Conquistar el trabajo es reconquistar la vida

Grupo de Cooperativas de las Indias

GNU social: Federation against the social model of Twitter

“Federation issues” may look like a “bug”, but they are really the result of an agreement, an implicit contract: to be part of a conversation on another node, I first have to have received the trust of someone who is taking part in it.

Hazte socio de «El Arte» y haz cerveza


The Facebook and Twitter socialization model, the FbT model, is like a large plaza where everyone can shout their slogans, while barely listening to each other and without taking responsibility for looking at context and understanding conversations. The result is like a big chicken coop, a “fray,” where any attempt to maintain a conversation on any topic is immediately cut off by an avalanche of slogans and aggression by users who, quite possibly, haven’t even read the article that led to the conversation.

Why does GNU social create more value in its conversations than Twitter?

lamatrizIt is no coincidence that what users most value is having “fewer links that on Twitter, more characters and more conversation,” “a space without noise for calm conversation,” “speaking calmly and dealing with other topics,” etc. All these messages point to the intimate relationship between the value of a conversation and the trust that has already been established within the nodes. It is a consequence of the distributed structure of GNU social. Thanks to it, GNU social is free of any recentralizing tendencies and builds the network based on independent nodes — generally formed by affinity between groups of friends who communicate with each other thanks to the federation of content.

What is “federation?”

villa locomunaThe connections between the nodes of GNU social are established by the users who follow each other. Through these “following” relationships, all nodes can communicate and form a network. It’s what’s known as “federation,” and could be understood as a network of agreements.

All it takes is for me to follow a user on another node for everything that that user publishes to be visible to all members of my node. Thanks to this, you can see not only messages from the people that you follow in your inbox or on your personal time line and messages that are published in your node on the public timeline of the node, but also a much broader collection of messages, “the whole known network,” where, in addition to previous messages, you’ll be able to see messages from people in other nodes who at least one user in your node follows.

This creates wonderful things, like “the whole known network” being different in every node, because its composition is based on the people you follow and who follow your nodemates (or “nodies”). This is a very valuable aspect because it means joint exploration of the network. And starting from the existing relationship of trust between the members of a node, each time a member of the node follows — which is to say, establishes an agreement with — a user on an external node, the space of trust is expanded.

The key to creating space and favorable conditions for conversation is that the federation of content is based on what the users of each node follow on others, and not the general aggregation of all content by all nodes. The result is that if a person that neither I nor anyone else on my node follows says something in a conversation, I won’t see their posts. This might seem like a “bug”, but it’s really the result of an agreement, an implicit contract: to be part of a conversation of another node, I first have to have received the trust of someone who is taking part in it.

“Federation issues”

federationissuesThis model of federation is criticized by many new users who land on GNU social having had the experience of socialization of Twitter and Facebook. They label this difference “federation issues” and complain that conversations they participate in only show messages from the person that they themselves follow or other people in their node. The solution is as technically simple to implement as it is dangerous.

What such a request would do, in reality, is break the federation of content based on implicit contracts and open the doors to the aggregation of everything, everywhere, breaking any chain of trust. That is, it would remove the basis for allowing the nodes to create spaces for real conversation. By breaking this model of federating content, we would be importing the social model of the great centralizers, the Facebook-Twitter model, into the spaces and networks that we built on the basis of tools like GNU social, Diaspora, Friendica, etc.

Massive socialization through Facebook and Twitter has impoverished conversations and cut off the birth of new identities. It has done so by imposing a narrative about how the more accessible any conversation is to anyone, the better a network and its interactions are. In other words, when it is not necessary to have a minimum of prior trust to be able participate or interrupt the conversation of others. However, the search for this kind of accessibility obscures the very basis of distributed networks: the fact that a distributed network is made up of nodes, of independent groups that communicate among each other.


The problems or defects of the federation of content are only such if we accept and approve of the FbT socialization model. Really, we should call them “federation advantages,” because if which we’re seeking is to build enriching and conducive spaces for conversation, what we have today in GNU social is the structure that makes it possible.

The federation of content based on following relationships — agreements between people — is the base on which to build enriching and conducive spaces for interaction and for conversation. This is a determining aspect to not give in to centralizing pressure and turn spaces built with GNU social into a new version of the chicken coop that Twitter or Facebook currently offer us. The distributed structure of servers is “invisible,” and if we change the spontaneous logic of federation so that the user sees the network and behaves the same as in a centralized network, we will have changed everything to keep everything the same.

The world of the federation of content is passionate, and will largely determine the future of the web. Speaking concretely of the model of the federation of content, we sincerely believe that the challenges that we have to confront are in developing private communication and enlarging the system of exchanging short messages to a system where we can share everything useful — creating networks of hospitality, supply and demand, music, etc. — for our circle of friends, associations, community and surroundings.

That is, we believe it would be a mistake to replicate the centralized model and its culture. That would serve information without agreements between people, and therefore, approve of irresponsibility and encourage confrontation. For us, GNU social’s priority should be on becoming the “Swiss Army knife” of distributed networks based on sharing, by developing a culture of socialization based on trust within the nodes and the responsibility for understanding what is being talked about when someone joins a conversation. And for that, the key is to connect through federation, as has been done so far, on the basis of the minimum responsibility that comes with the fact that, to be an equal on another node, someone from that node has to considers what I say interesting enough to follow me.

Translated by Steve Herrick from the original (in Spanish)

«GNU social: Federation against the social model of Twitter» recibió 16 desde que se publicó el Miércoles 8 de Abril de 2015 dentro de la serie «» . Si te ha gustado este post quizá te gusten otros posts escritos por Manuel Ortega.

Comentarios recibidos en este post y unidos a la discusión global de todos a través de la Matriz, nuestro espacio conversacional.

  1. manuel dice:

    #Federation issues or advantages?

    “Federation issues” may look like a “bug”, but they are really the result of an agreement, an implicit contract: to be part of a conversation on another node, I first have to have received the trust of someone who is taking part in it.

    cc @mmn @mattl @knuthollund @aroque @simsa0 @lnxw48 @mk @robmyers @rozzin @erkan

    !gs !gnusocial !fediverse !gscamp

  2. simsa0 dice:

    A good essay! I not necessarily agree w/ all but you found an angle from which to describe (and cover) a lot.

  3. simsa0 dice:

    As you yourself do in the last sentence of your text, I would rather use "interest" instead of "trust". #federation then becomes less this idealized way of enabling and fostering conversations that derives its "intimacy" from some mutual agreements. Rather, federation is a technical means to distribute interest. And as we can often observe, interest isn’t the same as trust. The latter *builds over time* (and isn’t there at the beginning), while the former can be intensive but also rather shortlived. Otherwise, every "un-following" would be a breakdown in trust, which I don’t find to be case. That doesn’t hinder, OTOH, that people who already trust each other will find federation to be good way of displaying it.

  4. manuel dice:

    @simsa0 I think that the key to understand why when you subscribes to someone you are showing not only "interest" but "trust" is to be conscious that the person who are subscribing to is nor just and only one more increasing the number of followers as in Twitter or Facebook.  In the !gnusocial federation model when you subscribes to someone your are adding a source into your node and telling the people in your node that "this in interest and I trust it". Two things:

    * It is desirable that there is a previous trust between the users that share a node. Only this way #gnusocial will be not a just a an free alternative to the social model of Twitter but a completely different social communication platform with a totally different social model. Because of this I think that the reminder that @knuthollund did a few days ago is important and paramount. It is «don’t forget to create more collaborating GNU social instances. Families, friends, clubs, communities etc.»

    * I hope that users do not just subscribes to others just they likes big numbers and wants to show how popular they are. I hope they are subscribing to other users because they have been reading and they find that should be great to join him or her to the conversation 😀 This is part o a share responsibility

  5. simsa0 dice:

    Honestly, I don’t don’t subscribe to someone on Twitter b/c I want to increase his numbers; nor do I want to be followed just so I can show off a bigger number. So I rather don’t agree with this supposition. And when I subscribe to someone in the federation I don’t do that to tell somebody else that I find the person interesting and *therefore* trust her. I subscribe to people b/c I find them interesting (which can be for different reasons and on several levels), not because I trust them. I don’t mistrust them, either but trust itself will build over the future communication.

    That the tone is more familiar in the federation, sometimes gentle, sometimes harsh and acrimonuos, seems to me due to:

    1) the people are not that different and share similiar attitudes;
    2) long-dents with more than 140 characters (which seeem the norm, not the exceptions in the nodes) allow for better and true communication (140 cs only fits for broadcasting)

    So to me it feels you’re reading a bit too much into it. But that we have this conversation proves your overall point 🙂

  6. david dice:

    @simsa0 @manuel I desagree with your idea that trust «isn’t there at the beginning». It is true in «public» nodes, but not in community nodes -already the biggest part of existing instances of gnusocial… which also are, from my point of view, its future, specially if -as manuel- proposed we develop all kind of «sharing economy» and «gift economy» plugins

  7. manuel dice:

    @simsa0 🙂 Yes, the artificial limit to 140 characters certainly is a big barrier for having a true communication. And I think we agree that the #federation as it works now, based on people connections, is paramount to make conversations possible. I think this is connected to your point (1), I mean, not all the people using #gnusocial have the same interests or contexts but the people you subscribes to and consequently the messages that you receive have at lest contexts in common.

  8. simsa0 dice:

    Perhaps this is really just a difference in emphasis. I will re-read your text & try more to suppose you were simply right … & then look what comes to mind thereof.

    In a sense I can see your stance: In Twitter you don’t need (much) trust in order to enter communication, & if something turns bad & the blocking starts, you may move to somebody else. In Gnu Social the level of trust we begin with when starting conversations seems higher — or is it simply the level of ease & relaxation?

    But your main claim has been that it is trust that underlies the willingness of people to subscribe to each other across nodes & thus drives the #federation. Here I am more reluctant, as the technical features seem to allow the same relations with or without the trust. But even then you may have pointed to something important: The features (like subscribing across nodes) wouldn’t be used in that extent were it not for trust rather than mere interest. Without trust, you might say, everyone would be more inclined to stay on his/her node, re-creating a silo-communication. I’m not sure, but will think about it.

    @manuel @david

  9. david dice:

    @simsa0 @manuel nodes as or created by pre-existing communities and networks for their members (I took the name from ). With the development of tools as !wpgnusocial is very probable that many other communities now organized around a blog, will install their own instances, ie has already declared that they will do it very soon, and I could name a couple of communities more only in the Spanish and Portuguese speaking world. In the anglophone world I can imagine, or doing something similar in some months.

  10. simsa0 dice:

    Ah, ok. I didn’t have the impression in the last years that nodes have been built around such (true) communities, rather that instances have been set up and people joined them for this or that reason. But given your take, I understand why you find trust to be the basic level of it.

    But even in your linked , there are more "public" than "community"-nodes. Unfortunately, ‘s compilation of social instances doesn’t tell if the nodes are individual, public or communal. So I don’t know what came first or what is more prevalent — which, OTOH is not really important either. Maybe the #federation and the communication just changes right now, so that trust will play a far more important and constructive role.

    @david @manuel

  11. steve dice:

    @simsa0 @manuel I’m joining the conversation a little late, but it seems to me there are different levels of trust. I wouldn’t trust someone on another node to give me financial advice or babysit my kid, but I would trust them to not be a complete waste of my time when I read what they write.

  12. After the publication of this post we have received important inputs for this conversations. Here I would like to point out some of them:

    @zoowar said «Have you ever been to a dinner party with more than 6 people? It becomes difficult to have a single group conversation. We can’t all speak at once, and when we do, not everyone hears us, so we break into groups. These groups are dynamic throughout the meal, but they form, and not everyone is part of every conversation. This is why good hosts spend time seating people near others with similar interests»

    @mmm said «Often claimed solutions are:

    * Spam-distribution is unacceptable.
    * Auto-fetching. Must be analyzed for issues relating to privacy (how/when is it fetched?) and performance (DDoS possibilities etc.). I for one don’t want to implement something that gets easily abused.
    * Distributed storage might solve the problem of "missing notices", but it causes several new issues which are even worse. I can extrapolate if desired, but I don’t want to rant too much»

  13. simsa0 dice:

    I wouldn’t put too much emphasis on the similarities & differences of GS-conversations & dinner party-conversations. Neither would I try to align them as closely as possible, as they are both Real Life conversations, just happening in different environments.

    However, at a dinner party one can see different groups of people & can leave & join them as one moves around. In a conv-thread one often wouldn’t even know that there are such other groups and would have to hop from node to node, suspecting that one is missing stuff.

    Not showing replies in a conv-thread OTOH is not the same as a private chat in some back room of the party or some group wandering off. It’s rather like people leaving the party all together.

    Showing all replies in a GS-conv doesn’t mean one has to read them all; one can skim over them, dive into the interesting ones and leave others aside.

    Showing them all is also not a matter of trust. Especially at a dinner party trust is likewise not an issue: You’re already invited and expected to wander around. The counterpart in GS is: one dents publicly, & people subscirbe to each other.

Pingbacks recibidos desde éste y otros blogs

Deja un comentario

Si no tienes todavía usuario puedes crear uno, que te servirá para comentar en todos nuestros blogs en la
página de registro de Matríz.

El Correo de las Indias es el agregador de blogs de los
socios del Grupo de Cooperativas de las Indias:
F-83409656 (Sociedad de las Indias Electrónicas S. Coop.)
F-85220861 (El Arte de las Cosas S.Coop.)
G-84082569 (Las Indias Club)