A Taxonomy of Social Media to help in understanding what Twitter is

Town square in front of Seoul City Hall on protest day

To understand online media -- which now date back close to 50 years -- it's worth looking at a taxonomy of sorts for all the parameters of a social or publishing platform. These can be viewed as axes though they are not quite independent. With the new controversy over Twitter, it's interesting to examine Twitter's place in the space of online social media, what may have made it a success, and what else might compete with it.

Many of these parameters also apply to traditional media, but some are new, or expanded by things going online. In particular, media that are social and not just publishing are new to the online world.

Online media began on timesharing services -- indeed, Plato, which was the first, dates back to 1960, though it become more sophisticated and social in the 1970s. The first multicomputer network social medium was Arpanet mailing lists, the first created by Dave Farber. The Late 70s saw the arrival of online services, dial-up BBSs, and USENET, which was the seat of community for the internet during the entire 1980s and some of the 1990s, though there were several other popular networks such as IRC, Fido, and more. The mid-1990s saw the arrival of the web browser and website and an explosion of media, as well as a few non-web-based systems like instant messengers.

Reader Friendly vs. Writer Friendly

Most media decide on a trade-off between being easy to create or easy to read. (I will say "read" though I really mean "consume" as later media will include audio and video and more.) Old school publications paid writers to work hard to make the stories good for readers, so they were aimed at being reader-friendly. As the online world arose, sites relied on users for content, and as such sometimes wanted to make it very easy to create, possibly at the expense of quality for the reader.

Most systems want as much of both as they can, and more recently software tools and AI have come forward to allow easy creation while tools help organize things to be easier to not just read, but manage. Nonetheless, there is usually a trade-off here. Writers need to do more work to make things easier for readers, and systems are not always sure who they need more.

Reading Order: Serial - Sampled - Sorted - Browsed - Searched

Most early media were serial -- messages appeared in a sequence and you tended to glance at them all, in order. Newspapers and magazines are serial. Mailing lists are serial, as was USENET, though it got very good at handling very large volumes. Libraries were browsed but the web brought browsed media to the forefront. In fact, many serial media became browsed, forcing users to have to figure out what they had seen and what was new, something quite reader-unfriendly. The RSS system allowed people to access content serially in the browsed environment of the web by having their computer constantly check for new messages.

Twitter was one of the first "sampled" media, which is to say a serial medium that goes by so fast you can only look at snatches of it from time to time, knowing you will miss things. The closest analog from before was channel surfing on a TV remote.

This led before long to the creation of sorted media, and the rise of the concept of "the algorithm," which is to say the computer attempting to figure out the importance of various messages in the stream, and presenting you only the most important, possibly by order of importance, but with some weaker chronology. This approach became common in all high-volume media, because people can't possibly read everything, but also don't like the randomness of sampling what's going by.

Facebook began as a browsed social network, but soon the feed become the core of it, and eventually that feed started to look a lot like Twitter but with two-way relationships and the potential for longer messages.

For some media, even though their underlying format may be serial or sorted or browsed, many readers enter from the viewpoint of search. That's sometimes internal search but often Google, which failed in all its social media plays but remains the leading path by which most of what is on the internet is read.

Search has also made global spread more immediate. People like that they can go to Twitter and search for up-to-the-minute posts on things, in many cases things not yet covered anywhere else.

Back in 1982, I proposed a social media system based on tags, rather than designated topic areas, with communities congealing around searches for these tags. It was too far ahead of its time and was largely forgotten. Much later, the same concept arose in Twitter in an ad-hoc way with the #hashtag approach. Eventual official support led to another generator of virality.

Brief - Long

Another function of Twitter that was special was its promise of brevity, which came about because of the 140 character limit on texts. With most media, a decision to "subscribe" involved a commitment, for the messages might be long and time-consuming. With Twitter, if you followed somebody they were generally forced to be brief, which made it easier to decide to follow them. Subscribing to blogs or mailing lists usually involved getting at least some long messages, though in general, the modern attention span encourages brevity to avoid scaring off the audience.

Naturally, many Twitter users don't want to be brief, even after the size was doubled to 280, and you see people posting long "threads" or using other tools to post long messages. On Facebook and others, long messages are possible but not particularly encouraged.

This axis has also had a big effect in video. TV and radio had long shows. YouTube brought in much shorter videos, but still fairly long. TikTok repeated Twitter's trick of demanding brevity to great success.


One approach to the brevity question is the use of headlines. Some systems demand headlines, or summaries, to offer multiple lengths. Some allow new headlines for sub-threads but I am not sure there are any who demand it. One interesting approach that will be possible in the future is to use AI summaries (which are getting quite usable now) to have the machine write headlines and summaries, or at least write drafts for minor tweaking. Then every post can be a tweet, summary, or full length if readers desire. Headlines do allow the consumption of large volumes of messages.

While headlines are ancient, and in newspapers are usually written by professional headline writers trying to help the reader get as much information in a small space, they have also become a reader-unfriendly feature in some systems to be "clickbait" where the headline is deliberately uninformative or deceptive.

Management: Open - Managed - Curated - Edited -- Pressured

Most early online media were largely unmanaged. Anybody could post and be read. This worked because groups were small and the people were trustworthy and known. This was the opposite of traditional media, which tended to have everything closely controlled by an editor or sole author.

Soon, different styles of control emerged. The most common was post-management, where moderators would delete or edit problem posts after they went out, and sometimes kick out offenders. There were also more fully edited platforms where the moderator had to approve everything in advance, which slowed things down but gave a very curated experience. There are also, of course, online media that look like old media, with an editor-in-chief responsible for all that is published and all who can contribute.

For most of the history of these media, management of the messages was not done due to external pressure, but for the users, or perhaps a vision of the operators of the system. Eventually, requirements came from governments to block certain content, and in the 2010s we saw the rise of public pressure to block content and ban users. While there have always been efforts to do this, it was more in the lines of "I don't want to read that sort of thing, so block it." The Trump election led to a strong rise in "I don't want other people to read this type of message" filtering.

There are many more policy decisions that differentiate the different platforms, though only a few are part of a taxonomy. This axis covers where policy decisions come from.

Viral - Direct

Some media reach only their subscribers or in-group. Others provide a function to make it easy for messages to spread, with features like "Share" and "Re-Tweet." While it's always possible to link to other posts, or to copy and paste them, one-button sharing speeds this up a lot. In addition, many systems use engagement (such as commenting or pressing "like" or "upvote" buttons) to trigger delivery of messages to a wider audience, even one beyond the intended audience. (In "Sorted" media, even the subscribers to something see only a fraction of it, but sorting based on engagement makes it much more likely something will be seen, which can then make it more likely it will be shared.)

Virality does depend on having a large and probably public audience. It allows messages to spread from sub-communities out to larger ones.

Most social media restrict conversations to being about single posts. A YouTube video does not go viral because of its comments -- though the more attention a video gets, the more YouTube promotes it to a wider audience. A Twitter thread, however, does spread because of replies. In the faded blogosphere, when people commented on a post in another location, rather than its original site, that took it viral.

Some platforms, such as Twitter, Reddit, and others have an idea of "trending topics" to increase virality and reduce FOMO.

Virality, of course, has also been a big negative in the online world, stirring mobs and falsehood while also bringing attention to hidden coolness. Some platforms have had great success with virality, while others try to avoid it.

Audience: Contacts - Community - Interest group - Curator - Public

Originally audiences for most media came from the public or large communities, and people subscribed to publications or tuned into broadcasts. Online media began with a mix of audiences that were either communities, or more commonly interest groups, but the interest groups were almost always communities too. This is to say that while a mailing list or newsgroup might be about a topic like Science Fiction, it soon also became a group of people who socialized because they were interested in the topic. Only later did we see the rise of communicating with more intimate groups, like a group of friends. That always existed but tools like buddy lists and Facebook made it central.

In some cases the audience is picked by the writer and reader together, which happens when "following" is symmetrical or a "friends" relationship. This also happens in invite-only mailing lists and was the original mode of Facebook. Much more commonly following is now one-way, and even on Facebook you can follow somebody without them following you. (Though there are still two-way relationships and people post messages only for friends.)

The size of a community plays a huge role. While Bob Metcalfe tried to suggest that the value of a network went up with the square of the number of things connected, with social networks it actually gets worse the larger it is, both because of poor signal/noise, but also because there can just be too much "signal." Ie. you can have a large group of top experts but you still don't have time to see 20 amazingly good essays on the same topic and debate them. In many cases people would rather discuss things with a smaller group that can be managed.

This has led to media where membership in the group is curated, not just messages. That concept dates back to some of the earliest mailing lists. Dave Farber, who created the very first mailing list maintains a list he calls "Interesting People" because he thinks the people on the list will be interesting to listen to.

While many social media are open to the general public, it is rare for one to get a truly broad audience across all populations and interests. Classic media had this to some degree, and newspapers like the New York Times got called "the Newspaper of Record," and even today stories in the NYT go out to a very wide audience far beyond the subscribership.

Systems that enable viral spreading can extend the audience of small groups to the public.

Only the largest systems like Facebook, Reddit, Twitter, YouTube, TikTok, and a few others have that global audience.

(See below for the question of the "global town square.")

Medium: Written - Audio/Photo/Video

An obvious and old distinction is the actual underlying medium. Early systems were all text just due to bandwidth limits, though text has many advantages which keep it popular today. Video creation can be very creator-unfriendly (it is a lot of work to make a good, popular YouTube video) but it can also be trivial if the audience will tolerate a video that's just gathered in real-time and unedited.

While many media allow visual forms in addition to text, platforms like Instagram put an emphasis on photos, and others have had emphasis on video, and podcasting on audio. Thanks to the cameraphone, photo posting can be both writer and reader friendly, though successful instagrammers put lots of effort into their photos.

Centralized - Decentralized

While each mailing list is centralized, the concept of mailing lists was highly decentralized. Most early systems like USENET, IRC, BBSs, and others were also decentralized. Web and app-based social media tend to be centralized, including Twitter, though the new Fediverse is decentralized.

There are upsides and downsides to both. Decentralized systems can be more robust and scale better, and are not subject to central control or easy censorship, but they also tend to innovate much more slowly, particularly if they require federation and collaboration. This was a key reason the web replaced USENET.

Funding: Philanthropy - Advertising - Micropayment - Subscription

A large fraction of social media were funded by philanthropy, or some mission. The big ones need revenue, for which advertising has become the dominant factor. While newsletters and publications often run on subscription, getting payment for social networks has been a challenge, because social networks depend on contributions by members, and you want to avoid charging money to those who make the system work. The push to advertising has altered the nature of most systems, due to the conflict of interest that generates the phrase "If you're not the customer, you're the product."

Latency: Immediate - Short-term - Long-term - Historical

Systems will have inherent latencies (how long it takes to publish a message) but today that's more rare. In the modern era, the key role of latency is how long people are expected to take to read and respond to messages. Some systems are immediate -- if you don't see it right away you are unlikely to ever see it. Some you may not get to for days. In the traditional publishing world, we have fast news, daily and weekly cycles, and media that are meant more as a recording of history.

Mobile or Desktop

Today, there's no real medium not found on mobile. But some systems -- including Twitter -- got their edge by being on mobile earlier. Twitter, in fact, began using SMS before the smartphone became a popular thing.

Motive: Provoke - Pursaude - Help - Request Help - Entertain - Earn - Socialize

Almost all platforms will have several reasons why people contribute, and read -- though the motive of contributors is the most important. A small number pay contributors (like YouTube for some of them) but most are there for their own purposes or gratification. We like to be important and have an audience, and we like to be social.

For some Twitter is social, but there are many who don't interact, they just publish. Almost all on Facebook want to socialize. YouTube has famously horrible comments and is not so much for social discussion. While many are there for their own benefit -- convincing or provoking others -- there are also those who like to give help or just to entertain or to enjoy the attention.

Identification: Anonymous- Pseudonyms - Real Names - Reputation

Systems will usually have a policy on how posters are identified. A few systems, like 4chan and its ilk focus on anonymity. Pseudonyms are also common, sometimes linked to a hidden email, sometimes just anonymous with a persistent name. Facebook famously insists on true names, though some get around that. Many platforms have reputation systems (such as reddit and slashdot karma) to boost pseudonyms. A few systems have authentication of real names, like the blue checkmarks on Twitter and Facebook, for well known people from the real world.

Specialty Features

Some features and policies are special to only one or two platforms and don't form the basis of a classification scheme. Emphemerality was a key feature for Snapchat, for example. Many message boards have very specific policies as well.

So, about Twitter

Using all these elements, it is possible to examine the value of Twitter.

Twitter's biggest distinguishing factors are:

  • The promise of brevity, which makes it very writer-friendly, and also reader-friendly
  • The global audience and the ability to reach more than those who follow you
  • The ability to make things go viral
  • Being the first of the sampled media, and eventually a sorted medium
  • Its early push into mobile, as perhaps the only social app for the phone in 2007

Global town square

Musk says he purchased Twitter because it was the "global town square." I prefer the term "seat of community" to describe something that is the leading place where people interact about other than specific topics.

In the internet's early days, there was little doubt that there was a seat of community, an online place where "it happened" and that started in mailing lists and was USENET from 1980 to 1994. After that, a wide variety of websites vied to be that place. In a way, the internet's early growth came because there were a million websites from a million owners, each innovating to attract readers, and there was no central place. Yahoo became the most widely used site, but not for socialization.

Message board websites started to take this role, certainly for given communities, with Slashdot being the leader for the tech community. The early "social networking" sites were not community sites. As Facebook came to dominate, it never had a central place where you talked to everybody -- indeed its virtue was you interacted only with those you knew.

Twitter arose to fill that role with the attributes above, especially brevity and being on phones. Even so, it has various competitors for community in places like Reddit, which for a time called itself "the front page of the internet."

There was also the rise of what people called the "blogosphere." This was a large collection of individual sites, each posting just the content of the blogger, and there were aggregating platforms like Blogger and WordPress which had large numbers of independent bloggers on them. RSS readers allowed people to subscribe to these blogs, and it was common for bloggers to reference other blogs to create a discussion. For a time, bloggers even automatically put in trackback links when they were referenced, though spam stopped that.

It's worth noting that Evan Williams founded Blogger and was one of the founders of Twitter. He later built Medium, another blogging site.

Another contender was LiveJournal, which had both blogging and social networks. It faded and particularly after becoming Russian-owned, it could not be the seat of community for the internet.

Twitter's claim to global town square comes because it is the most prominent place where anybody might read what you write, and you might read messages from anybody, and you might even have conversations, about any topic. This can happen on Facebook if your "friends" share the posts of others, but it's not as strong.

YouTube (and now TikTok and Facebook Reels) have some element of this. People have subscribers to channels but really hope their videos will go viral to a big audience. But YouTube, even with comments, is not a two-way conversation.

The previous "global town squares" of E-mail and USENET were not run by one company, or person, though. In the past, to be "the place" required a fairly open platform, with minimal censorship and nothing particularly divisive in its philosophy of operation. As soon as a platform becomes aimed at one group or disliked by another group, it won't get to be a global platform any more, which is a difficult path to follow.

It is also worth noting that when it comes to what people read and what links them to other content, this has also changed. By far the largest source of traffic for articles online today is Google. For news sites, this includes the Google News product but it's true all over. If one has articles on a web page, the traffic will come from Google, and then at a much lower level from Facebook, Twitter, Reddit, LinkedIn, and some other search sites. Even though Google is in no way a social or discussion site, it still drives a great deal of what is read on the internet about current events. Social media choose from the above parameters and they usually represent trade-offs. Commercial networks seek a large audience, but larger audiences just mean more noise to the point that readers can't handle the volume, or the volume has to be severely filtered. People want to feel engaged but can't handle too much engagement. They want just enough engagement to reduce "FOMO" but not a great deal more.

Networks that have had too many people in the conversation have tended to get overloaded. One of Twitter's successes has been to do this without as much overload through the promise of brevity. Nobody can handle reading 500 long think pieces, but you can scan over 500 Tweets.

Having been so involved in USENET, I find it interesting to reflect on how its attributes continued and didn't continue into the modern world. USENET had to deal with a much smaller population on the internet, allowing it to be the seat of community for so many topics. Part of this came from its simple text interface, the requirement for a subject line, and threading. In addition, it was strongly serial -- once you saw a subject line and didn't read it, all those messages were marked as read.

As a result, you could go to USENET and see all your topic areas quickly. You then entered one and might see a screen full of Subject lines for the threads which were underway. If you wanted to read any, you flagged them, and then you read those and the rest would not be shown again. You could easily get through thousands of long messages on hundreds of topics in just a few minutes, an ability rarely duplicated today. It is actually puzzling why it hasn't been duplicated. The all-keyboard interface was also very quick for a touch typist, quicker than typical mouse interfaces.

Users and platforms

One reason we don't get all we want is that the motives of users (readers/writers) and platforms are different. Platforms want to make money, which usually means they want to be attractive to advertisers. Writers want an audience and to be heard. Readers want to engage, be entertained, and keep up without fear of missing out. Successful systems find a balance among all of these, but no balance has been perfect.

Each new platform picks different positions in this taxonomy, and they often invent new ones to expand what is possible. While Twitter and Facebook have now held prominent positions for around 15 years, dominance in this space is usually short-lived.


Symmetrical vs. Asymmetrical following. This distinction is important because it determines, to a large extent, the type of community and culture that forms on the network.

At Google, I learned that there were two types of social networks: "Affinity" networks in which your connections represented people with similar interests, but which may not be known to you personally, and "Close ties" networks, in which your connections represent people you know.

Twitter uses an asymmetrical model - following is unilateral, you don't have to accept incoming friend requests. This is the best model for an affinity network, but it's a poor model for a close ties network. The reason is that for close ties, you really need for communication to be reliable - you don't want to miss important messages about your family.

Most networks use a hybrid model, where you can either "Friend" or "Follow" someone - however, most networks started with just the symmetrical ("Friend") connection model.

I should add a note about that. It is part of the "who are you writing to" question, and whether you are writing to people you know, or people interested in you.

The reason modern social networks do not make it easy to skim thousands of messages in a few minutes is the advertising business model. It is not beneficial to the bottom line to speed you on your way; they need you to linger and be exposed to ads. See also RSS and its gradual disappearance from most people's consciousness.


You could be right, but I am not sure they are right. At a certain point, being able to get through things quickly does not reduce the time spent, it increases the number of messages processed. But generally yes, those who fall into the ad-supported camp do serve the advertisers rather than the users.

(Though it is worth noting that Netflix, which has no advertisers -- yet -- still tries to keep you using it as long as possible.)

RSS was a kludge but yes, it is sad it has faded away. The reality is that pure centralization is bad, and pure decentralization is also bad because it slows certain types of innovation, which makes it lose the game.

There's a different approach where a writer decides for each idea they want to write about, what form they want it to take, and to the extent it can be shared with other people, they accept that. I went that way in 2017 after giving up peering with all the different approaches and their limits. There was no way to write for all of them. No single platform other than the one I wrote allowed links and styling, optional titles and enclosures, unlimited length, and allowed me to update my post if it changed.

Between them all was an ideal platform, if there had been even slight cooperation. Ultimately as a writer, while I do want the ideas to spread as well as they can, I write mostly for myself, to develop ideas, for fun, for a record so I can see how I was thinking. So it worked better. But now I'd like to try again to see if we can't get some interop at a meaningful level between the current set of popular writing platforms.

Also, how unfair of you to call RSS a kludge. A lot of thought went into it, and it has stood the test of time. You may not use it but a lot of people do.

Here's a summary of the ideal writing platform, one which can peer with other writing and reading platforms and doesn't lock in writers.


I view Markdown (which I am using to write this) as not part of the platform, but a creation tool. The rich text war has been fought and HTML won, so I think the right thing is just to define a subset of HTML and you can make that from Markdown, or your favourite wysiwyg or macro language. It is, however, nice if many platforms allow Markdown as a type of input.

The biggest problems with inter-system exchange formats though is that it's hard to do them without limiting innovation. The proprietary platforms have this advantage -- they can come up with a new idea, and it works fully within their platform. A big reason USENET died is it was impossible to innovate. It never even accepted HTML, or a text-only subset of HTML. (People feared rich text forms for fear they would make the text network get overloaded with images, and that's a valid fear, but there were ways that could have been avoided, and also ways that images could have been integrated usefully without harming the network as bandwidth got cheaper.)

Subject lines are an interesting one. I think they are very reader friendly, but not writer-friendly because they put an extra demand on the writer. As a reader, I would love it if every message in a thread had a sub-headline, but posters will never do that. The solution to that has arrived with AIs that can write decent headlines for you and we'll see the rise of AI (which is a whole new article) in message sharing to make everything as writer and reader friendly as possible. As well as in filtering.

The reason I used that term for RSS was that it used polling to implement publish/subscribe. That's the kludge part. And I understand why, because you could do that by just generating a page on a web site, so you didn't need to create a pubsub system. And because it's a bit more anonymous. It's just not efficient and also less immediate.

The use of polling doesn't require it, but it also led to systems having to implement their own "what's new" calculation rather than having that inherent in the push update approach. This has led to it being buggy in a number of systems.

But I get the tradeoff. E-mail lists, USENET, instant messengers were all pubsub, and reading blogs by polling seemed backwards to me at the time.

Not that this is easy. You may never see this comment because you would have to poll, or put the comment feed into your RSS. Because I don't have everybody who comments here create a login, I don't have the feature such sites usually have to let you get email notifications of replies to you.

It is unclear you would ever get a way to write for all of them. It's a lot of work for me to put my items on Forbes, my own blog, Facebook, Twitter, LinkedIn. The last two I do with an RSS to posting tool.

What would have been nice would have been a simple pubsub notification protocol, established enough so that ISPs and others maintained servers, and tools could manage all my subscriptions, easily creating and destroying them and doing it in an anonymous way. (There might be polling at the last level due to NAT, or a constantly open connection.)

There is rssCloud which provides instant updates. It was part of RSS 2.0, published in 2002, over twenty years ago. It works. That's about all I have to say about that. ;-)

To create nirvana for you. What can you do to create the writing environment you want.

RSS is just a syndication format. For such a modest thing it has helped a lot. When people said all the nasty shit about RSS, they are doing what you're doing -- giving it all their grief. It's just a freaking syndication format. Try being nice to it and see what happens instead of giving it all your grief.

And sorry for making you think I was faulting anything as a kludge except the use of polling.

It is useful, and as I wrote had the advantage that you could just do it on your web site. There's a lot of merit in that.

It's not RSS really that has been in decline, it's independent publishing -- blogging, etc. The old world of a million independent blogs, linking to each other, that did OK for a while and there are still a number of top sites that do well, but the big aggregators -- Twitter, Facebook, Reddit and a few others, took things over. Even Medium, Substack, Wordpress etc. are a minor part of the game today.

To make a better distributed system was possible, and technologically doable. But that's not where the problem was.

Add new comment