Social Networks

The Wire blows a fuse

This blog follows The Economist’s style guide, which prescribes the spelling “program” in reference to computer programming and “programme” in all other contexts. As I work for an American corporation, in my official mails, I use “program” for both senses. Keeping track of these differences in spelling does cause some mental strain.

The Wire, the torch bearer of the anti-establishment movement in India, is too busy to follow style guides. Its exposé claiming that Meta gives Amit Malviya, the head of BJP”s IT Cell, plenary authority to take down any Instagram post he reports changes the spelling from “programme” to “program” and back within the same article, indeed in one instance within the same paragraph. It is understandable that given the focus on saving the nation from bigotry and authoritarianism, the small matter of which colonial power’s spelling to use does not get much attention.

In response to their exposé, they have received an appreciative mail from a whistle-blower a “senior employee in Meta’s privacy and security team”, who worked for the company for a decade who mentions his disquiet with the “XCheck programme”. The whistle-blower is a man after my own heart. After spending a decade in an American corporation (perhaps in the US) he is fastidious enough to switch to the British spelling when writing an email to an Indian publication.

I had not heard of the XCheck programme till The Wire made me aware of it. Apparently, the world was made aware of the programme’s existence by the WSJ’s Facebook Files series. The story on the XCheck programme is not behind a paywall, at least for me. I highly recommend reading it if you are able to. Unlike The Wire, WSJ adopts a sober tone, is built on multiple sources and describes Facebook’s position fairly. Reading the WSJ report, I could actually sympathize with Facebook’s point of view.

Facebook, like any social network, has some challenges when moderating posts by political officials, celebrities and well-known people in general. For example, there are rules that users shouldn’t use the platform to promote violence. But for government officials, threatening violence is often part of their job. For example, in response to a terrorist strike, a leader may say that they will bomb countries that harbour terrorists.

Secondly, even when a celebrity’s post clearly violates the social network’s rules, it may serve public interest to keep it up because of inherent newsworthiness. Traditional media generally handle this problem through editorial control. But due to the scale of social media, they have to rely on automated moderation, which is inherently flawed. Strange moderation decisions by algorithms quickly turn to memes; when prominent users are subject to those decisions, they become prime time news.

XCheck, or “cross check” appears to be Facebook’s attempt to ameliorate this impossible problem. XCheck users’ posts aren’t subject to automated moderation. Instead, they are “cross-checked” by some humans. The Journal’s exposé claims that in an attempt to avoid false positives, this process has led to many false negatives, such as celebrities posting revenge porn that stayed up for too long awaiting XCheck moderation. It also presents internal memos where Facebook employees express unhappiness about this.

But it is one thing to have lower moderation standards for prominent users. It is quite another to allow them an automated way to take down posts of other users. If The Wire’s reporting is correct, it would imply that the Wall Street Journal, that found so many bad things in in its Facebook investigation, missed this truly evil practice that was occurring adjacent to the questionable practice they were investigating. Facebook employees who were so freely expressing their misgivings about other questionable policies failed to express their outrage over this, or the Journal failed to find the smoking guns.

The Wire’s reporting shows no awareness that it is making an extraordinary claim requiring extraordinary evidence. It is so focused on taking down Amit Malviya that it fails to ask the basic questions – how widespread is the practice? Do all XCheck users get the privilege? Only those in India? Only certain countries? Only Amit Malviya? Are they aware of their privilege? Do they get a special interface to do the reporting?

Let’s say that all XCheck users got the privilege. It is a widespread programme encompassing many users, including those who are critical of Meta. Surely at least one of them would have noticed and blown the whistle? If only a subset got the privilege, wouldn’t it have raised some questions (to put it mildly) within the organization? If there were the smallest grain of truth in the report, the WSJ and other news organizations that have done critical reporting on Facebook/Meta in the past would be asking themselves how they missed something so big and would be tapping their sources right now. Instead, we are seeing The Wire successfully creating a fog of war around the authenticity of mails, which is causing us to miss the elephant on the battlefield.

In response to questions about The Wire’s reporting, its defenders have claimed that it is known for high quality investigative journalism, and this reputation should shield it from having to show its work. If The Wire has any reputation for investigative journalism, this reporting gives no evidence of it. The evidence, such as it is, looks more like an artistic depiction of an evil corporation than a realistic portrayal of how wrongdoing actually occurs. For an actual example of investigative journalism, look at the WSJ report once again, and then when you go back to The Wire’s report, you comprehend the damage the Tehelka model of journalism, optimized for social media soundbytes, does to journalism. The point of The Wire’s report isn’t to convince anyone who isn’t already convinced that there is a cozy relationship between the BJP and Meta, any more than the point of RRR is to convince anyone who isn’t already convinced of the evil of the British Raj. It is to create cringy memes that will be used by its supporters to share among themselves. In that, it has succeeded.

The Network Effect doesn’t make you invincible

Unlike Socialism, which fails in similar ways wherever it is tried, Capitalism gives us new things to worry about every generation. In a dynamic economy, companies from different industries threaten to become monopolies every generation, giving us new reasons for why this time’s threat is different from that of last time.

The threat our generation faces is apparently from monopolies that benefit from network effects. We are particularly scared of social media companies that exercise control over freedom of speech, who cannot be dislodged from their monopoly perch because they have the power of the Network Effect.

The Network Effect

You are trapped in the Network Effect when being part of the network may not be your first choice, but you have to be there because all your contacts are there. Your contacts are there for the same reason. It is a collective action problem. There may be a network that all of you would move to if only someone could coordinate, but that is an impossible task.

So everyone hates LinkedIn but everyone is still on it because all their professional contacts are there. They are on Tinder because all potential romantic partners are there, on Facebook and Instagram because all their friends are there and on Twitter because all their enemies are there.

The second reason why all of you are still on the network is that you all would like to move, but each one’s choice is a different one, so you stay on the current one as a compromise.

Network effects are self-limiting

It is a meme, popular among young women, to show four photographs of themselves representing how they wish to appear on different social networks – a sanskari look on Facebook, a professional look for LinkedIn, an exotic photograph of themselves for Instagram, and a sexy look for Tinder. A Twitter look usually does not appear in this set of photos, presumably because people rarely have photos of themselves having a meltdown and screaming at people.

This meme illustrates a truth about social networks and network effects. Network effects aren’t always positive. There are some networks you would go to because all your friends are there. For some, that is a reason to avoid that network. No teenager wants to be in the same social network as his parents, as Facebook is learning. As networks grow too big, there is a real possibility that they become infrastructure – just as email is, or they are so crowded that no one wants to go there.

The meme also teaches us that there is a market for more than one social network. In theory, we would like to have different social networks for different purposes, just as the young woman in the meme has, the better to ensure that her sanskari look doesn’t leak into Instagram and her love-life is not revealed on Facebook.

The vector of attack

In any industry, if a challenger wants to dislodge the dominant player, taking it head on is rarely the right way. The better way is for the challenger to attack an area where the dominant player isn’t noticing, doesn’t consider it worth its while, or is unable to respond.

The limitations of the network effect suggest such a vector of attack for challenger social networks. You won’t dislodge Facebook by building a better alternative to Facebook. You will do it by finding a group of people who want to be on a different network from the others, or a theme that a network of people can coalesce around, and begin your attack from there.

In practice, this is not easy to do. Managing multiple social networks is hard, which makes it hard for a new social network to gain a foothold. The challenger social network, in its initial phase, needs to have a well-defined theme. If you have ever tried to moderate a mailing list and tried to keep people to stick to particular topics, you would know that it is difficult to do. It is almost impossible to do at scale. People want to do fraandhip on LinkedIn and they use Facebook for professional networking. If it were easy to do, we wouldn’t have been having arguments about network effects in the first place.

But it isn’t impossible. There surely are some natural communities you can begin with – the community of youth is one obvious way, but I am sure there are others that are not so obvious. The Network Effect isn’t some unprecedented power that successful companies of this generation have, a power that makes them so strong that we have to reevaluate everything that we have learnt about the wisdom of breaking up monopolies. It is a power. It is a moat that is difficult to cross. It does not make you invincible. One day, the moats will be crossed, and this generation of companies too shall pass into oblivion.

But Facebook is buying them all up!

But isn’t Facebook is buying all the challenger social networks? It bought up Instagram and WhatsApp! If the dominant players keep buying up the challengers, who will be left to challenge the monopolists?

Well, calm down. A monopoly isn’t much of a monopoly if it takes work to maintain it. If the monopolist has to constantly scan the market for potential threats, move nimbly and introduce new features to maintain dominance or acquire any companies that can pose a threat, it isn’t enjoying a natural monopoly.

Every organization gets old. Once it gets old, it gets flabby. The founder either moves on or doesn’t have the same instincts he used to. The people working there don’t have the same drive they do when young. The organization gets too bureaucratic and procedure oriented. New ideas are stymied by turf wars. It doesn’t move as quickly to acquire or crush competitors as it used to. It is slower to introduce new features, or even to copy them from competitors. It doesn’t see emerging threats as quickly as it used to because it is too focused inwards and too focused on existing customers.

This has happened to every organization in history and it is a fair bet that it will happen to the companies that run social networks as well. I believe that we should be concerned about monopoly power only if it lets the old and flabby organization stay a monopoly long beyond its sell-by date. I don’t see evidence that this is happening. An hour is a long time on Twitter, but in historical terms, the current social media companies haven’t really been in a dominant position for long.

Social media will destroy democracy

We are all very concerned about the effect social media has on society, our political system and our democracy. We are all very concerned that companies that run social media will dominate the discourse and decide who can say what. We are all very concerned that social media companies are monopolies. My belief is – no, they aren’t monopolies. No, they don’t dominate the discourse and determine who can say what.

Yes, social media will destroy democracy, but if it is any consolation, it won’t be through the monopoly power of the companies that run social media.

Postscript: Why isn’t there a social network in India?

I’ve argued that one way to challenge the dominant social networks is to find a group of people willing to break away and form their own network, perhaps around themes that has little overlap with those of the larger network. Now, there exists a large nation named India. It has a vast population that speaks many different languages. It has many interests and cultural issues many of which overlap with the rest of the world and many which don’t. While this country is fairly well-connected with the rest of the world, it also has many people who should be willing to join a social network that caters to their interests. It also has a government with a lot of influence, and interest in supporting an indigenous social network. If the officials of this government begin extensively communicating on a challenger social network from India, it should attract a nucleus of influential people who can seed the growth of this network. It also has skilled programmers and a startup ecosystem that can build a new network. Could the next social network come from India?

Just kidding. It won’t.