Figure 2-1. Twitterfeed: automatically tweet a link from an RSS feed

People check their Twitter feeds for two basic reasons: to connect with people, and to exchange information. The immediacy of news flow on Twitter, facilitated by the short messages and varied means of access, makes it a great source of information for many members. Connecting other content to these channels, therefore, has value.

One of the simplest and most useful third-party applications is Twitterfeed. This RSS parser will accept feeds from any XML source—such as Facebook comments, blogs, or other information streams—and automatically convert the newest information into a tweet (see Figure 2-1). Created by Mario Menti (@mario), a solutions architect at Global Market Insite, Inc., this web tool allows some customization in terms of how the tweet is constructed, such as how frequently it should check for new content and what text to prefix to every link. Twitterfeed is the most popular third-party publication tool, accounting for about 10% of all tweets in the public stream and sending 80,000 feeds to Twitter.

Figure 2-2. SnapTweet: publish a link to Flickr photos

One of the important constraints that makes Twitter attractive is also a limitation. Unlike some other microblogging services, such as the now-defunct Pownce, Twitter cannot be used for sharing media files. Digital photography, though, is something many Twitter members do: they take pictures of important events in their lives or interesting things they encounter, and often upload them to photo-sharing communities such as Flickr. SnapTweet, created by Damon Clinkscales (@damon) of Austin, Texas, attempts to bridge the gap to Twitter (see Figure 2-2).

SnapTweet lets you automatically or manually post your latest Flickr photos as tweets. After signing up for a SnapTweet account, you can send a direct message to @snaptweet specifying the text you want to include with the link to your most recent Flickr upload. You need to provide your Twitter username and password and a link to your Flickr photostream. SnapTweet will use this information to publish a tweet with a URL to the new photo. You can also configure SnapTweet to scan your photos every 20 minutes and look for the presence of a special tag. Every photo it finds with that tag will automatically be posted to your Twitter stream, with the title and a link directly to the photo.

Figure 2-3. SecretTweet: tweet anonymously in a shared account

Twitter is a transparent medium. Not everyone likes sharing personal moments and various minutiae with the world, and there are those who won’t tweet simply because other people will know who wrote the messages. One of the benefits of the Internet, however, is the ability to be anonymous when desired.

Inspired by a friend, student graphic designer Kevin Smith (@mozunk) of Marshall University created SecretTweet, a communal Twitter account fed from a web interface that allows people to share secrets anonymously (see Figure 2-3). The site already had 750 posts at the height of the Fail Whale period in mid-2008, and another 4,000 were added over the next six months. The following has grown to 1,200 members, and SecretTweet is broadcast in multiple languages. Smith also created Voolia, a tool to collect multiple URLs into one tweetable link.

SecretTweet is not entirely automated, as Smith discovered the need for content moderation early on. When a person submits a secret using the SecretTweet web form, an email is sent to Smith with the text and a link to approve the tweet. Clicking on the link triggers a method request to the Twitter API to post the status update. All secrets are saved, stored in a database that feeds the “unabridged” version of the published information stream. Smith censors submissions that contain links to questionable material, racist language, or content not in the spirit of the channel. He also relies on community vetting and feedback, providing a means for commenting on and red-flagging submissions posted to the website.

Figure 2-4. Twittervision: tracking tweets around the globe

Among the earliest efforts to view the public timeline was Twittervision, created by Dave Troy (@davetroy). This visualization merges the location information in the authors’ Twitter profiles with a world map from the Google Maps API to display tweets based on the authors’ geography (see Figure 2-4). To be seen on Twittervision, you must have a public account with a custom profile picture and location text that can be parsed. Up to 100 of the most recent public updates during the last half-day are displayed.

With early API developers concentrating on the most easily accessible data, a number of projects similar to Twittervision cropped up in 2007. TwitterFaces, Twitter Planet, and Twitter Earth all did the same thing, albeit with different map platforms. This kind of visualization is interesting because it leverages an available but hidden bit of information—author location—and presents it in an engaging way. However, a lot of filtering is required to find relevant content when reading individual tweets from the public timeline. The real-time map visualization also requires constant attention in order to be of benefit to the user; there is no aggregation of data or any way to replay past tweets. What you see is what you get.

Figure 2-5. Twitter Matrix: tweets flow vertically, green text on black

In June 2007, Biz Stone (@biz) used the weekly Twitter newsletter to highlight a unique delivery of the Twitter stream, inspired by the movie The Matrix (http://espion.just-size.jp/files/js/matwitter/matwitter.html). Initially the treatment was only for the author’s own stream (in Japanese), but it now allows members to interact with the Twitter API directly and turn their own tweets into green text flowing vertically on a black screen (see Figure 2-5).

The Twitter Matrix application, created by Kyosuke Takayama (@takayama), has no practical value; text rendered in this way isn’t readable, nor does it provide any new insights about how you tweet or patterns in your content. Still, it’s wicked cool to look at for a while, and it is clearly a unique way to visualize the data flow.

Figure 2-6. Twalala: filter out tweets based on topic or author

Simplicity is an attraction when it comes to posting content, but not so much when it comes to managing a growing number of followers and followed authors. All it takes is one chatty friend attending an all-day conference to realize that your options for getting rid of the noise are limited. Unfollowing that person for a while might be the answer, but she’ll find out that you did so when you start following her again. Awkwardness can ensue, which is why a less formal way to filter other people’s tweets is a welcome feature of Twalala.

On the surface, Alex Taylor’s (@goldenmeanie) web interface is just another way to view your personal timeline (see Figure 2-6). Click on the settings link, though, and you’ll realize what Twalala contributes to the Twitter culture. You can filter tweets from your stream by designating keywords, phrases, or specific individuals to block at the presentation level, rather than dealing with a noise problem through Twitter’s servers. You can also create a whitelist of designated people or words that you want to exclude from filtering.

This could become a must-have feature for more established tools, like Twitterrific, as business and marketing folk start to push advertising through the Twitter streams: the same hashtag technique used to extract and group status updates by topic can be used against advertisers to effectively remove their content from Twitter. This means businesses such as the new Magpie network, which pays individuals for the right to occasionally post in their tweet streams, could find their ads hidden from the eyes they want to reach.

Figure 2-7. Track This: track packages over Twitter

For active Twitter members—over 200,000 different people are tweeting on any given day—communicating through status updates and direct messages is often more convenient than sending an email or loading up a website. Most of what we do on the computer involves tasks and information retrieval where connection with others is not the obvious goal; however, it makes sense that the Twitter channel might become useful for these activities, too.

One innovative example is Track This, a tool by Fragmented Tech (@pb30) that lets you tweet to get updates about where your packages are as they travel around the world (see Figure 2-7). After following the @trackthis account, you can send direct messages with a tracking number and a label that makes sense to you (as in, “d trackthis 123456789123 New Macbook”). Any time your package changes location along the route to its destination, a message will be sent to you with text describing its updated status.

In the first two months after its beta release in April 2008, Track This sent out over 10,000 updates for about 1,500 packages. Major delivery services such as FedEx, UPS, the USPS, and DHL are integrated with the service. In addition to Twitter, the tool accepts tracking numbers through the Web, email, IM, or SMS.

Figure 2-8. LiveTwitting: collect notes from a conference session

There was once a time when conference attendees used pen and paper to take notes on what a speaker said. Then came live blogging, which enabled a person with a laptop to not only capture those notes for posterity but also to share them with others not in attendance. Some people also use Twitter for this purpose, posting short comments throughout long talks. The problem with this is that it can become quite noisy for followers who love you for your occasional tweets about cute kittens and baking bread, not for a stream of snippets about the latest advances in your profession.

Michael Jensen (@mdjensen) and Danny Sullivan (@dannysullivan) developed LiveTwitting with this in mind (see Figure 2-8). All of the commands to tell the system what to do with your notes are sent through the direct message back channel—no more flooding your followers with tweets while you enjoy the conference. There is an added benefit of organizing all the tweets into a somewhat coherent document of notes from the live session: you wind up with the same kind of output as you would live blogging, without having to keep a laptop open.

The commands allow you to annotate your messages to @livetwitting, not just saving your text but also essentially meta-tagging your content. You can name your session, flag a new topic, identify the speaker, and mark when the question-and-answer segment starts. There are also pause and resume features to allow you to use direct messaging for other purposes before the session has ended. The aggregation of all of the content and formatting can be viewed in the Events section of the LiveTwitting website.

Figure 2-9. FoodFeed: share and track your eating habits

The American people are overweight (well, many of us), and good nutrition is a global issue. Although we techies may eat a lot of unhealthy stuff (mmm, Pop-Tarts) and lead sedentary lives, at least we have some self-awareness about the whole thing, and we occasionally try to get back in shape. For every fast food burger sold, it seems there is a matching tip on how to live a more healthy life. Some of the more successful strategies involve sharing your battle of the bulge with a community.

Vitor Lourenco—a young interaction designer from Brazil who did some coding for Twitter in 2008—built FoodFeed to encourage people to share details of their eating habits with their followers (see Figure 2-9). Vitor (@vlourenco) was responsible for the well-received Twitter website redesign in 2008 and was on the team that created the successful Election feed for the 2008 presidential campaigns. There are a few similar health services out there, but I chose to include FoodFeed here for a few key reasons.

First, it has a great username for the Twitter account that accepts the posts—@having—allowing you to both tweet in a more readable way and to answer Twitter’s prompt question (as in, “@having scrambled eggs”). The second great decision its developer made was to give users the option of posting what they’re eating as replies or as direct messages. This has implications for use, because you don’t have to worry about bothering your followers with a lot of food-related data, and also because the private communication option encourages people to feed FoodFeed with more and better data without fearing what others might think.

One thing FoodFeed doesn’t do is keep stats. In the site’s own words, it doesn’t actually do anything: “It’s not really useful. It’s just fun.” If you want to keep track of details such as your calorie intake, try Tweet What You Eat.

Figure 2-10. TwitDir: a directory of public members of Twitter

A year before Summize started searching tweets, a French marketing account manager began creating a directory of all Twitter members. Laurent Pantanacce (@loran) launched TwitDir as a way for users to search member profiles that match a given keyword (see Figure 2-10). The site can narrow the search to just parts of the profile, such as the name, username, location, or description. Initially, this was the only way to find other Twitter members in a certain geographic location.

More importantly, TwitDir became a de facto authority for Twitter statistics. By monitoring the public timeline, the site could identify new members posting to the stream and thus keep a running count of memberships. It wasn’t precise—private, dormant, and some early adopters’ accounts are not visible in the public stream—but it was accurate enough to suggest the arc of Twitter’s growth curve. Bruno Peeters, author of the Twitter Facts blog, has tracked the numbers TwitDir has published since its launch, using that obscure little number on the main search page to project milestones as Twitter grew.

TwitDir also keeps tabs on the leading users in the four main profile statistics: followed, updaters, favoriters, and followers. For the first several months, the Top 10 and Top 100 lists had some meaning. Today, however, even the top 1,000 users lists represent only about .03% of the total membership.

Figure 2-11. Green Tweets: streams tweets about environmental issues

In one sense, Twitter is like a multiverse. There are so many different uses and overlapping interests in what flows through its streams that Twitter has the potential to be all things to all people...but not at the same time. The things I find interesting (Chicago Bears victories, Web 2.0 startups, Joss Whedon) may not be the same subset of things that interest you. We are all using the same channel, but we each find ways to extract the bits that are most relevant to us. One way we do this is by assembling our follow networks, selecting other members whose tweets we want to read. But it is more difficult to filter based solely on content.

Green Tweets is one example of a topic filter: it tries to connect what all Twitter members are saying about the environment (see Figure 2-11). The custom information stream applies semantic language analytics to display tweets that mention “green” issues, from global warming to sustainability. The green stream was built by developers at Ecovian (@ecovian), a collection of city guides created by local communities of people who are passionate about sustainable lifestyle. Each guide contains user-generated reviews and resources to support local green living. Green Tweets is a great example of how Twitter can be used to help identify a community for another social networking platform, built around a domain topic.

Figure 2-12. TweetBeep: sends emails about keywords found in tweets

Two early features of Twitter that are missed by many were IM integration with Jabber and the ability to track tweets by keyword. Once upon a time, an IM window would pop up with a chat message from Twitter any time someone mentioned a term you had chosen to track, such as “informatics” or “Bloomington.” There are other ways to track this information now, but few of them come to me.

TweetBeep is billed as “Google Alerts for Twitter” (see Figure 2-12). It is more robust than Twitter’s tracker was, with controls for how frequently you are notified, what terms to exclude, and whether you want alerts constrained to a particular location. After you create a new alert, TweetBeep will email you a list of all of the tweets that should be on your radar. Separate domain alerts are available for mentions of your blog or other relevant URLs. Site creator Michael Jensen (@mdjensen) is also responsible for LiveTwitting (see Tools of Appropriation) and TweetAnswers, a community sourcing of questions and answers.

Figure 2-13. Tweet Scan: search public tweets and member profiles

Like TweetBeep, Tweet Scan offers a way to track use of keywords in the public timeline (see Figure 2-13). Enter a keyword to search both tweets and member profiles; the results are available in several formats, including email, RSS, and JSON. Tweet Scan allows you to track up to five phrases at a time and get the results delivered daily or weekly. The site has added support for Identi.ca and other Laconica-based microblogging channels, which gives the scan a reach that goes beyond Twitter. Tweet Scan maintains its own content database.

Tweet Scan has some other utilities as well. Although the user search hasn’t worked well for me, it does offer a tool that lets you download your historical archive of tweets, dating back to December 2007. Creator David Sterry (@weex) has also made it possible to include Tweet Scan as a list in your browser’s search box, letting you quickly search for terms without loading up the website first.

I have been using Tweet Scan since February 2008 to find local Twitter users. Even when Twitter’s own user search is working, it is only as good as the whimsy and diligence of each member choosing to set their location fields. Ironically, iPhone and BrightKite integration wreaked havoc on the trustworthiness of the member location by making the field more accurate. Now that location can change automatically as people travel, tracking who is tweeting as a resident in a college town means many false hits whenever the university hosts a conference or a sporting event. Tweet Scan’s daily emails allow me to look at what people say, pointing me to anyone who mentions my hometown so that I can examine the context of their timeline.

Figure 2-14. Favrd: daily lists of most favorite tweets

A lot of attention has been paid to three key stats that every member compiles: follower count, following count, and total number of updates. These numbers have been prominently displayed on every Twitter member profile page from the start and, as a result, are the most commonly utilized bit of information mined from the API. The black sheep of the profile stats is the favorites count. This is partly due to only a small percentage of people making use of Twitter’s bookmarking function, but it is also true that people don’t use it because comparatively fewer programmers develop for it.

Dean Allen (@textism), a 42-year-old Canadian living in the South of France, developed a website to start mining this underutilized Twitter feature. Favrd aggregates the favoriting activity of Twitter’s members and pulls up the most memorable tweets of the day (see Figure 2-14). Allen tips the hat to Ono Matope’s website, Favotter, which has cataloged almost 900,000 favorited tweets by concentrating on the 950 users that use that feature the most.

Favrd screens out the flotsam and jetsam that distract from the intended use of Twitter: real people commenting on life in their own voices. In addition to automatically crawling to find users who make use of the favorite button, the site also invites any users to add themselves to the voting list by verifying their Twitter credentials once. The site keeps track of the most favorited tweets from that community each day, letting you know who else has favorited the tweets you liked.

Figure 2-15. Twappi: estimates emotion of Twitter users

Twappi, launched in spring 2008, is a simple tool that scans daily posts from the public timeline and makes a summative assessment of the mood of Twitter’s public users (see Figure 2-15). Twitter’s advanced search tools allow for search by emoticon, so the “happy” percentage is calculated as a ratio of the use of :) characters out of the tweets that include emoticons.

Twappi isn’t just a reporting tool, though. Jacob Friis Saxberg (@webjay) of Denmark designed the site to include a single post by a Twitter member who is not in that happy percentage. The text of that user’s downer status update is followed by a link—with the text “cheer up”—that invites the viewer to take action. Clicking on the link brings up your Twitter home page, already pre-addressed as a reply to the unhappy tweeter. This is what makes this application so appealing, despite the relatively little it offers in terms of functionality.

Figure 2-16. Twitscoop: current buzz as a dynamic tag cloud

In 2007 and again in 2008, Twitter got some press for how quickly its members were able to report on and disseminate information about earthquakes and fires, well before the more traditional media could turn the events into stories. The attraction of writing a short tweet as opposed to a longer blog post or newspaper article gives microblogging channels an edge in the speed of information flow. Third-party application developers are trying to leverage the API to allow users to taste some of that breaking news goodness.

Lollicode’s Twitscoop is arguably the best at doing this. Inspired by the mathematics of financial markets, the Twitscoop algorithm crawls hundreds of tweets every minute, filtering out the common words and extracting the most often mentioned terms (see Figure 2-16). The results are displayed in a dynamic tag cloud and in a list of top phrases, giving viewers an idea of what people are talking about. One of the better features is the mouseover action on the phrase lists, which reveals some examples of specific tweets and a chart of references in the last 24 hours. Whereas Tweet Scan (see Tools for Search) is most useful for collecting references to specific keywords, Twitscoop is the most advanced tool for understanding what is happening in real time.

What makes Twitscoop so influential and useful, however, is not simply the fancy web work and computation. Twitscoop has its own Twitter account, which it uses to broadcast the more interesting changes in public chatter as they are detected. I’ve found that I am much more likely to go and explore the site when I’m first prompted that something interesting is going on there. Now, whenever I want to check whether some repetition in my personal information stream has become widespread, I visit Twitscoop to look for the terms in its cloud.

Figure 2-17. Twist: compare graphs of keyword references

Some of the early attempts at trend mining were fixated on basic counts of the number of times a specific word found its way into a tweet. The first tool to do this was Tweet Volume, which compared the total number of references to a few terms with a bar graph. The problem with this approach is that it loses meaning over time. Counting references to Barack Obama isn’t meaningful if he is mentioned constantly; the raw numbers of mentions will just continue to go up. The element of time is therefore an important dimension to consider when evaluating a trend.

Flaptor released a similar version of the keyword search that incorporates time into the equation. Twist lets you enter multiple terms and graph the number of daily references in tweets (see Figure 2-17). This allows you not only to see how use of each term has ebbed and flowed over the last week or month, but also to see how that activity compares with that of other terms. Twist highlights the most popular period (sometimes several days long) with a clickable box that reveals more detail. The graph automatically scales to fit the most popular term.

The big question during the early life of the company was, “How is Twitter going to make money?” That question is still open, although an answer is expected in 2009. Inevitably, some members have followed it with a second question: “What’s in it for me?”

Advertising on Twitter appears to be an emerging industry. Businesses have formed around selling advertising space on Twitter members’ profile pages (Twittad), sponsorship of moments during live-tweeted events (such as Glam Media’s Academy Awards coverage^[56]), and even selling individual tweets (Magpie). To bring those dollars from advertisers to users, however, investors need some means of evaluating which members will provide the best return on investment.

Figure 2-18. What’s Your Tweet Worth?: calculate your value

Lava Row, an emergent social media consultancy firm in Iowa, built a tool specifically for this purpose. What’s Your Tweet Worth? calculates how valuable you are to the Twitter community and spits out a price tag that you can put on your profile background for Twittad (@Twittad) (see Figure 2-18). Although the formula is hidden, the site returns the basic profile statistics, presumably as an indication that they are important in determining worth. If true, this application’s contribution is to turn multiple metrics into a single, easy-to-digest dollar figure.

At the close of 2007, Damon Cortesi (@dacort) became curious about his own Twitter use and built a little Perl script that would scrape and digest the tweet footprint for a given account. About a month after he released it, he used that script as the base code for a project to teach himself Ruby on Rails, and that in turn became TweetStats.

Figure 2-19. TweetStats: graph your tweet history and posting patterns

This web tool examines archival data for a given user and aggregates information about when that user tweets into graphs for different dimensions (see Figure 2-19). TweetStats also shows the people to whom you reply the most and the various ways you publish your status updates. TweetStats graphs are a nice way to understand one’s longitudinal usage of Twitter. Cortesi has also added a trends tracker, which tracks the top 10 keywords returned by Twitter’s trends API over time. The site includes a historic look at what has been important to twitterers via a tag cloud that includes such terms as iPhone, Obama, and Christmas.

Examining your Twitter history with TweetStats can provide some interesting insights into your daily routine. For example, for the six hours between midnight and dawn, my wife has almost no Twitter activity. Any time she spends conscious during that period is reserved for mild insomnia and needy kids. On the other hand, I have a spike in activity just as my wife is winding down for the night, followed by a low but steady murmur throughout the night. This reflects not only my general lack of sleep but also my irregular sleeping patterns (sometimes I stay up late, sometimes I get up early). It makes me wonder whether TweetStats could be used to reveal patterns in couples’ use of Internet technologies.

Figure 2-20. Follow Cost: how annoying is it to follow someone?

Whenever I am notified by email that I have a new follower, I do the same few things. First, I pull up the user’s profile page and look for a real name and bio information. The more complete and human the profile looks, the more likely I am to keep looking at the rest of the page. I then glance at the statistics for signs of a spammer: high following counts coupled with low follower counts are a red flag. Finally, I start looking at the quality of the member’s tweets, shying away from people who only post replies or automated tweets. Not everyone values the same things, but most people have some kind of system they use to determine whether or not to reciprocate.

Follow Cost facilitates this process: it attempts to assign an aggregated metric to your Twitter use to give other people another way of evaluating your value. This site only measures tweeting activity. The unit of measurement, however, is defined in a creative way: using über-user Robert Scoble (@scobleizer) as the baseline. According to creators Luke Francl (@lof) and Barry Hess (@bjhess), as of late September 2008 Scoble had tweeted 14,319 times in 675 days, for an average of 21.21 tweets per day. One milliscoble—their chosen unit of measure—is defined as 1/1,000th of the average daily Twitter status updates by Robert Scoble, or .02121 tweets per day.

The evaluation tool also has some hidden features. For instance, Figure 2-20 shows my rating shortly after the 2008 U.S. elections, when I spent most of the day and night tracking the Electoral College updates on Twitter. It was a personal record for number of tweets in a day, almost all of which were about voting and the election. Follow Cost picked up on that surge, raising my longer-range stats by 3–4 daily tweets and flagging me as a political junkie. It makes me wonder if I would get a background of footballs instead of flags if I spent a Sunday afternoon tweeting play-by-play for the Chicago Bears.

Follow Cost has at least one more surprise. When I checked, the follow cost for Robert Scoble showed a high level of activity for his last 100 tweets, with a rating of 76.46 posts per day, or over 3,600 milliscobles. The background for that stat was a mushroom cloud, and the title changed to “nuclear follow cost.”

Figure 2-21. Twitter Grader: turn your activity into a 100-point grade

One of the more popular statistical tools to pop up recently is Twitter Grader, an evaluation site that accumulates several stats and turns them into a single grade on a 100-point scale (see Figure 2-21). Built by the founders of HubSpot, which offers similar graders for websites and press releases, Twitter Grader collects data and adjusts your standing based on your data set. After about a month of service, it had already collected 300,000 profiles.

The grade it gives is the sum of several factors, including the number of followers you have, the power of those followers in the network, how often you update, and how thoroughly you filled out your profile. The single-figure statistic is nice because it is easy to wrap your head around. The metaphor of a test grade, though, unfortunately implies that any number lower than 60 is a failure. Twitter Grader, Twinfluence, and a few other member-ratings tools have drawn criticism for assigning values that some believe are best defined by individual users rather than algorithms.

It isn’t the grade that makes this site so interesting, though—HubSpot added a few other features to go with the rating and profile stats. The tag cloud isn’t that useful, probably because it doesn’t go deep enough into your tweet history to find much useful information to summarize in that way. The list of suggested people to follow, on the other hand, has had an impact on Twitter users. Before Twitter Grader, the people choosing to follow me were largely confined to three groups: people I knew, people from my home state, and spammers. After this app hit the scene, a fourth group became prominent: active users with moderately sized follow networks and some relevant tweet content. I suspect this spike in random follows is in fact not so random, but rather is attributable to people using Twitter Grader to find new members to follow.

Figure 2-22. Twitterank: using PageRank to calculate importance

Sometimes the best inspiration comes from an example of what not to do. One day in early November 2008, the Twitter community took a new third-party application to task, shortly after openly embracing it. At first Twitter was ablaze with signs of interest in the form of self-promoting tweets about a new user-ranking site, but only a few hours later came the backlash, as many speculated about whether the simple application was really a phishing scam. Panic ensued, especially after ZDNet’s Oliver Marks posted an article about user gullibility.^[57] Passwords were changed. Twitterank’s programmer responded. A parody was created.^[58] Twitter spent the latter part of the day responding to user complaints about something it wasn’t responsible for, and the Internet had a nice little meme about trust and authentication.

Twitterank was billed as a “PageRank for Twitter users,” claiming to ignore follower counts and instead use other indicators to gauge importance in the network. The initial page looked hastily assembled and was littered with a lot of tech slang; it warned users to take caution when giving away their Twitter passwords—and then asked them to enter their passwords. The ROI for granting your trust to this site was a page with a cryptic number and nothing more (see Figure 2-22).

To his credit, the creator of Twitterank—Ryo Chijiiwa (@ryochiji), a programmer at Google who made this tool as a side project—responded to the criticism and histrionics about his site being a phishing attack with some fast changes. Not the least of these were adding a method for calculating rank without requiring a password, albeit less accurately; adding context (he included a percentile to let you know where your number fit into the known Twitterverse); and a link to his identity. Those are the types of clues people look for when determining whether to trust a site.

The Twitterank of yesterday certainly won’t be the Twitterank of tomorrow. It is unknown whether Chijiiwa’s project will evolve into something useful or permanent, but it is clear that the debut could have gone better. The beleaguered developer tweeted at the end of his long day: “I have a new-found appreciation for people who do PR.”^[59]

Figure 2-23. Does Follow: confirms whether one user follows another

A web application doesn’t have to be complex to be useful. A bit of a coding meme started when Twitter API guru Alex Payne (@al3x) launched a tool that lets visitors type in an inaccessible web address and confirm whether or not the problem is on that site’s server (http://downforeveryoneorjustme.com). During Twitter’s darker tech days in the first half of 2008, Payne was frequently asked that question about his own company’s site. He built Down For Everyone for a friend, after realizing how widespread that line of simple questioning was.

Damon Clinkscales (@damon)—the same developer who brought us SnapTweet (see Tools for Publishing)—created a similarly simple web tool, Does Follow, to check whether one Twitter member is following another. It accepts two usernames and then calls the Confirm a Follow API method (see Confirm a Follow) to check with Twitter. The response page returns either a green “yup” or a red “nope.” There is also a button to easily switch the order of the names, to check for mutuality.

There’s nothing much to Does Follow, but it’s useful enough to be bookmarked as a resource (see Figure 2-23). Sometimes simple is good.

Figure 2-24. Qwitter: monitor who stops following you on Twitter

Managing your follow network is not an easy task on Twitter proper. Sure, you can add someone’s content to your personal information stream, and you can unfollow those folks who in the end don’t float your boat. When those things happen to you, however, you will only be notified of the follow, not the unfollow. Psychologically, it’s probably a good thing not to know when someone loses interest in your Twitter stream or what pushed them over the edge. Sometimes, though, you just need to know.

Enter Qwitter, created by Contrast. Signing up for Qwitter just requires you to enter your Twitter username and your email address (the destination for any future unhappy news when you lose someone from your follower flock; see Figure 2-24). The service monitors your follower list, checking routinely to see if anyone is missing and adding anyone new who might go missing in the future. If Qwitter detects that another user has rejected you, you get an email with a message like:

This tool fills a real gap in the features already offered by Twitter and the extended developer community. Be warned, though: the warm fuzzy feelings you get when someone follows you are twice as intense in the opposite direction when someone unfollows you!

Your Twitter profile page used to show you up to 100 avatars of the people you follow. When the site was redesigned in mid-2008, that number was reduced to 36. That means that in most cases, many of your flock are hidden from quick view.

If you want to figure out quickly who among that full list is also following you, the fastest way on Twitter is to page through your tweeps 20 at a time and look for the “Direct message” link, evidence that a user is following you back. It’s very tedious.

Happily, Dusty Reagan (@DustyReagan) has devised an easier way for you to make sense of your contact list. The Texas developer is responsible for Friend Or Follow, a tool that organizes a follow network of any size into three tabbed pages, separating everyone in your network into three categories. The first tab shows the profile images of all the people you follow who don’t return the favor. The second tab shows the people who crave your tweets despite you not liking theirs. The third tab shows the mutuals: those people with whom you freely exchange your little 140-character missives (see Figure 2-25). Members of this last group are the only people with whom you can have a two-way conversation through direct messaging.

Figure 2-25. Friend or Follow: a pictorial map of your follow network

One of the more personable and effective discovery tools for Twitter was launched in late 2008: Mr. Tweet. Developers Yu-Shan Fung (@ambivalence) and Ming Yeow Ng (@mingyeow) made an impact in this new development space through a quality product as well as a proactive, open relationship with people using the site (see Figure 2-26). Mr. Tweet quickly jumped onto the crowdsourced customer service site Get Satisfaction, and its developers have been very receptive to making changes to rapidly improve the young application.

Mr. Tweet requires only that you follow @MrTweet to join. This allows notifications to be sent via direct messages. Server limitations created an initial backlog of registration requests, but as of January 2009 the site sends out notices of updates to network analyses every two weeks. Mr. Tweet highlights changes in your follow network and gives you details about each Twitter member that follows you to help you make an informed decision about whether or not to follow them.

Figure 2-26. Mr. Tweet: gives you info to decide about following

Initially, the site provides three distinct views of your data. Mr. Tweet shows you a list of people that it recommends you follow, but the site also gives you a ranked ordering of people who are following you who might be worth of reciprocation. Finally, it provides a view to show you how others will see you in their lists. The value of these recommendations has improved dramatically since the initial launch, when the same crowd of über-users was always listed first. The algorithm has become quite a bit more nuanced. It is a very stylish site with a respectful appreciation of the Twitter community.

The idea for Omnee came to creator Mark Hawker (@markhawker) after he’d compiled a list of top people in health care worth following on Twitter. He quickly realized that maintaining such lists would prove an intractable problem for one person, so he turned to the wisdom of the crowds for help. The tag-based director attracted 100 users on the first day, and its use has grown steadily every since (see Figure 2-27).

The genius behind Omnee is the simple method of self-reported descriptions that each contributing member can add to the site. Rather than completing lengthy web forms or having a central authority assign statuses and classifications, Omnee asks everyone who wants to participate to send a direct message to @omnee (or a tweet with the #omnee hashtag), using a plus sign (+) before each new descriptive tag. Each submission through Twitter can contain as many keywords or phrases as Twitter length constraints allow. You can use the minus sign (–) to remove existing tags. An automated task runs every five minutes to process new information.

Figure 2-27. Omnee: an organic tag directory of twitterers

Hawker remained busy iterating the site after its launch. Most remarkably, he sought out the second-generation Twitter APIs containing data from other third-party tools to augment his site and give Omnee even more value. These data sources included TwitterCounter, Twitter Grader, Twinfluence, and Monitter.

From the island of Mauritius off the eastern coast of Africa, developer Asvin Baloo (@asvinb) released a tool for Twitter that makes the old manual way of searching for new people to follow a lot more efficient. Twitree starts with the 100 most recent additions to your friends timeline and lets you browse their own following lists for new names. There is nothing new there. However, because Twitree treats each individual as a directory in a recursive tree structure, you can dive down into the pathways one branch at a time (see Figure 2-28). I navigated through two dozen branches in the tree fairly quickly before the browser response slowed and took another minute to fill the horizontal screen with two dozen more.

Figure 2-28. Twitree: explore your part of the Twitter network

Twitree also adds some other functionality. You can update your Twitter status from its web form, and you can access a suite of actions—follow a twitterer, get profile information, send a direct message, reply, or retweet someone’s most recent message—by right-clicking on any of the user icons. Since Twitree knows who you are from the Twitter username you gave it initially, Twitree will also underline in green any Twitter members you encounter whom you are already following.

Much of this interaction only requires you to enter a Twitter username to get started. Authentication is needed so you can do things like post a message, but otherwise Twitree appears to be a well-conceived network exploration tool that takes advantage of what the Twitter API provides and accesses that information in a very efficient manner.