BookBlog

Data Portability Summit: Data Sharing, Privacy and Context

At the Data Portability Summit, there was some excellent discussion about Data Sharing, Privacy and Context.

In conventional wisdom, data sharing and privacy are seen as black and white opposites. Everything is locked down, private, non portable. Or everything is open, public, and free-flowing. But data sharing and privacy are not black and white. In real life, people share and present information based on social context. There are gradations of privacy and information sharing.

Here are some of the stories we came up with regarding gradations of privacy and sharing. The ideas came from the session, plus pre- and post-conversations with Joseph Smarr and Thomas Vander Wal

Truly Private information

There are times when it is right to share data in a way that preserves privacy. Family members use different photo services, and want to share photos with each other but not the rest of the world. A group working on mergers and acquisitions absolutely needs to keep information confidential. In these cases one give permission to family, friends, or business associates based on membership in a group.

Signal to noise, social context

There are many circumstances where information isn't truly private. But people choose to share with smaller groups. Someone doesn't want to bore all of their friends with information about knitting or rock climbing, when that information is relevant only to a few. Information about one's political or religious affiliation isn't a secret, but it may not be the information one chooses to share when meeting new people at a professional conference. In these cases, it would be useful to have the ability to create tags for the relevant groups, and share by tag. The tags can capture the nuances of subgroups: knitting hats vs. knitting sweaters, say.

Progressive disclosure

There are circumstances when people want to start by sharing with a smaller group, and invite more people. Or start by sharing a little bit of information about common interest, and later share more sensitive information.

Stream filter

The signal to noise and progressive disclosure patterns are about the person sharing information. Stream filtering is for the recipient. Sometimes one wants to "people watch" a diverse stream of information. And sometimes one wants to focus on the current work project, or upcoming social events. Stream filtering is used by individuals who want to apply a context to the information they receive.

Persona

People use identifiers -- dress or email address -- to represent more than one persona. The same person wears different clothes, with co-workers, at a customer meeting vs. a barbecue.

Personal vs. organizational control

In organizations, there are some things that an individual may want to control, and some things that admins want to control. A person might want to share soccer pictures with the soccer league. An admin may want to ensure that people aren't sharing the sports illustrated calendar widget.

Wiki notes on data sharing and privacy

Data Portability Summit: Everyone is famous

The session on data sharing and privacy was combined with Kevin Marks session on digital publics. We talked about people's experiences handling the increased visibility of internet life.

Managing reputation

People share about their experiences in order to get their side of the story out and create a public image. Among digital natives, "it's not a real breakup until you've listed it on facebook.

Handling fame

Before the internet, there were only a small number of people who had more followers than people can comfortably manage socially. Now many more people do. More widespread fame means that more people have the issues with stalkers and pestering fans.

Cautionary and instructive tales

At the session at the data sharing summit, the conversation turned to cautionary tales about social data sharing gone wrong.

Failed white lies

Someone begs out of a work-related social event by claiming the flu. His boss discovers a picture on flickr of the guy wearing a skirt and holding a drink. The picture is timestamped at the same data as the work party. His boss sends him a note suggesting that that may not be an effective way to recover from the flu. The lesson here is that some things that feel private are more public than we think.

Social network molting

It is socially awkward to unfriend people. Some people get around obsolete lists of friends by "forgetting" their password and needing to invite their current lists of friends with a new password. The lesson is that declared, public friends lists are in

The ex-girlfriend effect

The list of "people you should know" in social network recommendations often includes exes and enemies. These are people who are part of your social graph - but you are not connected to directly. In an organization, similar algorithms might locate internally competing projects. The algorithm doesn't know that some gaps in the social graph are deliberate.

Recent Changes Camp: Media and Science progress toward open content

At Recent Changes Camp, I heard signs from two very different directions -- fan fiction and biomedical research -- that open content business models are finally reaching the mainstream.

Laura Hale of the Fan History Wiki talked about how, since 2005, big media companies have stopped harrassing independent fan groups with takedown notices and other threats. Instead, they have set up their own commercial fan hosting sites, and use those as a way to promote their brand's content. Independent communities still exist. The big media companies can dismiss them as "unauthorized". The indies think of themselves as un-coopted. The main story is, the business model has changed, and the media companies think of communities as ways to make money, and not worth legal prosecution.

Ehud Lamm, a researcher in the philosophy of science, shared that the NIH, after years of debate, had mandated that researchers taking NIH money must make their papers available to the public. This decision was ratified into law in December of last year. Since the NIH is a major funder of biomedical research, this will have a transformative affect.

When the internet reached the mainstream, models based on open access to content and active user communities became more powerful than models based on limited access to physical artifacts. It's taken a decade for institutions and business models to adapt. But the times are changing, and the participants in wiki communities are seeing up close the results of the change.

Social software as a collaborative game

Ralph Koster gave an interesting talk at etech about lessons from game design for social software. There were several things that seemed right and useful.
* challenge. Overcoming challenges and learning are key to fun. Games are designed to provide a successive series of challenges. By contrast, the software design paradigm is focused on ease of use. This is right for an ecommerce site that a user uses once, but is wrong for applications that people use and learn over time.
* contextual interfaces. in a game world, the monster acts differently if you approach from the front or behind. Game designers create information architectures that present different behavior depending on context. By contrast, the software IA paradigm is about consistency.

The talk was also missing a few things, I think. His psychological model was individualistic. It was all about the individual player, and didn't talk about the social factors - decoration, storytelling, came creation. And his social model was purely competitive. "Of course,", he says, "people are playing the game to win. " But people play games with a variety of motivations, and social software includes ways to play individually and collaboratively, in addition to competitively.

This topic is really interesting and needs some more fleshing out.

Inevitability, social network merging, and Jane Jacobs

Inevitability is a rhetorical technique used when someone is trying to steamroll some highly debatable activity.

The first time I really noticed the technique being used was when doing public interest lobbying. Legislators pushing a bad bill would say that passage was inevitable. This tactic would put inexperienced activists into a tizzy. But it wasn't fact. They were just trying to get you to give up. opposing their deal.

Another way this is used is technical determinism. A given action is technically possible, and therefore it is inevitable that it will be used the way the speaker wants. This is bogus. Automobiles can easily do 70 mph. This makes it possible to construct many-lane, banked boulevards that allow cars to careen through neighborhoods. But it doesn't make it inevitable. The design of the road system is a social decision, not a purely technical decision. Localities can choose speedways or traffic calming.

The folk pursuing the social network graph experiment are claiming that reducing the inefficiency of digital social networks is inevitable, just as 60s traffic engineers claimed that reducing the inefficiency of local roads was inevitable. Some amount of social network friction is socially beneficial. Someday, digital networks will need to make this decision as policy choice. There will be the network equivalent of robots.txt, or some other aggregation calming technique. Brad Fitzpatrick is acting as the Bob Moses of social networks, someday we will need social network Jane Jacobs.

Facebook Developer Bubble

From Facebook Developer Garage a presentation two developers about going into business on their own. "You can write your application nights and weekends. Then, once it gets big, quit your day job. One of the most important things we learned was "time management 101 - the need to leave time to have conversations with people other than each other."

There was an early-boom land-rush vibe. A tall VC in oxford and chinos lured developers with a pitch that Facebook will go public for billions, and users on Facebook apps are worth $1 to $10 each. The race is on to attract millions of users with viral applications, and cash out before users get annoyed and quit the app. Virality is becoming social spam -- one of the apps has come up with a way of doing "reply-all" via SMS. Uh, thanks for bringing that into the world. There was envy and hatred for the "big incumbent players" (who started 3 months ago), who create new million-user apps by cross-promoting from their existing million-user apps.

The atmosphere was reminiscent of dot.com -- without some of the sleaziest aspects of fraudulent dot.com business models, but also with less of the idealism about opening information and improving the world. The median age at the event was about 21.5.

There is also a somewhat dizzying look into the future. The handful of developers at Facebook who took questions at the last session were talking about allowing apps to add elements to the API (yowza!), hosting applications (!), strengthening groups. Platform looks like this, not like Microsoft. The decisions that the crew of developers on the couch in front of the room make now will affect how software works years from now.

The euphoria and the leap from virality into social spam still isn't the interesting bit about Facebook to me. I've been walking around with various ideas for social applications for years, and Facebook provides a platform that makes it easy to bring those ideas into reality. There are all kinds of useful things that can be done, with an application based on people being able to communicate with their friends. Some of those things are socially useful, and some remunerative. The "get a bazillion users and sell out to a greater fool" model isn't so interesting.

I guess that makes me a chump from the perspective of profit maximization, but I feel the same way about quizzes and horoscopes as about celebrity weddings and movies where things go boom -- I don't mind, as long as it doesn't get in the way of doing things that might be less momentarily popular but are worth doing.

Connecting the Social Graph: the cure is worse than the disease

Who cares most about putting together a holistic picture of your friends and associates:

• you? You can eliminate the repetitive task of adding friends to networks
  
• marketers? They can infiltrate the social network and spam you through your networks of friends
  
• government analysts? They get to more easily trace people who know people who oppose government policies, attend anti-war demonstrations, protest factory farms.
  
• insurance companies? They can tailor your coverage based on whether your friends smoke or are sexually active

Brad Fitzpatrick and David Recordon are enthusiastic about reducing the inconvenience and friction of a disconnected social graph. But it seems to me that the cure is worse than the disease. Reducing friction introduced by different services in different social contexts is moderately convenient for individuals, and very handy for institutions that don't have peoples' best interests at heart.

Ross Mayfield pointed to privacy concerns here, and Danny O'Brien of EFF talked about them in a privacy session at BarCampBlock, the raw notes are here

An SNS caste system?

danah boyd has written an interesting and controversial essay documenting an observation she has made in recent months that there seems to be a socioeconomic/cultural division between young people using MySpace and Facebook. Privileged kids are gravitating to Facebook; lower-income and otherwise marginalized kids are staying on MySpace. There are a lot of very serious concerns about increasing inequality and decreased social mobility in the US. And if I was looking for domains to worry about it, Facebook and MySpace would be somewhere near dead last.

Like other folks who have commented on danah's essay, I would be watching out for change. The demographics of Facebook changed rapidly when they opened it up. The population is likely to change further with new applications. Maybe developers will add apps to Facebook that have more media and decoration features, so kids who want music and more pictures will be able to have those things on Facebook. Without more research, it's hard to say how much of the relative preferences have to do with overall visual style, vs. features, vs. preferential attachment. Not to mention, sns's are the subject of fashion, like physical clubs. What is considered "cool" will change in different social groups, too.

I don't see an increased concern about the creation of an SNS-based caste system. People group themselves, that is nothing new. A person will go where their friends and perceived peers are. We're talking about MySpace vs. Facebook, so digital divide access issues are factored out. Free social network services have much less built-in stratification than: selective colleges; the ability to pay for higher education or private education; racial profiling in shopping areas and on the street; clothing; transportation; neighborhood safety... any number of factors in the real world that differentiate strongly by income inequality, and are much higher, more persistent, more tightly closed barriers than social groupings on Myspace or Facebook.

One of the reasons that the article was controversial was that danah wrote about her anecdotal observations before going and getting quantitative data. I think it's fine to blog prepublished, unfinished and informal things. I would very much look forward to danah's cut of the analysis based on data. Once source might be the Pew data set which has socioeconomic information. The Pew study doesn't have the psychographic categories that danah is talking about, but they do have household income, education level of parents, race and ethnicity, and age, and might be a place to seek to validate some of the hypothesis.

Why shouldn't Toyota foster a Prius users community?

TechDirt has a snarky article about Toyota's effort to create a web community with Prius owners, referring to previous flops: a failed Walmart customer community, and an "anti-social software" application that let people sms others based on license plate number.

Toyota's implementation sounds flawed, but the idea has merit. Walmart shoppers have little in common other than they like cheap stuff and are willing to drive to get it. Prius owners, on the other hand, may have more in common, including maintaining a fairly new product, as well as interests in other green purchases and green policy.

The Toyota site, by news report, allows toyota owners to create profiles of themselves, and search profiles of other people. But what Toyota owners have in common isn't the desire to date or hire other Toyota owners (the motivation in MySpace and Linked In sites with this format. It's to share information about owning a Prius, and being generally interested in responsible household energy use.

The need would be better served by a traditional blog/wiki setup, where owners could tell stories about their Prius use, Prius products, and other experiences using and seeking green products. A profile might be a feature for the system -- and individuals could choose how much to disclosed about their personal identity -- but it wouldn't be the first thing that a user would do. There are other features more important than profile detail and profile search -- the ability to create local "prius club" events, for example. A built-in wiki would help Prius users build information about hybrid technology.

There are reason for Toyota not to host the site themselves, as TechDirect suggests, but to sponsor an independent site. Prius owners might be interested in third party modifications, such as plug-in conversions, that would void a Toyota warranty. Toyota might not be willing to foster plug-in mods under its own roof, although they would certainly benefit from learning from those early adopters. It would be useful to have a ratings service for mechanics and third-party products, and Toyota might not want to sponsor this directly, either. Prius owners might be interested in organizing to advocate green policy at a local or national level. Toyota might or might not be interested in being directly assocated with this.

So, the best solution might be for Toyota to be a sponsor in a third-party hosted community, rather than hosting itself. And, while profiles could be a useful part of the tool set, it wouldn't be the place to start. Prius owners are probably more into their shared interests than personally interested in each other.

Netvibes and portals

Like Yahoo and Google, but
* with widgets from anywhere to get out of the walled garden
* with lovely slideable ajax boxes that want to be a common design pattern

I hate the "everything-one-one-browser-screen". It is horribly distracting. Multiple Windows are good.

Back in the day, I used to use a 3M sticky note application that let you type things on stickynotes and organize them on bulletin boards. Only problem was it had godawful memory behavior and if you had too much information it would kill your computer's performance.

So, like NetVibes but with desktop widgets you can move around, bundle and snap together. Or a physical gizmo, the size of a deck of cards, with 5 or 7 or 12 display cards held together with a keyring. Plastic cards with a matt finish, slightly raised edges like coins, and a display. You can shuffle them so you see one or two or three at a time, or collapse them together. Each one is a widget; a clock or feedreader or a calendar. You configure the widget set from your desktop computer.

In front of your attention when you want it. Away and closed up when you don't.

VOIP your congresscritter

Does this exist yet? It's like the widgets that let you email your legislator, but instead it uses Skypeout, Gizmo Callout, or other service. You have reminder text about the bill number and the topic. You push the button to ring out to the congresscritter and talk to a staffer or leave a message. If the service was developed in a brandable fashion, it could be subsidized by EFF or the Sierra Club or the NRA or whoever.

For that matter, Congress needs VOIP service. A public office could setting up a voice presense handle and use it for inbound calls. Is anybody doing this yet?

Wiki investigative activism

At the Sunlight Foundation event, several tools were using wiki and wiki-like features. Eugene Kim and I led a session on effective wiki practices. One of the people at the session was at a traditional NGO, and wants her group to be more open, potentially using wiki for distributed information gathering around their core coverage. One of the barriers to wikification is the fear of "anyone editing" and the potential impact on quality. The NGO benefits from the specialized nature of its information. There are 10,000 regular users of the database, which feels like a lot to the NGO. But with a community of 10,000 readers, one might expect to start small, and build to a group of contributors in the hundreds to low thousands, which is quite a manageable size. Also, the "watchlist" feature of larger wikis enables subcommunities of specialists to watch and protect the areas they care about most, and develop social norms to maintain quality. Another cultural barriers to wikification was that the staff considers readers -- mostly journalists -- as an audience, and not yet as a community. The main cultural transition is to consider the readers a participatory community, and to evolve from a private, letter to the editor model to a peer contribution model.

The barriers faced by this NGO seem common to the sector. Traditional public interest organizations see their audiences as traditional journalists and legislative staff. Their content is not-very-accessible databases and pdf reports; the websites have little about the people involved in the organization; the membership, when their is one, is seen as a source of credit cards and petition signers, but not organizers and active participants. NGOs need some new cultural concepts to take advantage of the new tools.

There were some interesting wiki ventures at the event, and new in the world. CongressPedia was at the event. One of the strengths of that model is the ability to build a persistent and easily google-able reference source. Another Sunlight-funded tool, Open Congress, is aggregating bill and blog data to provide more visibility into the bill-making process. The foks building the tool plan to add a "wiki" feature for collaboration. Not sure how that would work, that seems like an invitation to edit wars and disinformation because of the frequently competitive nature of the legislative process.

I'm most excited about the potential of the Adopt a Committee project of the Daily Kos community. This is starting with a community of likeminded folk to grow and protect the information; enables distributed information gathering and group memory for a large number of people with mixed amounts of time; and seems like a great way to shed light on the committee process where bills take shape. Perhaps the role models for successful use of read/write web tools will be new groups that aren't burdened by traditional top-down-media orientation. If the dot com revolution is a model, startup organizations will pioneer new techniques, which will eventually get rolled into the ordinary way of doing business.

WeMeme and interestingness

Ross asks for a meme aggregator that shows him what his social network is looking at. In order for this to be really useful it needs a feature that Amazon had in its apparently now defunct purchasing ciricles. You could look up what people in Austin, or people at Cisco were reading. But the algorithm stripped out the most popular products, and showed the things that people in Austin had distinctively in common. So you wouldn't see the non-information that a lot of people watch the superbowl, but you would see that a lot of people saw a particular Jon Udell podcast.

Rashmi Sinha on Designing for Social Sharing

Rashmi Sinha hits the nail on the head. Being social is more about sharing than declaring. Sharing meals, sharing music, sharing gossip and news, sharing activities, all of these kinds of shared experiences are the stuff of social life, beyond the "hello" and the tribal handshake.

Rashmi's insightful presentation on "Design for Social Sharing" explores the design patterns of "second generation social networks that put objects at the center: tagging, video, news creation". Design patterns include passing on a cool video, tagging and rating. Social sharing apps combine personal social value. Tagging a link helps me remember it, and helps others find it too. Creating a playlist or group of pictures helps the person who makes the list, and other who come later.

One insightful pattern is sensing the presence of others. The is the magic of recent changes in a medium-sized wiki, where you can have a window into what your colleagues are thinking. The conventional wisdom is that "presense" means synchronous presense -- I can see that you are there and I can interrupt you if I want to. Asynchronous presense is differently good, you can see the flow of others' activities without interrupting.

Rashmi Social Sharing as a "second generation" of social networks, beyond the first-generation of explicit tools like Friendster. I think the generational terms are more about the hype cycle than what's been going on. While the Friendster fad flared, LiveJournal and Flickr fostered community and fun, and MySpace skyrocketed. Now, the design patterns and the nature of social apps are better understood - it's not just about saying hello.

Washington Post writes the "fear of wikipedia" article

Sunday's Washington Post is running a version of the mainstream media "concern troll" article about Wikipedia. Parents and teachers are dismayed that kids are plagiarizing homework assignments from Wikipedia. The answer comes from a new product from AOL, called "Study Buddy", which, er, contains authorized material for kids to plagiarize from? The answer, of course, that kids are supposed to be learning critical thinking skills. They're not supposed to be plagiarizing articles from the World Book Encyclopedia either. They are supposed to be learning to look up more than one source.

The article sniffs that "User-created content, such as entries found in Wikipedia, the open-to-most online encyclopedia, comes with varying degrees of trustworthiness." Of course, and human-created content comes with varying degrees of trustworthiness.

Amazon still doesn't get links

One of the persistently frustrating things about Amazon's reader reviews is that they don't have permalinks. Reviewers can't respond to other reviews, or even bring in references to reviews of other books. This prevents people from talking to each other. It prevents flamewars, and it prevents community.

Recently, Amazon has added several new "features" that borrow from the forms of social software. A "plog" looks like a weblog. It is a listing, in reverse chronological order, of posts about products you've seen or expressed an interest in. Unlike a "blog", which consists of posts the author has written, a "plog" is a marketing newsletter, with messages from authors and others who are trying to sell you things. Apparently the right to write is bestowed by Amazon upon slected authors or marketers. The recipient of this unnatural hybrid has very little over the content. You can make a comment, and comments even have permalinks. But there is no venue inside the Amazon sprawl to use these links to write back. The user doesn't have obvious ways to write or link. This is the opposite of user-generated content, it is content inflicted on the user.

In the same family of mutant social software is Amazon's wiki feature. The so-called wikis appear near the bottom of a well-shaft-long scroll of various product description and review features. If you log in with a username and credit card (!), you can edit a page about that product. I need to upgrade my credit card, apparently, in order to see if the wiki even has a linking feature. It is clear from perusing the top wikis that linking isn't part of the idiom. People who are writing collaborative commentary about, say, the XBOX, aren't building a rich , interlinked history and knowledgebase of the games market, trends, and technology, unlike the WIkipedia entry. Instead, the Amazon "wiki" is a short and shallow review that happens to have been written by more than one person. The Amazon XBOX wiki doesn't even have it's own link as far as I can tell, all you can do is get to the xbox page and scroll all the way down. This is the opposite of the design pattern of atomic entries, identified by links, and interconnected by links, that allows the Wikipedia entry to grow and deepen with links to Microsoft, components, games, market trends, and related information.

The problem with Amazon's reviews is that the absense of links inhibits the creation of community. The wikis are antithetical to the concept of building a rich knowledgebase using shared vocabulary as links. The plogs don't allow user-generated content. In all of these "features", Amazon's interface designers have borrowed the appearance social software but missed the meaning and the social dynamic that makes the whole of blogs and wikis to be greater than the sum of the parts.

Affective computing: the mood thermometer in the lecture hall

Here's the anecdote that was most telling about the wrong track taken by Affective Computing, the book by MIT's Rosalind Picard about the digitizing of emotion.

Picard writes about giving a lecture as part of an elearning program. She was troubled by the fact that she could not read the emotional responses of people in the audience, unlike a physical lecture hall, when you have visual signals of interest. Her suggestion was to wire the audience, and get a digital readout of the emotions.

In recent years, conferences and remote meetings have developed a different mechanism to read the response othe participants. A simple text chat enables people who are engaged to show and share emotions with smiles praise, questions, or heckling. In great presentations, the backchat is dead silent, as the audience is spell-bound.

Unlike a physical room, where even someone silent may reveal emotional signals through physical signs of boredeom, excitement, or anger, a backchat reveals nothing from someone who is silent. That's not quite true - someone who wants to telegraph excitement or displeasure in a 3D meeting room will also use backchat signals. In small groups, silence in a teleconference and backchat is also revealing. A group leader can ask someone who is unusually silent to say what they are thinking.

Another big difference is that in a lecture hall, or a remote presentation with backchat, participants have substantial control over the emotions they display. Picard's hypothetical mood thermometer might pick up on involuntary emotions, or emotions the participant might want to hide. A participant might be feeling angry at a family member, or lustful for a fellow member of the audience, or exhausted because of a small infant at home. Picard's hypothetical mood-reader would transmit those emotions to the lecturer.

Picard herself notes that all known emotion-detecting technology can be fooled by skilled humans. So emotional surveillance in the virtual classroom would lead to unnatural emotional repression. what's weirder, the mood thermometer is one-way -- the lecturer can see the mood of the audience, but the audience can't read each other. Backchat is very different from the mood thermometer. Like same-place emotions, backchat allows participants to feed on each other expicitly. This difference comes directly from Picard's belief that affective computing is "personal" -- her model doesn't include the social aspects of much of human emotion.

The backchat model can be extended to include emoticons, color feedback, and other signals to share emotions with other participants and the leader. These nonverbal signals can complement text chat; allowing people to do the thing they're good at, combining thoughts and emotions in communication.

Picard's theories rely on complex tools to automate emotions, rather than on simpler tools that allow people to share thoughts and emotions with each other. I think her theories are misguided, in an interesting and revealing way.

Campaigns without people

I've found three slick websites supporting greenhouse gas policies. This Climate Choices site from the Union of Concerned Scientists, focused on the California Legislative session is nicely designed, links to the bills, has videos and action alerts. But no obvious people, and no direct feedback.

This national site from Environmental Defense focuses on getting signatures in support for the languishing McCain Lieberman bill in Congress. It has a petition and videos, as well as gizmos you can put on your site, including banners, instant message icons, and PC wallpaper. But no humans, no way to provide direct feedback, no obvious way to meet fellow activists. I can see that over 100,000 people have signed the petition in California. Uh, yay, I guess.

This National Resources Defence Council has an action alert, a postcard to send, and a place to sign in and see the history of actions you've taken. . But no humans, no way to provide direct feedback, no obvious way to meet fellow activists. I can see that over 100,000 people have signed the petition in California.

These campaigns use the internet as if it were a fancier form of direct mail. A beautiful brochure, with some more widgets and animation. No opportunity to take advantage of the ability to meet to the people behind the scenes and to talk to each other. None of the messy, potentially unpredictable consequences of actual political organizing.

Jimmy Wales on Wikipedia Community

Jimmy Wales, the founder of Wikipedia, gave a talk at Stanford last week. The bits that struck me from the talk were about the human model of community that makes wikipedia work.

Jimmy described two models of online communities. In one model, people are ants. Information emerges from the unwitting contribution of the masses. Individuals are not powerful. In this model, reputation is a number. In the second model, people are a community, a few hundred active volunteers who know each other and interact based on kindness and trust. In this model, reputation is human.

Despite the high traffic - 5B page views per month at last count, Wikipedia is a tight community. 50% of edits are added by 615 people, and 73% of edits by 1746 people.

Wikipedia is a network of encyclopedias in different languages. The "tipping point" when an encyclopedia has enough critical mass to succeed is a community of 5-10 people, producing about 1000 articles. An encyclopedia gets reallyl useful at 500,000 to 100,0000 pages.

Jimmy gave two stories that showed the value of open access and community-defined process.

Say you're going to open a restaurant, and you're going to serve steak. There are steak knives that can be used to kill people. What do you do? You lock people in cages (he shows a very sad gorilla in a small cage on a concrete floor) By increasing barriers the barriers to doing bad things, you prevent people from doing good things.

Wikipedia has a very simple and flexible model for voting about whether a page is to be kept or deleted. It is just a wiki page, where participants note whether they believe the page should be deleted. Then an admin makes the decision. More weight is given to evidence that a topic is valid -- if there are 8 people who think it's hoax, and two people who prove with citations that the topic is valid, the page is kept. Programmers regularly ask whether they should write a voting widget, and Jimmy says no.

The wikipedia model is less common than the traditional software model, where access is restricted as much as possible, and permissions are restricted as much as possible. As wikis become more popular, some people gravitate toward the familiar pattern to manage content by keeping people out. It's inspiring to look to wikipedia's overwhelming success for lessons of the benefit of access, flexibility, and human community.

Squidoo and self-promotion

I understand Squidoo a bit better now that the fact that it's Seth Godin's company has sunk in. In the social software equation, the individual motivation is self-promotion, and Godin is hoping that others will be as entertaining in their quest for fame and fortune as he is.

Siderean: but what does it do?

I ran across while preparing for a panel on "tagging 2.0" at sxsw. They sell a faceted classification product used for information retrieval. They did Facetious, a cool del.icio.us mashup which applies facets to delicious tags.

In order to describe their hack as part of my talk on collaborative tagging, I wanted to know how it works. Does the software create the facets? Does the software suggest facets, which are then selected and edited by human administrators? Or do humans create the facets for Siderean to fill in?

I ordered collateral from the website -- needed to fill out a form for each piece of collateral, and talk to a salesman first. The collateral didn't answer my question -- it was all benefit/result marketing, with nothing about the software itself. So I emailed the salesguy. He said he could only answer the question if Socialtext was seriously interested in partnering with Siderean.

So I didn't include Siderean in the presentation. Does anyone have a clue how the software works?

SXSW 2006: RSS, it's not just for blogs anymore

At SXSW, I facilitated a session on "RSS: not just for blogs anymore", with Chris Frye of Feedburner, Scott Johnson of Ookles (formerly of Feedster), and Robyn Dupree of Bloglines.

The panelists gave many examples of the non-blog uses of RSS. The most prevalent is audio/video, which currently represents 20% of feedburner-measured content. "Structured blogging" applications like calendars and restaurant reviews are nifty and seeing experimentation, but not big takeup yet. RSS on mobile devices is used by a small number of users, but those users are very heavy users.

The most fun part of the session was the "fireside demo" format modeled after Wiki Wednesday. This being SXSW, there were plenty of people in the audience with interesting RSS applications. People lined up at the mic to give examples of RSS uses and hacks, including:

* Stuffopolis, a utility for keeping track of lending personal stuff
* Toolshed, a toolset for online music promotion
* 30Boxes (calendar) and Dodgeball plugins for Wordpress by Andy Skelton

Here's the delicious list

Speaking of invitations...

Upcoming has good attributes of both guy culture and Silicon Valley culture.

In girl culture, there are infinite nuances to invitations, non-invitations, and anti-invitations. Girls invite to pre-parties, afterparties, and other intimate moments that complement official get-togethers. When a traditional girl wants you to come, she looks you in the eye an invites *you*. When a traditional girl wants to know you're only sort of welcome, she tells you that there's an event going on. When a traditional girl wants to snub you, she tells the person standing next to you that they are invited to her event. Bonus girl points for making sure you hear, while whisking the other person away.

Guy invitations are delightfully simple, by contrast. A guy describes an event, and assumes that's an invitation. Hey, there's a tailgate party. Or hey, we're going riding.

Upcoming does the delightfully simple, guy-like version. The event is declared, with a time and a place. The attendee declaresly, openly and impersonally, if they are "watching" the event, or, "attending" the event. Here it is, come if you want to, no big deal if you don't.

Upcoming also has aspects of bay area culture. People have a facility at declaring parties. There's Mobile Monday, and Tag Tuesday, and Wiki Wednesday, and "bar camp". People pick themes or places and flock, often enough that you need some automation to keep track of it all. And the communities are big and dispersed enough that a public bulletin board is helpful; in smaller communities, it's easier and more obvious to let the subcommunity who needs to know more directly.

The one thing I wish it did was slurp events into a personal calendar. Today's task was copying and pasting sxsw sessions and parties.

Incentives to decentralization

Edgieo is a conceptually nifty decentralized listing service -- like craigslist, but it aggregates listings you put on your blog. The bit this misses is how and why an individual would put a listing on his/her own blog rather than craigs list or ebay, when those services already have a large audience and easy listing tools.

In order to get adopted, a standard needs to be:

• supported by an existing community
• supported by a widely used tool, or
• of benefit to the user him/herself, or
• a mix of these. As far as I can tell, edgeio doesn't have any of these, just a sprinkle of decentralization magic pixie dust.
  Posted by alevin at 08:20 PM | Comments (1) | TrackBack

Antisocial software - calendars

Did some conversational noodling about why group calendar software is so bad. The trouble is, it doesn't support the social processes that people use to schedule events.

First of all, the negotiation required to schedule a meeting. Today, this is done with the brute force method of giving the meeting-ee access to your whole calendar. All free time looks equivalently free. And, exposing one's whole calendar is "too much information" unless one manages nuances of calendar entry types. It would be nicer to be able to do an automated version of the "pick the best option" drill -- you publish some candidate times and the meeting-ee chooses from among them.

Second and worse, the affordances needed to quasi-schedule and unschedule a meeting. How do you set up a request for a casual lunch sometime that a few busy people happen to be free? How do you politely defer a meeting? Scheduling software is a clumsy, second-best replacement for the refined and obsolete skills of personal secretary or traditional wife.

Delicious acquisition anxiety

Phil Edwards writes about the discontent felt by del.icio.us users when the big bad Yahoo buys out a social software web service. It's tied into a critique of web 2.0 as an exploitative phenomenon.

I suspect this pattern arises from client-server architecture and server cost, regardless of malice. Successful client-server apps like google and del and flickr wind up costing someone a truckload of money. They need to do something to pay for the servers. There's hardware and backup and patches and air conditioning and so on.

Even if you factored out venture money and outsourced r&d and the other artifacts of high-tech commercial culture, you'd still need someone to pay for the servers. Thus the classic phenomenon of a successful, idealistic web app provider doing a begathon when the server goes down.

The governance issues posed by server ownership get particularly strange when it comes to online games and communities. Eventually it could lead to political governance, where costs are paid via taxes to a democratically chosen government.

Some applications (aggregated comments) might be done decentralized. e.g. a shared bookmarking service that aggregates the bookmarks in each of our browsers, and allows browsing and querying of the virtual db, or a decentralized aggregated comment tracker.

When these apps are conceived after there's an installed base of tools, it requires painful standards work to make this sort of thing happen, and then the installed base turn adoption process can take years. Data standards are political; the user base needs to have enough power and organization to create and demand the standard; this can take a long time. In many cases it's easier to throw up a server, which gets us into the economic bind.

Sxore skepticism

Pesonal digital identity is on the wrong side of the network effect. It's a chicken and egg problem. Individuals don't pick a digital identity solution, just like they don't pick an ignition system. It's plumbing that's provided by the tool vendor. There isn't end user demand for it.

Meanwhile, social software tool vendors haven't felt enough incentive to use a third party digital id system. The path of least resistance for a tool vendor is to implement its own internal single signin system. SixApart has TypeKey. Blogger has its own login system. Yahoo just merged Flickr's login system.

Sxore is an attempt to stimulate end-user demand by providing a a solution to a real problem -- comment spam. Sxore is a cross-application comment system with capchas, moderation, whitelist and blacklist features. The idea is that if a user signs up to comment on one blog, they'll be able to comment on other blogs. Sxore will work with WordPress and MovableType, so someone who likes it can use on their own blog.

This would have been brilliant 18 months ago, before the major tool vendors and projects added anti-spam features. Today, end-users will tempted to follow the path of least resistance, which is to use the features that come with their tool. Perhaps one opening is open source projects interested in some Sxore features. But there's no evidence on the Sxore site that Sxip is offering code.

The handiest -- and maybe the creepiest -- feature is the ability to follow comments for an individual user. Sxore creates an RSS feed for each user. Presumably you can follow comments made by that user across different blogs. So, if you think someone has good ideas about blog visualizations, you get to read what they also think about President Bush.

Asterisk VOIP application links new immigrants with 2nd generation translators

Over at Worldchanging, Emily Gertz reports on a social VOIP application that lets Chinese immigrants use their mobile phone to find an available English-speaking volunteer. The Guides help new immigrants with school registration and other practical puzzles of American life. The software uses ad hoc conference calling to patch together the caller, the volunteer, and a third party such as a local business or government agency.

FlickrFrame

Imagine a flat panel wall-mounted screen with a very slowly alternating selection from Flickr. The FlickrFrame could come with a remote control that lets you fast forward, pause, and navigate through friends, interesting photos, themes.


flight of the rainbow, Originally uploaded by linny.


Traditional art selection is a commitment. Unless you are wealthy enough to rotate a collection, you get a photo framed or buy a piece of art and live with it for years. The FlickrFrame would provide visual variety without Martha Stewart's budget.

There are images that are emotional or loud, that I'd want to look at sometime, but not everyday all year long. The FlickrFrame would allow the viewing of jagged and soothing images, without being locked into states of permanent angst or tranquility.

Looking at the Flickr API docs, someone has done a little bit of this with a hack that lets you display Flickr photos on a TIVO. "You can choose to display pictures searching by tags, groups, sets, users or just the most recent photos. This is configured by a GUI on the PC, or command line options for the adventurous." It doesn't have the very-slow-rotation feature, and it requires a Tivo.

The idea of transient art is implemented in 3d by the Canvas Gallery, a cafe and gallery in San Franciso that has art for sale or rent. I haven't seen the model in other places, not sure why. If there is furniture rental, surely there should be art rental.

MiniFlickr

Thinking about different settings to surf image, it would also be cool to have a Flickr browser for a handheld or screenphone. This would want a similar interface as the FlickrFrame's remote control, allowing navigation of tags, people, and other streams with a few keypresses and good lookup. Good for meditative time on trains, in line, and other time spent otherwise waiting.

this looks like one way to do it.

Thanks to Peter Kaminski for Flickr-inspired brainstorming.

Conversation Clouds

Here's what I mean by conversation clouds:

The cloud would be a picture of a conversation surrounding a person or a topic. The picture would show the relationships between the participants in a conversation. The densest areas would represent people who frequently cross-reference each other over time.

You can start with a participant (the url of a person's weblog), or a search term (a word or tag) Nodes are clustered based on closeness, measured by number of links and reverse links over a period of time (comments, too, if you can measure them).

If the picture starts with a link, then that link is at the center of the picture. The picture shows the links between the first node and the other nodes, and between other nodes that are connected to each other.

If the picture starts with a word, topic, or tag search, then the cloud contains a cluster of blogs that include the term or tag in the last time period. The picture shows lines between blogs that link to each other. Unlinked blogs are thrown out.

The cloud is built from a data set over a time period; the user should be able to scale the time (conversation over a week, a month, six months) The conversation cloud would need to provide ways to navigate through conversation space. If you click on a blog, perhaps you re-center around that blog's conversations. If you click on a tag or topic, you search based on that. You'd need to experiment with several ways of allowing browsing out from the first cloud.

This type of picture would not measure rank. Instead, it would illustrate the connections within subcommunities.

Cloud-browsing represents a pattern of blogsurfing. A reader might start with Mary Hodder's post on blog metrics, and then traverse to Dina Mehta, danah boyd, Stowe Boyd, Ross Mayfield.

The cloud would show in graphical form what a Technorati or Blogpulse search would -- who linked to the post. And it would also illustrate the repeated links and cross-links as people reply. If you zoomed out the time horizon, you'd see some relationships become more obviously dense, with repeated patterns of links and counterlinks.

I think this sort of presentation would get more of what we're looking for -- a picture of the relationships in a community that reveals participants, both loud and quiet. The ability to browse the conversation.

The results would be more interesting than a diagram of an email thread -- where participants already know who's talking to whom. It woudn't be particularly rankist, since webwide popularity isn't relevant to the picture. It would let you browse to related people, or related ideas that the same people are talking about.

The next step is to test this idea, maybe with a manually drawn picture, and then with a dataset and a toolkit like TouchGraph. This seems like a good experiment to me. It could be somebody's done this already. Or somebody's tried this and proved that it doesn't work. Please share if you know.

p.s. Zawodny talks about the need for content discovery. I don't know about you, but a lot of the content that I discover comes from browsing through a conversation and finding voices that I want to keep hearing.

Unlocking the value of topic blogs

Carson of Buzzmetrics talks about the financial value currently hidden in the midlist. A blogger in the so-called midlist might be highly influential in their subcommunity.

One thing which I think might be interesting to add to the discourse, would be something around topicality. i.e. "influential on what?" Because BuzzMetrics is typically answering questions of influence within a commercial setting, we are rarely looking for "top bloggers." We are looking for "top influencers amongst wireless application developers who happen to be positively inclined towards the Linux platform."

Exactly. Many bloggers are not general celebrities but are influential in some domain. Compared to traditional research, blog search is a very low cost way of finding those networks of influence.

Technorati's broken for the midlist

Lately, I've found that Technorati searches to find who's responded to my posts have become unbearably slow. Often it takes a few searches in a row for results to show up.

A Technorati employee explained that they've got a problem in the queue to fix that affects only midlist blogs. Currently, searches are more efficient for blogs with a great many links, and those with only one or two links. Searches are painfully slow in the middle.

This makes Technorati less useful for conversation discovery, particularly for the people who desire it most. Will Wheaton is a celebrity with high link rank from adoring fans. He probably isn't interested in talking back to all the fans who write about him, except in a selective "fan letter quote" manner. He's probably most concerned with the size of the audience, because that helps drive the audience and word of mounth for his books and television shows.

Midlist bloggers probably care most about conversation discovery -- they are blogging in order to participate in a conversation, and each cogent reference is valuable.

Whether the segment is valuable to Technorati depends on their business model. Niche blogs with subcommunity connections ought to have value -- more value than can be unlocked yet. The question is whether Technorati's customers are marketers and advertisers to whom they simply sell metrics -- in which case it doesnt' matter if the system performs poorly for the midlist. Or whether there's value with those users directly, by showing ads to them or providing paid services.

Conversation, not rank

In Mary Hodder's roundup of comments on the discussion of community metrics, I agree wholeheardedly with Dina Mehta. Dina says that the value of Technorati to her is conversation discovery.

For instance, I have no interest in what my ranking on Technorati is, but I do visit it daily to see who is linking to me and how they might have progressed a thought. Yet, I'm not so happy when these get transformed into lists, ratings and rankings. Are you merely well-known, or well-read?

Yes, exactly. I use Technorati to see who responded to what I wrote, to discover distributed comments. I also use Technorati to find out who's written about something I'm interested in at the moment. Then (if I have something to say), I'll comment on their blog or link to them. Technorati is for discovering and continuing conversation.

Link rank is a not-so-interesting byproduct.

Purple pro and con: the insight and the argument

Chris Dent writesin praise of purple numbers. These paragraph-level identifiers enable re-use of content. Chunks of good ideas are locked inside larger units, within documents and discussion threads.

The benefit of purple numbers is that they unlock insights, increasing the liquidity and flow of ideas. The drawback is that they break apart arguments. Insights may be captured in paragraphs. But arguments are conveyed across multiple paragraphs. You need more than one paragraph to provide context, to set up a contrast, or to draw a causal connection.

Sometimes, picking apart the individual points is what's needed to find the holes and strengthen understanding. Sometimes, picking at individual points is a sign of a flamewar -- people are searching for points of disagreement. Picking at points can increase the quality of thought, or reduce the quality of thought by reducing the incentive to build toward a larger theme.

In general, the wiki form is conducive to concensus, by bringing people literally on the same page. It will be interesting to see how wiki+purple affects the quality of thought and level of agreement.

Avoid rankism with clouds

In response to Mary Hodder's concern about "rankism"... I wonder whether rank is the wrong presentation, and clouds are right.

A cloud presentation would primarily show the communities that a blogger is in. It may show secondarily the influence strength within that community, but that should be secondary in the presentation.

A cloud presentation might enable navigation along topic axis. For my blog, you'd be able to traverse to social software and austin clouds.

Influence would be calculated within the cloud. So, Jon Lebkowsky would have separately-calculated influence level within Austin and environmental blog communities.

Perhaps the presentation would allow the browser to traverse communities. One could find "blogher", and traverse to the "sepia mutiny" south asia community.

Time would be an interesting factor. Perhaps one could view the cloud by week, month, or year. See how participation ebbs and flows over time. A longer time frame would be interesting -- I wonder whether other bloggers are "bursty" in their topics of interest. A long time frame would catch people who come and go.

In sum, a cloud presentation would avoid the worst of rankism, because it would focus on the community more than the individual, and allow a browser to traverse communities.

Mary Hodder on Blog Community Discovering

In a thoughtful essay, Mary Hodder explores what it will take for blog search to go beyond the "top 100 syndrome" to discover the interesting patterns of influence and community.

...this is about going beyond lists and links, to understand that the social relationships of expression between and across blogs is really about searching for a "metric for identity" or "metric for affiliation", "metric for community", or "metric for influence".

Mary is ambivalent about creating new forms of "rankism".

I have to say, I've resisted this for the past year, even though many people have asked me to work on something like this, because I hate rankism. I think scoring, even a more sophisticated version of it, akin to page-rank, is problematic and takes what is delightful about the blogosphere away, namely the fun of discovering a new writer or media creator on their terms, not others.

The algorithm would weight links in posts higher than blogroll links, and new blogroll links higher than old ones. It might include new terms like time read, comments, and topic score.

Hopefully, the tradeoff for more rank-ism is better discovery. This weekend, I spent some time exploring Sepia Mutiny - a group blog for South Asian writers - and its cousins, after meeting one of the authors at BlogHer. This form of indirect discovery is delightful. A tool that helps with such serendipity would hopefully be more like the joys of a used book search database, and less like "sororitization", the turning of social groups into popularity contests.

I wonder whether rank is the wrong presentation, and clouds are right. Clouds would primarily show the communities that a blogger is in -- and may show secondarily the influence strength of that community?

Blog search: Tell me something I don't know

I got an email about a new blog search engine called Blogniscient, so I clicked through to try it.

On the home page, it tells me that the top 10 political bloggers are:

#1 Michelle Malkin
#2 Captains Quarter Blog
#3 Eschaton
#4 Powerline
#5 Crooks and Liars
#6 Austin Bay
#7 Think Progress
#8 TPM Caf�
#9 The Anchoress
#10 Daily Kos

You can drill down and find the top liberal and conservative blogs. Two clicks later, I find that the top liberal bloggers are (the list goes to 20):

Liberal Politics
#1 Crooks and Liars
#2 The Left Coaster
#3 Eschaton
#4 Think Progress
#5 Daily Kos
#6 TPM Caf�
#7 Talking Points Memo
#8 Political Animal
#9 The Huffington Post
#10 America Blog

So please, Mr. Search Engine. Tell me something I don't know. I knew that Daily Kos and Atrios/Exchaton were very popular. I had no idea that Atrios was two places ahead of Kos, and... I don't care. It's not like baseball heading up to the playoffs, where there's going to be a single winner.

Where are the good centrist blogs, like The Moderate Voice and Ambivablog? They don't fit into the impoverished taxonomy, let alone sites like Booker Rising, a site focused on moderate-to-conservative African-Americans.

Here's the problem. The top 40 blog list is boring. It's stable. We know who they are. The job of a search engine is to tell the user something they don't already know.

Splitting up the top 100 into big themes is somewhat more interesting than the general-purpose Technorati 100. It's more meaningful to look at top political blogs, sci/tech blogs, entertainment blogs. But it's still stable, and doesn't convey much new information.

The top news stories is a bit more interesting, since that churns daily. That's an interesting zeitgeist check, and may be worth checking back.

The bulk of the site misses the glory of the web. With a vast amount of human knowledge there for the mining, please tell me something I didn't know already

The algorithm of network power

danah boyd just made a striking point at Blogher.

The link algorithms that drive "Top 100" lists at Technorati and other services are based on a broad and shallow pattern of linking. This is characteristic of male patterns of networking. By contrast, characteristically female patterns of networking are smaller and denser.

The "Top 100" pattern recaps the hit-based attention and financial economics of the mass media. It just doesn't measure the sub-communities that should be visible out of the "Long Tail."

Mary Hodder says that she is assembling an algorithm that will highlight the subnetworks and the long tails, using critera like comments and interlinks.

This is needed. Today's algorithms are missing communities of interest. And frankly, it's missing opportunities for power and money.

Technorati and the discovery of community

Thanks, Chris Anderson for the kind words. I'll have to repay them by explaining where he is wrong again.

Chris writes:

Technorati is a blog aggregator without a community

.

This is true when you use Technorati as a pure zeitgeist-check, to find what the blogosphere is dithering about today (Karl Rove and Windows Vista).

But it is false for one of the most interesting and valuable applications of Technorati -- conversation discovery. Bloggers use Technorati to find which other blogs are responding to their posts, so they can continue the conversation.

In an era of comment spam, Technorati has become a primary method of knitting together cross-blog conversation. Technorati helps make conversations and subcommunities visible. The "community" of Technorati is not a feature of the service itself. But Technorati is a key, and hidden component of the blogosphere's long-tail communities.

The "long tail" is social

Chris Anderson writes a refreshing rant about the misuse of the Long Tail. But he's partly wrong.

Anderson writes:

There are many distortions of the term, but the most common one is to use it as a newly-positive synonym for "fringe". Invoking the Long Tail is not a magic wand to explain away the apparent lack of demand for what you've got. The Long Tail is not a get-out-of-jail-free card for poor-selling product. Or weak sectors. Or bad ideas.

Anderson goes on to say that business models that focus only on fringe content are doomed to fail. Effective "long tail business models", like Amazon, combine popular content with niche content, and use the popular content to draw people in.

Anderson's right -- Indy-only online music services draw much less business than providers like Amazon that can use popular content as a draw. A customer might check out a Britney Spears album, and then use the recommendation engine to traverse to related and much less well-known music.

But Anderson is partly wrong. LiveJournal and Flickr disprove his theory. LiveJournal is an online journal community that has historically had a large population of young people. They congregate in social groups, often starting with people who are friends offline. The software gives users tools to control the level of privacy. A user can define which friends can see private content, what content to share with intimate friends, and what to share with the wider world. Similarly, Flicker is an online photo sharing community, where users can share photos with their friends and the world.

Cultural preferences are social. When people like strange music, unusual fashions, or minority religious practices, they most often do so with a subculture of like-minded folk.

This is hard to see in the mainstream commercial economy because of the history of technology. Until now, mainstream marketing has had two main kinds of choices.
* Mass media is used to reach wide audiences. Coarse-grained targeting is used to reach market segments -- viewers of the Cooking Channel, or readers of Parenting Magazine. The audiences for these niches is still quite large, many thousands of people.
* Direct marketing is used to reach individuals. Direct postal mail, telemarketing, and legitimate targeted email is used to reach individuals who are selected by personal history (e.g. bought the product before), or by membership in a targeted demographic group.

Until now, the smaller social networks in which people share culture have been largely private and noncommercial, with a small number of exceptions, like Tupperware parties and Amway.

What's worse, the content industry has done its best to make sure that social content-sharing is illegal. Rather than seeing opportunities in tools that let people share content, the industry sees all sharing as piracy, and tries to stamp it out.

So, the successful examples of social content-sharing are based on non-commercial content, like LiveJournal and Flickr. There are also grassroots networks of cross-linked music blogs where people review and recommend music. And there are networks of cross-linked knitting blogs where people review and recommend patterns. Classic long-tail stuff.

So, Chris Anderson is right that catalog retailers like Netflix and Amazon need to have hits, which help draw users to the niche. Their recommendation engines serve as an automated proxy for the natural social recommendations that people make every day.

But that's true only when you start with the content. When you start with groups of people, then opportunities for "long tail" are abundant, and don't depend quite so much on mainstream content.

Search vs. RSS?

Kevin Hale writes that RSS is becoming more important than search.

I think that gets something not-quite-right. It's true that when you find a superb resource, RSS lets you