study found that a lot of time is spent on description and others just don't care. The problem is that people want to be heard whether or not they have a willing audience. There are two possible errors: excluding important items and including unimportant items. Their prescription: Solicit agenda items, vote, follow voted agenda (must have plurality to make it). This process cut meeting duration in half while increasing the perceived quality.

I returned to other room during Return on Investment and Organizational Adoption by Jon Grudin of MS Research. In a nutsell, ROI is a bad measure because it solely looks at performance, failing to catch the real value. He had a larger model in which performance was just one aspect. Organizations that scream about needing ROI before making an investment decision do nothing about measuring it after a decision has been made.

Leveraging Social Networks for Information Sharing by Jeremy Goecks of Georgia Tech was next. His interest is in using social networks to mediate and direct sharing -- items are shared selectively via social networks.

CSCW - Tuesday morning - second session

First up was a study of increasing participation in a social community using Movielens as a source for two studies. The first one focused on generic encouragement. Unexpectedly, emphasing the individual's uniqueness and personal benefit produced a negative effect. One potential issue is that they were mentioning extrinsic benefits and mayber raters had intrinsic benefits for rating movies. The second study focused on goal setting. As expected, specific goals produced better results than "do your best" goals. Surprisingly, those with group goals actually outperformed. There was very wide variability in individual raters. Design implications: embed goals. An interesting question was raised: is undercontribution a bad thing? If everyone rates, are the ratings less valuable? The speaker disagreed.

Next up was a blogging paper from a group at Stanford, which discounted the notion of blogs as diaries, since blogs are intentionally public. They contain social actions - salutations, advice giving, invitations. Group blogs also confound the diary hypothesis. 20% of the bloggers in the study started in response to encouragement by others. Audience affects content, blogs are discussed in other media, and bloggers seek feedback from others: none of these are at all like a diary. The one thing that might be diary-like (although I don't think it was mentioned in that context) is thinking by writing. The interactivity of blogs happens in other channels more often than in blogs or comments. Future research: Blogging as reading and the relationship between readers and writers. While none of this was really new to me, it is nice to have a real study instead of speculation.

The third paper was on flash forums. Key features: diffuse authorship, large size, focused topic, short duration. Analogous to a flash mob. Flash forums are not conversational or deep. Forum reader software stinks. The group developed Discusson Reader software, which has a map that's tied to the discussion. Users valued the map and search features as much as essential features like the scrollbar. Users used the map, moving non-linearly. One goal that wasn't successfully addressed was facilitating the finding of novel content.

Tuesday, November 09, 2004

CSCW - Tuesday update

I have a bunch of notes from earlier sessions, but they need a lot of cleaning up before they are blog-ready. I plan to get those posted tomorrow morning before the sessions start, but there are no guarantees.

I used the IRC backchannels today. The experience was pretty hit or miss. From what I saw, all of the action tended to be on one channel, which typically was the channel corresponding to whichever room had the highest concentration of people who'd contribute rather than lurk. Whether it was a certain critical mass or specific instigators who drew others in, I don't know. While there were the expected humorous asides and some minor snarkiness, the conversation was generally pretty focused. The fact that all of it was being logged undoubtedly helped set the tone. The back channel really added to the experience of the last panel session, enriching the experience and eventually contributing points to the main discussion. We even had a panel member participate.

Before the panel session, Liz Lawley and I discussed the disconnect between what's happening between in this academic arena and what's emerging in technology. That became even more clear as the panel progressed. For instance, social network theory is much more robust than any sort of technology mainstream attempt to model it would have you believe. Why does this gap exist, and what, if anything, is anyone doing about it?

CSCW - Tuesday morning - first session

Today's first presenter is Laura Dabbish, with whom I talked last night. Spontaneous, informal communication has both benefits and downsides. Her focus is on interruptions. Typically, the interrupter benefits more than the recipient. 45% of the time the recipient didn't return to the task they were doing previously. She tried two approaches: workload displays and shared social identity. How much info can you give the interrupter without hurting his effectiveness? If the interrupter and recipient have joint goals, how does that effect their effectiveness. IOW, how do you maximize the effectiveness of both parties?

Ideally, the interrupter would use the workload info to time the interruption to minimize the impact of the interruption. People self-reported using the information displays, but the data indicate otherwise. The question rate was 22% less for team members, and they did use the info to better time their questions. The abstract display of workload worked best. Higher bandwidth was too distracting on the interrupter end, impacting the interrupter's performance.

A question was asked about how you summarize a more complex task. The answer (by Bob Kraut, her advisor) is that machine learning can give approximation of workload. Really??????? C'mon. What about when you are doing a task in your cube that doesn't involve the computer, like, perhaps, deep thinking? Even though sometimes interruptions are actually beneficial to the recipients, this research assumes that all interruptions are bad.

The second talk in this session isn't all that gripping. I'm spending more time on the backchannel discussing IM usage in call centers. The presenter is discussing an aware phone prototype which uses manual settings of availability since it is very difficult to automate it in a medical environment. Evaluation: the younger docs really liked it, because they feel bad about interrupting the older docs. IM might not be the best channel, and privacy might be a problem. Social awareness includes physical location as part of availablity.

The third speaker, Quentin Jones, immediately has me engaged. He is presenting Design Requirements of Location Aware Community Systems. He calls his approach P3 = People to people to geographical places. Uses Harry Potter's Marauder's Map as an example. How cool. Just another sign that any sufficiently advanced technology is indistinguishable from magic.

Team workspaces update

We are still having trouble getting our team workspaces pilot off the ground. It's very frustrating that no vendor really embraces our vision of simply enabling non-employee access to the workspaces they need. This shouldn't be a huge deal from a security perspective, but it is for reasons both technical and political. The political issues will almost assuredly be more difficult than the technical, but they can be overcome. I won't bore you all with details, but consider this: the alternative to giving these external folks access to a site is to email them the content. Hmmm. The difference in the risk profile is minimal.

So, we will descope a bit and break the project into phases, focusing on internal teams only at first, but making sure that whatever we do will work with the eventual external access plan. We must show some tangible progress. Meanwhile, we will be working in parallel to resolve the access issues and set the stage for evolving our extranet architecture.

Monday, November 08, 2004

CSCW - Monday evening

This was the poster and demo session. There were demos of some interesting software such as CoWord and CoPowerPoint, which enable multiple people to simultaneously edit a document, as opposed to the control switching that happens with NetMeeting. But for the most part, I enjoyed the interactive posters more. Various researchers, many of them doctoral students, summarized their research on posters, which they'd then discuss with whomever wandered by. My best conversations:
  • The user experience of instant messaging with Amy Voida of Georgia Tech
  • The effects of interruptions and how they differ based on organizational culture and individual differences with Laura Dabbish of Carnegie Mellon.
  • Ubiquitous mobile collaboration with Davor Cubranic of the University of British Columbia.
  • The use of instant messaging on large displays with Elaine Huang of Georgia Tech.
  • Information architecture with Sam Harvey of the University of Technology, Sydney.
  • Feedback acceptance by virtual teams with Matt Bietz of Michigan.

CSCW - Monday afternoon

I went to the panel discussion on whether CSCW needs organization theory. In a nutshell, CSCW is rather practice oriented and could use a cross-discipline theoretical boost. Borrow from useful looking theory in other disciplines, hypothesize about it's design implications, test, and voila! prescriptive theory for CSCW.

Somebody named Barry in a bright yellow t-shirt over a button down gave the panel some grief for being too focused on small group and co-located technology and not focused enough on the Internet. The SpeakerID application that was discussed in the morning session would have been very useful for getting complete and accurate names! Anyway, I'll have to see if I can track to Barry for a deeper dive into this topic.

This session wasn't as good as the others. Perhaps I should have gone to the evaluation methods session instead. Maybe if my battery hadn't died, I could have listened in or placed a query on the IRC backchannel to figure out if a switch was in order. I'll try to sit next to a plug tomorrow and see if the IRC channels add any value for me.

At CSCW 2004

I'm in Chicago this week, attending the ACM's Computer Supported Cooperative Work conference. So far, it's been pretty interesting. I seem to be a rare specimen here: a non-academic who doesn't work for an IT company's research group. Here's a quick review of what I've heard so far:

Mitch Kapor opened the conference with a presentation on The Open Source Society. There wasn't a ton of new stuff in here for me, but it was still interesting listening to Mitch talk. His presentation was all black text on a white background, except for pictures of Richard Stallman and Linus Torvalds. Amazingly, more folks recognized Stallman than Torvalds. Wow.

Content-wise, the key point for me is that Open Source is a meritocracy of coding, but doesn't really value the knowledge of the particular domain for which one may be coding. IOW, from a corporate perspective, one of the worst aspects of IT. Another gem: more folks on blogs increases noise, but more folks on Wikipedia improves discourse. This is due to the strength of the Wikipedia community and its social norms.

Next, I attended the "Collaboration Using Large Displays" session. This was a great session, even though this is an area that we really don't do anything with from a corporate perspective. A group from NASA presented the MERboard, which senior scientists working used to help determine the instructions to be sent to the Mars rovers in the daily batch. Key stat: 25% of the documents had multiple owners, and those documents averaged 5.7 owners. Because new versions were created on each save, over time, users started trusting the versioning rather than using "save as."

"Augmenting the Social Space of an Academic Conference" discussed the use of RFID tags at the Ubiquitous Computing conference to identify speakers and share user-selected personal information near the coffee pot. Very interesting stuff, especially around how they missed their goal of having the technology mesh with existing practices. You can find more info here.

Another group put a pair of large displays into a high school common area. Interestingly, the seeding activities they planted never got used, but they system was heavily used, and in manners the researches never anticipated. Again, the group had well developed social protocols, and applied them to the new technology.

Tuesday, November 02, 2004

Service, not device

Jon Udell has now chipped in to the Tim/Jeremy/Russell debate on do-it-all cell phones:
Pervasive connectivity and ubiquitous computing don't depend on the device you tote in your briefcase or clip to your belt. Or anyway, they shouldn't. The environments we visit, as well as the ones we live and work in, could provide for our communication and computational needs. Service portability would, in many cases, trump device portability.
Absolutely. Why should I have to worry about toting a device everywhere? One day the phone will really be a viable all-purpose computing device, and that could be way cool. But for now, I have a tiny cell phone that is voice-only but ultra portable. I only carry my Blackberry when I want email access, and there are a lot of times when I don't. Jon's vision is appealing:
The ultimate freedom, to me, would be the freedom not to have to tote my TiBook everywhere, because I'd know its generic equivalent would be waiting for me in all of these places. Identify to the device; flow your preferences and data and applications to it; use it for a while; wipe the slate clean.
I'd extend that vision to include cooperative computing. Because I will likely have some device, but a that device isn't going to have all of the capabilities of a stationary device, if for no other reason than form factor. For instance, the portable device should be able to take advantage of free WiFi VoIP or a larger screen. The fixed device should be able to take advantage of the cell phone's radio if the fixed device loses its standard connectivity.