Liberationtech

Info Tech for Social Change

How the Next Generation Diaspora* Should Be Built to Help High-Risk Activists

An online privacy activist recently asked me: Suppose you were to build the next-generation Diaspora* — i.e., a secure, private, and decentralized social network — how would you go about it? 

The question is an important one, especially considering that many projects preceded Diaspora* but failed to gain traction, along with the skepticism with which Diaspora* has been met in hacker circles. 

Hacker News has been particularly vicious, with attacks on Diaspora*’s security and privacy code implementation from the get go and with criticism of the Diaspora*’s team ability (or lack thereof) to implement its vision

Criticism has also come from the mainstream media, where reporters have wondered “whatever happened to Diaspora*" and "what’s taking so long,” as though building a secure, private, and decentralized social network were as easy as building a centralized alternative like Facebook.  In this context, credit should be given to the Diaspora* founders for trying to advance the vision by learning from the mistakes past projects have made in this space.

II. What is the goal?

One of the first steps to undertake when answering the question posed by my online privacy activist friend is to determine what the goal of such a next-generation Diaspora* would be.  For example, if the goal is to gain traction among mainstream users, as Diso creator Steve Ivy has suggested, then the focus would not be so much on the merits of the technology at ensuring security and privacy, as it would be on its ability to use decentralization to overcome Facebook’s considerable network effects.  As one of Liberationtech's coordinators, I'm much more interested in the former than the latter, for there are many people in the world who care about security, privacy, and decentralization for its own sake, and there is a very compelling reason for giving these people such a solution, i.e., their very lives depend on it.  Before we delve into that topic, however, I'd like to address the question of how one would overcome Facebook's network effects. 

III. How does one overcome Facebook’s network effects?

As you may recall from economics, a network effect is the effect one user of a good or service has on the value of that product to other people.  A network with a lot of people has more value than one that has fewer people.  For example, if you are looking for a job, or searching for people who share your interests, you are more likely to find them in a larger network than in a smaller one.  Since people will choose to join the larger network at the expense of the smaller one, one will ultimately end up with one giant network, as barring some kind of niche offering in smaller networks, one is unlikely to find any value in the smaller networks as the number of users on those networks dwindles. 

This process also illustrates how difficult it is to persuade one person to switch from one network to another.  A person benefits from her participation in a social network because she has ties on that network.  That person’s friends benefit from their participation in that network because of their ties.  As such, to persuade someone to switch from one network to another, you must not only persuade that person to make the switch but also that person’s ties, thereby creating a chicken-and-egg problem: That person will switch only if her friends switch, and the friends will switch only if that person switches.  Thus, overcoming network effects is a group problem, not an individual one: You must create a social movement of sorts to encourage people to switch from one social network to another, or at the very least, create an information cascade or bandwagon effect that encourages people to switch.

For those who may be skeptical about the strength of Facebook’s network effects, consider the following:  Polls regularly find that approximately 70% of users are concerned about their Facebook privacy and security, yet according to a proprietary Forrester study only 4% quit Facebook for this reason.  In fact, nearly half of those who quit Facebook do so because they were bored with Facebook or found a better niche site elsewhere.  These numbers suggest the strength of Facebook’s social network effect.  Given the seriousness of people’s security and privacy concerns, one would expect a much larger number of people to quit.  Yet they don’t do so because quitting would mean losing touch with your friends and other contacts on the Facebook network.

You may immediately notice, however, that this economics story is one-sided: The assumption is that advantages in network size will create an inexorable trend towards consolidation, yet the disadvantages in network size that could create an equally strong or more powerful effect away from consolidation is left unexplored.  We know, however, that such effects exist.  Otherwise, how would MySpace have replaced Friendster in the first place?  Or how would Facebook have replaced MySpace?

The question that arises then is the following: How does one overcome Facebook’s network effect?  There is less research on overcoming network effects than on their inevitability, but some possibilities immediately come to mind. 

A. Deep Pockets

One obvious possibility is that a competitor may come along with pockets sufficiently deep enough to challenge the entrenched network.  Such a competitor could spend considerable resources on marketing and advertising to attract users to switch from the dominant network to the competing one.  Yet, as Google+’s experience has shown, this process may be neither immediate nor successful.  The verdict is still out, but Google+’s recent experience suggests that deep pockets may not be enough to counter a leading network’s network effect.

B. People Discovery

A second promising alternative is people discovery, i.e., a social network that enables you to meet people you don’t know.  Despite an extensive academic literature that suggests that people are distrustful at meeting strangers in real life, proprietary Mintel data suggests the opposite:  Nearly 50% of those surveyed say they would like to meet strangers online, and many admit to “friending” strangers on a regular basis, including women, who are generally assumed to be much more distrustful of strangers.  Web inventor Tim Berners-Lee sees people discovery, or stretching one’s ties to meeting people who are different from us, as the next social networking frontier, and companies have heeded the call, as Altly’s transformation from a private social network to a people-discovery engine named BetaBeat has shown. 

It’s unclear, however, whether people discovery will be sufficient to overcome Facebook’s network effects, especially since Facebook has sufficient resources to copy any social network innovation in this area to its advantage.  Moreover, by virtue of Facebook’s larger pool of users, the company should be in an ideal position to introduce people to others they do not know. 

As Twitter has shown, however, people discovery has the distinct advantage of circumventing Facebook’s network effect.  If a new social network focuses on people who want to meet those they don’t know, then those people are also less likely to care whether their existing ties are on that new network, since by definition they don’t care as much about their existing ties as they do about establishing new ones.  Twitter has capitalized on this insight, and Facebook has recognized it, as the latter’s ongoing transformation from a private, close-tie, college-based campus network to a public, weak tie, international network has shown.  In economics, some have argued that countries pursue protectionism when they are poor and free trade when they are rich.  Similarly, Facebook was for privacy as a small network and is now for openness as a large one. 

C. Technical Superiority

A third possibility is to find a technical feature (or set of features) that are more valuable than those offered by the dominant network.  In other words, one would focus on technical advantages that overcome the social advantages created through network effects.  An example of this can be found in how Google overcame Yahoo at search.  Google had a search algorithm that generated better results, and over time, people gravitated to Google over Yahoo. 

One example that comes to mind in the network space would be the application of natural language processing to enable users to get more benefits out of their social network ties.  Facebook does this through filtering, albeit not as transparently as many would like, leading thinkers like Eli Pariser to complain about the dangers of “filter bubbles.”  In fact, there is a battle brewing between Facebook and Google+ in this area, as natural language processing is seen as the potential driver for a new wave of social network interactions.  Nevertheless, just as Google did to Yahoo, it is entirely conceivable that a new network could come along with a proprietary algorithm in natural language processing that could give it a similar technical advantage over Facebook or Google+ in the social networking marketplace. 

Moreover, there are many unexplored innovations in this space:  While computational researchers have made advancements in the study of syntax and semantics, pragmatics remains a relatively black box, despite media hype to the contrary.  In fact, the most sophisticated research in this area comes not from computer science but from social network analysis in the behavioral sciences.  Surprisingly, however, programmers have yet to mine this extensive literature for behavioral insights with which to construct better social networking sites. 

D. Total Institutions

One final possibility comes from the realm of total institutions.  A total institution can be defined as a place of work and residence where a great number of similarly situated people, cut off from the wider community for a considerable time, together lead an enclosed, formally administered round of life.  Examples of total institutions include monasteries, the army, prisons, and psychiatric institutions, among many others.  Total institutions are dense locations of activity, where ideas can spread quickly, and thus they are ideal locations for fostering the growth of social networking sites.

You’re probably thinking: “That’s crazy. Are you suggesting that we build a social network out of an insane asylum?”  But before you discount the idea, remember that this is exactly one of the reasons why Facebook became a dominant player in social networking.  Facebook, in fact, capitalized on the most influential total institution of Western society, i.e., the college campus.  On college campuses, students work and live together, and they share similar values and engage in similar activities, cut off from the wider community for at least four years.  Facebook’s strategy, as I have explained elsewhere, focused on controlled growth and saturation from one college to the next.  To the extent that female students had privacy fears about joining Facebook, these concerns were assuaged by the fact that Facebook only allowed people who had university email addresses to join, such that the number of potential whack jobs were limited to those that you knew on your college campus, not those that you did not.  This may also help explain why Facebook, unlike other social networking and dating sites, is predominantly female, and why men — as even the Facebook founders themselves acknowledge — were so attracted to joining Facebook in the first place.

This story raises the question:  Can a new social networking site challenge Facebook by taking over college campuses again?  The answer to that is unclear.  Facebook remains strong among college campuses, though the bulk of its growth is now coming from older demographics, such that the proportion of college students in the network has fallen.  You could say that Facebook’s strategy has now shifted from the campus as a total institution to the elderly home as a total institution.  At the same time, college campus-specific social networks have been launched in recent years but have made scarcely a dent on Facebook’s college-age numbers. 

As such, a better question to ask is the following:  Are there other total institutions out there that social networking entrepreneurs can tap into to challenge Facebook’s dominance?  I don’t really have a good answer to this question, so it remains rhetorical.  But to the extent that Diaspora* has gotten more traction than other social networking sites, it is because it has tapped into the free culture movement, hackerspaces and maker spaces, and so on.  Similarly, though Silicon Valley has an aversion to politics, a social networking site that is built out of movements such as Anonymous, WikiLeaks, or the Occupy movement may be able to attain significant traction, if timed properly.  In short, while the number of pure total institutions in our society is limited, it is clear from Diaspora*’s experience that a group-based social networking recruitment approach may work better for social networking entrepreneurs than the traditional individual-based approach they have followed to date.

IV. Getting back to the task at hand

Overcoming Facebook’s network effect, however, only matter to the extent that you want to build traction to supplant Facebook.  But suppose you’re not interested in traction.  Suppose that what you care about is to create a secure, private, and decentralized Facebook alternative that protects high-risk activists fighting for freedom, democracy, and human rights in oppressive, dangerous environments.  Then, many of the mainstream users’ considerations drop out of the equation, and the problem becomes much more focused and manageable — albeit still difficult.  At the same time, however, mainstream users who care about privacy and security can still use the solution, if they are so inclined. 

As a Stanford liberationtech coordinator, you can see why I would be so interested in such a solution.  The goal of our program is to conduct research and design of information and communication technologies to foster freedom, democracy, human rights, development, and effective governance.  In other words, we want to figure out ways in which technology can support the dangerous work that activists conduct every day to create a better world.  A secure, private, and decentralized communication platform would help support activist efforts to this end.  And such a platform only needs traction among activists, not all mainstream users, to succeed.  In other words, it needs to solve the activist problem, not the mainstream user’s problem, to be most effective.

A. Organizing versus broadcasting

But, you may ask, aren’t movements like the Occupy movement, the Arab Spring, or the Indignados more interested in spreading the word?  As such, how can you give up on traction in pursuit of this goal?  To answer these questions, it is important to differentiate between what activists do before a movement and what they do during a movement.  As my doctoral dissertation shows, before a movement, an activist needs a private and secure platform to organize with a small group of people.  These are the people who lay the groundwork for what the movement is to become.  Authoritarian regimes understand this, which is why they seek to stamp out the early-movers, and why they immediately crack down on any signs of free assembly.  When groups of people are able to assemble in such environments, that’s when the regime’s days are numbered. 

If people are able to assemble, then the activist’s task changes from organizing to spreading the word.  It is at this point that traction, or the broadcast capabilities of a social networking site, become important.  But as we have seen, large mainstream social networking sites like Facebook and Twitter are more effective at doing this task.  Once activists get to the broadcasting stage, what becomes more important to them is to protect their identities as they spread the movement’s message.  But the organizing task is never completed.  The organizing task continues.  And it is this organizing task that I care about most.  This critical organizing task is done by a small group of people that need to be able to maintain strong ties to one another in a secure and private fashion if they are to succeed.

This explanation starts to draw the raw schematic of what a next generation Diaspora*-like private, secure, and distributed social network should look like, if it is to achieve Liberationtech ends.  The network should facilitate the communication of a small group of people seeking to organize social change and subsequently enable them to broadcast that message through larger mainstream social networking sites to generate the strength-in-numbers that can help movements grow and ultimately bring about political change.  In other words, it must be a secure and private social networking site with HootSuite-like capabilities that can protect the anonymity of the person broadcasting messages to the larger and more mainstream social networking sites.

B. Decentralization

So far, my emphasis has been on security and privacy alone.  But decentralization is inextricably tied to security and privacy and equally important.  What do we mean by decentralization?  Decentralization means that instead of having to post a message to a central server like Facebook, and then wait for that server to transfer that message (or not, in the case of censorship) to your friend, you send that message to your friend directly.  To achieve this, communication must be machine-to-machine, where the sender controls the first machine and the recipient controls the second, and the message that is transmitted is encrypted to ensure that only the sender and the recipient can read it.  In other words, the sender and recipient must have an easy and fast means to install and manage the software on their machines — whether these machines are servers, computers, or phones, as in the FreedomBox vision.  Furthermore, the sender and the recipient must have the ability to stop using their machines and seamlessly use new ones, should the original machines be compromised for whatever reason by an authoritarian regime.  The software would need to have an easy “self-destruct mechanism” such that the data can be destroyed immediately in an emergency.  At the same time, the “right to forget" would have to be embedded from the get go, such that the data would self destruct after a certain period of time to prevent a trail of communication that would make it easy for an authoritarian regime to track down the activists.  As such, the next generation of secure, private, and decentralized social networking site would create a one-click turnkey solution for activists that could easily be discarded if compromised and whose data could be destroyed automatically as the utility of the data diminishes while organizing unfolds.

C. Mobility

There is one final consideration.  Activists are constantly on the move, such that the social networking site will need to be mobile from the get go and have the capability of synchronizing data on multiple machines simultaneously.  Thus, if the activist needs to coordinate with others elsewhere, she must have the capability to access her data from the alternate location.  Similarly, there will be times when the activist will attend a street protest, and the relevant social networking data will need to be accessible on her phone.  Other times, the activist will need to go to a “safe house" and access her data from there. 

Moreover, connectivity will vary greatly.  At times the activist may have access to broadband Internet, but other times, she may need to connect via a 56K modem, a mobile connection, a mesh network, or perhaps even a satellite link.  The social networking site will need to be accessible regardless of the connectivity, which means significant work on data compression will be required to ensure that the software’s performance remains nimble under such disparate conditions.  This creates difficult challenges for the developers of such an application that developers of mainstream applications would never have to encounter. 

D.  Cooperative

So how does one draw the necessary resources required to overcome the aforementioned challenges of security, privacy, decentralization, and mobility to build such a social networking site?  Western society gives us two main legal-institutional vehicles for tackling the problem:  i) a for-profit firm a la limited liability company or C corporation; or ii) a non-profit firm a la private foundation or 501(c) organization.  (Another possibility is a hybrid for-profit/non-profit model a la WordPress or Mozilla, but let’s set that aside for now.)  In either case, a group of individuals — usually, the founders — become owners of the organization and raise the necessary resources needed to execute the organization’s mission, implement its strategies, and reach its goals.  

A for-profit organization like a C-Corp is ideally suited for the task because the founders can sell an ownership stake in the firm in order to raise the requisite resources.  But as the saying goes, there’s no such thing as a free lunch.  The resources come at a cost in terms of the organization having to perform in a reliable and accountable fashion relative to the expectations of its shareholders.  In the pursuit of profit, principle can easily be abandoned since, at the end of the day, all the shareholders care about is obtaining superior returns relative to what they could receive by investing elsewhere.  If the firm is able to secure superior returns, however, other prospective investors will be attracted to the investment opportunity, thereby providing the organization with the resources to grow over time.  In the end, shareholders matter more than customers — and for our purposes, the activists risking their lives for freedom, democracy, and human rights — for without the shareholders there is no business.     

On the surface, a non-profit organization looks better on principle grounds because the organization is not acting on the basis of profit alone.  Nevertheless, a non-profit organization is still owned by a small group of individuals, and as in the case of a for-profit firm, controlled by its board of directors, which means all decisions with respect to the organization and its customers — or in this case, activists — are made by the board.  This means that, in both the for-profit and non-profit cases, the product at the end of the day is determined by decisions resulting from the good will and discretion of a small group of individuals.  You’ll probably be surprised to hear that it’s often also the case for open-source projects like Diaspora*.  As Karl Fogel’s book “Producing Open Source Software: How to Run a Successful Free Software Project" teaches, most open-source projects are run by "benevolent dictators" in whom "final decision-making authority rests" and "who, by virtue of personality and experience, is expected to use it wisely." Even in the case where the project is Affero General Public License (or AGPL), the benevolent dictator can make decisions that can prevent the right technologies from being implemented over the course of the project.  The project may even create disincentives for open-source involvement by creating restrictive intellectual property (IP) assignment contracts that require developers to give up all rights to the code they produce.  And worse, a non-profit organization cannot sell shares, which means that there are no financial incentives other than the generosity of donors to raise the resources required to develop it.

So to summarize, on the one hand, there’s the for-profit firm that can sell shares to raise the necessary revenues to develop a product but in many cases may sell out principle in pursuit of profit, and on the other hand, there’s the non-profit firm that has to depend on donations but, as in its for-profit counterpart, still makes the activists beholden to the actions of a few individuals.  Given this predicament, what are we to do to ensure that the organization is accountable to the activists it serves and can mobilize developers to contribute in an open-source manner to the project?  One possibility is the cooperative, a business organization owned and controlled democratically by its members for mutual benefit.  The cooperative can range from for-profit to non-profit, depending on the project’s ultimate goals.  Thus, while the cooperative is not a magical solution to all of the aforementioned problems, it can help ameliorate many of them, when correctly designed and executed.

The advantage of the cooperative for purposes of the task at hand is that it can ensure that the organization operates in a democratic and accountable fashion relative to the developers who contribute the code to solve the aforementioned technical challenges and relative to the activists who risk their lives using the technology to do their jobs on the field.  The developers can transfer their IP rights to the cooperative, knowing that such rights will not be exploited for financial gain without them.  Similarly, the activists can know that the organization has their best interest at heart and thus can trust that the solution will be built and subsequently developed with their needs and concerns in mind.  

V. Conclusion

So in response to my online privacy activist friend’s question about how I would build the next-generation Diaspora*, my answer is this:  I would create it first and foremost as a secure, private, distributed, and mobile platform with HootSuite-like (but anonymous) broadcast capabilities and fast and reliable performance under rapidly changing conditions.  But I would make sure to work within a cooperative legal-institutional framework to find the correct design that makes the organization accountable both to its developers and customers, i.e., the activists that the social networking site is meant to serve. 

About the Author:  Yosem Eduardo Companys is a PhD student in engineering at Stanford University and a coordinator for the Program on Liberation Technology at Stanford University.  Yosem also worked as adviser, mentor, and consiglieri to the Diaspora* founders and as President & CEO of Diaspora*. He may be reached at companys[at]stanford[dot]edu.