Twitter Activity Streams: Surfacing Social Gestures Like Tumblr

Twitter is preparing to roll out a fairly significant rethinking of the user experience for the microstreaming service. They are planning to bring the social gestures that users make out in the open. These gestures are the actions of following people, favoriting tweets, retweets, or adding people to lists. Some of that gestural information has been available in Twitter to date, but most of it hasn’t been found in the stream along with the tweets themselves.

The change will come by changing the ‘@mentions’ tab into two:

MG Siegler, Twitter Comes Alive With Realtime Activity Streams

Specifically, the “@Mentions” tab on twitter.com is being replaced by two new tabs: “@USERNAME” and “Activity”. These two streams will add an additional layer to Twitter and to Tweets themselves, a layer showing the social activity around them.

The @USERNAME (obviously, USERNAME will be replaced by your Twitter name) stream will still show your @replies, but it will also show things like when someone follows you, when someone favorites one of your Tweets, when someone retweets one of your Tweets, or when someone adds you to a list.

The Activity stream will show you all of those things, but related to all of the people you follow on Twitter. In other words, you can see if a connection has retweeted a Tweet, or if they followed someone new, etc.

Siegler doesn’t say that the current Timeline tab — which shows the tweets from you and all that you follow — will remain unchanged, but that is my interpretation at present.

Surfacing social gestures in general — and making favoriting a social preoccupation instead of a not very robust bookmarking tool — is a great way to make Twitter a richer social experience. In fact, this shift feels like Twitter has taken a long hard look at Tumblr, and has decided to capitalize on that social networked blogging platform’s success, which is driven to a great extent by the richness of social gestures, which are presented in stream. Here’s a snippet of my Tumblr stream, showing gestures and a post:

I wrote a piece not too long ago, What Twitter Could Learn From Tumblr, which focused on the efforts that Tumblr has recently put into its support of tags, and curation of tagged topics. (For those still not familiar with Tumblr, you might read Comparing Tumblr To Wordpress.)

But it seems like the social gestures of Tumblr — which are natively presented in the Tumblr stream — will be the first innovation to jump from Tumblr to Twitter.

I wonder if Twitter will take the ‘notes’ idea from Tumblr, as well? In Tumblr, all the social gestures associated with a post can be displayed on that post’s page (depending on the template settings). So If I post something that garners a great deal of interest — getting liked and reposted a great deal — there is a long series of gestures shown on that page. In a sense, the post has it’s own associated stream: all the gestures that it caused.

On Twitter this would mean that the page associated with a tweet — the one reached by clicking on the tweet’s timestamp — might show all the favorites and retweets tied to the tweet. Will have to see if this will be done.

And oh, there is still all that work to be done on tags, which Twitter still doesn’t seem to be very interested in, yet.

What Twitter Could Learn From Tumblr

Twitter is on a fast growth path, as shown by recent data, but then the same data show Tumblr growing even faster.

What’s the story?

Twitter and Tumblr strongly diverge in their treatment of tags. Tumblr has implemented tags as first class metadata, explicitly supported by the Tumblr system, while Twitter continues to treat tags as microsyntax: text conventions invented by users, embedded in the messages. And I think this is a mistake for Twitter, and for the community.

You might counter my claim by saying, “Hey, wait! I use hashtags all the time, and so do others! Twitter supports their use!” But you’d be wrong. Twitter treats hashtags as text, just like all the other characters in a tweet. So if you write a tweet like this —

@JohnFontana: @DeepakChopra channeling his inner @stoweboyd #140conf

— and the ‘#140conf’ text represents that the tweet pertains to the 140 Character Conference (where I spoke yesterday, and so did Deepak Chopra). The important thing to realize is that Twitter does nothing special with the hashtag: it merely retrieves tweets that have that text in them during searches. Period.

[Update: 17 June 3:41pm EST — I have been informed by a commenter that hashtags are parsed by Twitter, and any hashtags embedded in the text of a tweet are accessible. But my real point stands: Twitter doesn’t develop that into a rich user experience. And the other ways that hashtags could be used in the API — like given a hashtag, show me all the tweets using it — would have to be implemented by an external program.]

Contrast that with the convention of the at sign (‘@stoweboyd’ ‘@deepakchopra’) that originally arose from users indicating who a tweet was intended for, but which Twitter adopted and built into the system at a deep level. Ditto for retweets, which was originally ‘RT’ text, and now is now implemented operationally, as a kind of message. Not so with tags.

Tumblr, on the other hand, like most blogging tools, has rich and deep support for tags. In the editor, the user can add tags to posts:

And knowledgeable users can take advantage of the tags, for example, typing in the URL to access posts with a certain tag, like ‘www.stoweboyd.com/tagged/curation’, which leads to Tumblr creating a tag page (or pages) with all the posts with the tag.

Perhaps even more interesting is the recent push by Tumblr to integrate tags with curation in the relatively new Explore capability. Basically, Tumblr has decided that a list of a few dozen very popular and broad categories — like ‘Tech’, ‘LOL’, ‘Comics’, and ‘Fashion’ — should be curated by a mix of algorithm and editorial oversight. Like a media company might do.

Below, you can see the Explore page for Tech, with the Featured tab selected. This is the view that is curated by a group of Editors, selected by Tumblr’s staff, and provided a different version of the Tumblr dashboard (something I have yet to see, either directly on in a write-up).

You can see that I am featured as a Top Contributor this morning, along with Smarter Planet and a bunch of other folks.

Note that I carefully called the Tech page on Explore a category, and not a tag, per se. I think that what Tumblr has done is create a mapping from a long long list of tags, like ‘apple’ ‘pc’ ‘iphone’ and ‘twitter’, and mapped that to the Tech category. That means I don’t have to explicitly tag my posts as ‘tech’ to be included.

And tags can be pulled from across the entire Twitter universe, using URLs like ‘www.tumblr.com/tagged/paris’ or ‘www.tumblr.com/tagged/liquid_media’. These are examples of tags that have not been promoted to curated categories, like ‘Tech’ or ‘Fashion’, but in the future, Tumblr could always expand the roster of curated categories.

So, Twitter could learn from this in the following ways:

1. Tumblr tags are metadata, and could be built-in more natively into the Twitter experience. For example, just as Twitter is now analyzing URLs and shortening them in the various clients, a similar analysis and indexing could go on for hashtags, either at the point of posting or at the point that a tweet from an external client enters the Twitter API.
2. The API could be extended so that tweets could be explicitly associated with tags, and these could be used to form queries, as well, like fetching the list of all the tweets I have made tagged ‘#140char’ or ‘#paris’.
3. Tag streams could be configured in a way similar to Tumblr Explore categories, and these could be either totally automatic — like the ‘All’ tab in Tumblr — or could be curated. Twitter could play the same role in curation as Tumblr is: picking editors and allowing the editors to identify top contributors.

Point 3 — where Twitter builds and manages its own liquid curation system, right in the Twitter application, as another set of Twitter owned-and-operated streams — is an enormous opportunity for Twitter, and one that would drive a stake in the heart of a dozen start-ups that are trying to make a business around topical influence on Twitter, like Klout, or media businesses, like Flipboard, Xydo, and News.me. But Twitter has not showed any reluctance in clobbering the ecosystem of quasi-parasitic companies living lamprey-like on the Twitter underbelly.

And, if coupled with a few other flourishes — like Flipboardish social journal display based on the URLs in the stream — Twitter could also destabilize the tablet media market pretty dramatically, and increase the company’s valuation dramatically.

Ultimately, everything important will appear in the streams first — like the stream of URLs in Twitter, and the stream of reblogs and likes in Tumblr — and those companies that own the streams will be in the best position to provide the complete liquid media user experience to users.

By exploiting tags and their role in curation, and quietly repositioning of the company as a media player, Tumblr is a giant step ahead of Twitter.

Yet Another Earthquake: The Pressing Need For Emergency Codes

During the Haitian Earthquakes first frightening days, the web world reacted in nearly instinctive fashion, clamoring for help, money, and technological approaches to mobilizing. The world’s larger response led to emergency crews trying to assist, doctors and nurses working in makeshift hospitals, and people the world over sending money and emergency supplies.

Even so, the devastation was unimaginable. And the Haitian people are in a terrible state, and likely to be facing years of rebuilding and healing. Hundreds of thousands are dead, and millions have been injured or grieving for loved ones.

We cannot predict disasters like earthquakes, tsunamis, and tornadoes, but we can do a much better job of coordinating our responses, and the natural communication that accompanies these disasters.

A number of approaches have been developed in recent months for trying to structure messages in Twitter to better relay emergency needs or requests. Many of these are based on the now commonly used hashtag.

My perspective is that hashtags are not really an appropriate way to encode complex information in Twitter or other streams of communication. In everyday use, hashtags are used to indicate the theme or topic of a tweet. They are not used to denote different parts of the tweet, they way that prepositions are used in natural language, for example. And perhaps worst of all, hashtags are based on natural language words or acronyms. So hashtags like ‘#have’ or ‘#need’ are understandable only to those who read English.languages. A disaster in Dagestan (a Russian republic on the Caspian Sea) might involve 15 or more language groups, with Russian being one of the smallest.

I read a piece in the New York Times recently that included a map of the most likely danger spots for catastrophic earthquakes. Here’s a map:

 

Sources: Koeri-Bogazici University, Istanbul (Istanbul analysis); Center for International Earth Science Information Network and Center for Hazards and Risk Research, Earth Institute at Columbia University. Via NY Times.

I studied linguistics in college, and I estimate that at least 50 languages are used in these areas by hundreds of millions of people. Spanish, Creole Friench, and various Native Amercian languages in South America are dwarfed by the Asian and Indonesian languages. Arabic, Mandarin, Hindi, Urdu, and other widely known major languages are involved, but most Amercians can’t even come up with the name of the national language of Indonesian — Bahasa Indonesia — which is spoken by around 200 million people there. It is the fourth most populous nation on earth, and an earthquake like the one that shook haiti to pieces could kill millions there.

My point is that we should develop a better way to transmit messages related to emergencies, and it should not be based on natural language keywords, which is what hashtag-based approaches do. A better microsyntax should be developed, using only common special characters, as we do with commas, apostrophes, and question marks in written language.

People have been using hashtags because 1/ they are fairly well understood, and 2/ they are supported in an obvious way by Twitter search and other search tools.

However, we should push ahead with a better approach and build tools to support it. That support might include adopt by Twitter and other tool vendors to support the use of these emergency codes in a direct fashion.

Therefore, I making the following proposition: A working group of interested parties should be formed, including representatives of various emergency and charitable groups involved in disaster recovery, to collectively develop a workable approach to messaging during emergencies.

I believe that such an approach has to have several characteristics:

• A workable approach cannot be based on keywords that are derived from natural language, like hashtags.
• The microsyntax should be distinctive, and unique: it should conflict as little as possible with other uses of punctuation, for example.
• Various scenarios of use should be developed based on the experiences of those involved in disaster response and recovery to make certain that the broadest collection of use cases are covered.
• Open source software to support this system should be designed and developed. This could include the development of an emergency codes server, which could be collecting emergency codes messages from Twitter and other services, in collaboration with Twitter and those services. This would potentially offload demand from the everyday services during emergencies, and allow for integration with other emergency-oriented applications. (This would also allow for blocking individuals or applications who might seek to exploit the service for spamming or outright disruption of messaging.)
• In such a model, victims, families, press, and responders could use everyday communication channels — cell phones and PCs with Twitter clients or via SMS — while those involved in mobilizing relief, coordinating materials and personnel, or tracking the status of people and places could be provided with specialized applications that could aggregate emergency encoded messages into a better big picture.

I have proposed the outlines of a microsyntax for emergency codes (see Disaster Microsyntax: Project EPIC, Tweak The Tweet, And Emergency Codes). It is very provisional, but has some of the characteristics needed. Here’s a sample tweet, based on a hypothetical hurricane called Bette that has struck New England:

!bette /usps, provincetown MA/ !@hassan haque: compound fracture of the lower right leg

Emergency messages (in this proposal) start with the unambiguous ‘!’ as the first character, and then the name of the disaster: naming must be undertaken by some international body. Then there is a location tag (or geoslash) indicating the US post office in Provincetown Massachusetts. Then there is some information about a specific individual, Hassan Haque, indicated by ‘!@’, and followed by a text field, indicated by ‘:’.

The specifics of this proposed microsyntax for emergency codes are less important than the fact that this — and other possible approaches — would work just as well for Bahasa Indonesia, Spanish, or Turkish.

And the microsyntax might not be seen by the originator of the message. If I were the person in the USPS building, typing this on my iPhone, I could be using a free application that formats emergency codes, or a feature of a Twitter client to support them.

In that case I might be provided with a simple form interface where I pick the sort of message — information about a person — and I merely key in the data: I type in location, the person’s name, and the text. That app would formulate as a standard emergency code, and out it would go into the Twitter stream. The official emergency codes server would gather that and other emergency codes, and other applications would slice and dice the information to display it, make planning easier, and to serve as a repository for others. Hassan’s family might access that information to find his status and location, for example.

The crisis in Chile is another wake up call. It doesn’t seem to be anything like what we have seen recently in Haiti, or what we can anticipate if we get a serious temblor in Instanbul or Jakarta. The NY Times piece I mentioned above stated that seismologists estimate that a nighttime earthquake of the sort predicted for Instanbul would lead to at least 30,000 deaths in the city, not including the surrounding countryside, and unknown levels of injured and displaced. A similar quake in Indonesia could mean 10-20 million deaths.

We should take steps now to build the system we need, instead of responding instinctively at the time of the next catastrophe.

—-

Update 30 March 2011 — See Bang: A Microsyntax For Emergency Messaging

Threading Blog And Twitter Streams Together

Stuart had some interesting observations about my recent posts on Ditching Technorati Tags:

Stowe, I ditched Technorati some years ago (feels like years). I have
found I really want two titles now. The title for my blog post and the
title for my Tweet. Which means right now I add a #hashtag or @name
into my blog title. As the title is always a tweet. Adding #hashtags
into the end of a post or even upfront would be interesting. Perhaps
more so if you used the SEO type functionality that is available.
However, too much thinking doesn’t help tags much. I use Delicious
still as a bookmark service and the tags potentially help me find some
article I want later… but really again it is not all that useful. I’m
also starting to consolidate categories for it keeps me more focused.
What’s probably relevant is we see a blog today as a super tweet. It
may be more thoughtful.. it may be a title or one liner supported by
more detailed notes. It’s just more valuable sharing it via Twitter
than most anywhere else.

I agree that I, too, have started to consider blog posts as a pair: the blog post and the Twitter post that I use to carry the link out into the stream. That tweet is like a needle and thread, sewing the two streams together — the shared and fast stream of Twitter, and the slower, publishing-style stream at my blog. 

Stuart has gone a step farther than me, considering the blog post title as being the same as the accompanying tweet. I still contrive a different tweet, usually one with some come on: a peek into the subject of the post or my feelings about the topic.

For example, the post on Saturday about attention economics and overload was called ‘The False Question of Attention Economics’. Crafting that title took some time: I think I changed it four or five times before settling, and I think its works on a number of levels.

One of the levels that it works on is that it is short enough to fit in a tweet with room left over for a short URL and a one-liner about the post. So they usually have this form:

[original title] [shortened URL] [quip]

I posted this tweet, which follows the form exactly.

The False Question Of Attention Economics http://sto.ly/8R3Mab We’ll solve overload culturally through a richer social web

And another example:

Its Only Blurry When You Expect It To Be Clear: Generational Multitasking http://sto.ly/4FRdht younger kids multitask more

Occasionally, the title is so snarky that there is no need for a quip, as in this case:

The Recession Is Over. I Feel So Much Better, Mr Summers. Thank You. http://sto.ly/7ctPDQ

Regarding tags: as I said in the post that Stuart responded to , I have decided to ditch Technorati tags. For those that haven’t fooled with them directly, Technorati tags are just an HTML href that point to corresponding pages at the Technorati website, like “http://technorati.com/tag/spaghetti” or “http://technorati.com/tag/bicycles”.

Hashtags, like ‘#futureshock’ or ‘#mcluhan’ are not HTML. They are strictly text, a form of microsyntax, like pronouns in English sentences. They intended to convey the idea that the tagged writing is about the topic that the tag represents. And that indication is available to human readers immediately, or for various search tools that might seek out tags in tweets or posts, and use them as a significantly more relevant handle for indexing than other words in the text.

So if I had thought to include it, the second example about could have read:

Its Only Blurry When You Expect It To Be Clear: Generational Multitasking http://sto.ly/4FRdht younger kids #multitask more

However, I forgot to include the tag.

Stuart might have written the original title with a hashtag includedm so the tweet would have looked like this

Its Only Blurry When You Expect It To Be Clear: Generational #Multitasking http://sto.ly/4FRdht younger kids #multitask more

So the tags are an additional way that the blog post and the tweet are linked, but that’s a weaker connection, since the connection is one of membership in the set of posts and tweets that are tagged #multitasking. On the other hand, there is a very tight connection between a tweet that has a URL referencing a particular post and the post itself. That’s the strongest tie.

Stuart has jumped ahead to thinking of blogging as, more or less, the place we create long format writing (or other creative works, like video, audio, etc.) in order to drop it into the faster stream. The slower stream of blogs is just a tributary to the faster, more social stream in Twitter (and other highly social streaming applications like Twitter, too).