Social networks face the wisdom of crowds

Elon Musk wanted to make sure his newly won control of Twitter would ‘sink in’ as he staggered through the doors of the company’s San Francisco office with a heavy ceramic kitchen basin. He left it in the foyer with the photographers before beginning a programme of layoffs and policy changes.

What may have yet to sink in with Musk in the following months is not just that running a social media enterprise like Twitter is surprisingly costly and beset with social issues, but that a previous CEO wound up bankrolling what may turn out to be Twitter’s replacement in the market. If successful, it might take other companies of the same Web 2.0 generation with it.

Technology executives like to talk of disruption. To break through, they need to disrupt whatever is in the lead spot now. Disruption involves not just taking over the top spot: that’s just regular competition. Disruption means breaking the old business model so badly the incumbent has nowhere to go but accept its fate if it does not adapt to the new environment.

At the end of 2019, former Twitter CEO Jack Dorsey used the shortform-posting social network to announce the creation of the Bluesky Initiative as a spinout from the company. Bluesky’s head at that point was Twitter CTO Parag Agrawal, who two years later would take over Dorsey’s role until his displacement by Musk.

As well as disruption, executives who grew rich from the growth of the web like to talk about the decentralised nature of the internet and its apparent ability to route around failure. In that context, the generation of so-called Web 2.0 operations headed by Facebook, Google, Twitter may represent an aberration – or at least that is how their would-be replacements like to see things.

The original crop of protocols developed for the internet and the worldwide web were all designed for distributed systems. Anyone with access could send emails, something we know now to contain some serious problems when it comes to getting emails you don’t want. Anyone could build a website that would respond to requests delivered using the hypertext transfer protocol (HTTP).

One of the early problems faced by the first crop of websites and other information repositories using standards such as Gopher or the file transfer protocol (FTP) was findability. Even armed with the Whole Internet User’s Guide and Catalog, you pretty much had to know where things were stored to stand a chance of finding them.

Search engines fixed one part of that problem, opening the door to Google’s later dominance of the space. If you knew roughly what you were looking for, you could probably find it. Social media delivered on the other half of the problem: delivering stuff you were not actively looking for but might be interested in. All the companies took one pre-web law of communications to heart: Metcalfe’s Law.

Though it was technology commentator George Gilder who popularised it, ethernet pioneer Robert Metcalfe formulated the law in 1980 around telecom networks, concluding that the value of a network was roughly proportional to the square of the number of users.

Examples abound. Fax machines would ultimately be swept aside by email for most purposes because email became more prevalent. Social networks occupy a space that is even more weighted towards the winner: that of the power law. Writer Chris Anderson presented the concept of the Long Tail as a key driver of Web 2.0 during its early days, and a means of ushering in a period of artistic expansion where creators could flourish no matter how small their niche. The long tail represented the massive difference in distribution between winners and everyone else in an environment characterised arithmetically by a power law. Because the businesses were driven by advertising while concepts such as micropayment foundered, it became a winner-takes-all market where all money went to those at the head of the power law. Digital sharecroppers picked at the scraps in the long and extremely thin tail.

It’s not all good news for those at the top. Building content-serving businesses on Web 2.0 can be fragile, as cryptocurrency channel Bankless discovered last May after YouTube banned the channel and removed access to the videos – and any streaming royalties that would go with user views.

Bankless founder Ryan Adams declared on Twitter: “They can deplatform you any time for any reason and they don’t have to tell you why.” Though the Google subsidiary reinstated the channel within a few hours, Adams added the episode provided one major reason “why we need Web3”.

In theory, Web3 will fix the problems of Web 2.0, though this has been the promise for a while. Ten years ago, the third generation of the web was synonymous with Tim Berners-Lee’s concept of the Semantic Web, and he maintained that vision is alive and well at the Web Summit in Lisbon late last year. For many others, Web3 rather than Web 3.0 is where the future lies.

What the promised web of the future is morphing into exactly is not entirely clear. However, a key principle behind most of the proposals for Web3 lie in the ability for end users to manage their own data, even to the kinds of information governments typically hold today. It might take a while to get there even if large swathes of the population choose to embrace it instead of staying with the services they already know.

One part of the migration to Web3 is the decision by many Twitter users to set up accounts on servers that run software from open-source project Mastodon. Twitter provided a reminder of how inconvenient migration can be if the existing service decides to dig its heels in: treating any links to Mastodon posts as pointing to ‘malware’ and banning accounts that mentioned they had moved.

Mastodon’s key difference is that it supports a federated model supported by a variety of software projects that come under the ‘fediverse’ banner and which, for the most part, use the same ActivityPub protocol to flag updates up on any part of the network. Much like blogging platform Wordpress, which now also supports ActivityPub, you can run Mastodon where you like, and if you do not want to run a Mastodon server you can sign up at someone else’s if they are taking on users. The servers exchange messages with each other so that, in principle, you can subscribe to posters on different servers in much the same way you can subscribe to them on Twitter.

However, critics of the federated model point out that there is still an element of control the end user only has if they are running their own server. Each Mastodon-based service has administrators and moderators that police activity on their nodes and they can block other servers completely as well as remove accounts they believe to be behaving badly.

Those who want to avoid the possibility of the rug being pulled out from under them by a service provider are looking at a further step away from the centralised Web 2.0 systems, one where the systems are not just using federated servers but operate on the same peer-to-peer principles as distributed file storage, accessed using BitTorrent and similar protocols, or cryptocurrencies.

It is not necessarily an either-or choice between a network of federated servers and full peer-to-peer decentralisation. “It’s more a gradient than a binary decision,” said Shermin Voshmgir, founder of Berlin’s BlockchainHub, at a seminar on Web3 organised by the European Union in the autumn.

The BitTorrent of chat is Secure Scuttlebutt (SSB), developed originally by New Zealand-based sailing-boat enthusiast Dominic Tarr, who wanted to see if it was feasible to have a chat system work well where many nodes were poorly connected, with limited access to servers. The result was something that bears a passing resemblance to blockchains like Bitcoin and Ethereum, but without the requirement for an expensive consensus-based update mechanism. A cryptographic key accompanies each update, which can come from anyone anywhere, that goes onto a steadily growing list of shared content. Each node updates itself by looking through, hopefully reasonably local copies, much like the chunks of files catalogued by BitTorrent, and grabbing those that seem relevant.

Though developed for chat, SSB has made its way into a variety of client-software programs that aim to emulate the social networks of Facebook, Instagram, TikTok and Twitter. Some of them, such as Far-caster, have opted to incorporate blockchains into their decentralised social networks alongside the core SSB update protocol.

This alignment with blockchains has two purposes. One is to provide a source of funding from users and so avoid the need to raise venture capital in the early stages, and sidestep the advertising-driven business models used for Web 2.0. The second is to provide a way to build a system where no-one must give permission to sign up or post: that permission in principle comes from the community as a whole.

In practice, Farcaster minimises the on-chain activity its network performs, largely to keep costs down. At this stage, Farcaster uses the Ethereum blockchain transactions primarily to give each new user an account ID. The other is to buy a non-fungible token (NFT), currently set at the equivalent of $10, that is used much like a domain name would be for someone constructing their own website or getting a custom email address on Gmail.

In his Lisbon speech, Berners-Lee criticised the apparent alignment between the distributed web and blockchain, and he is not alone. Jay Graber, CEO of Bluesky, argues the key to Web3 does not lie in blockchains, but the cryptography that makes systems like Ethereum work and the ability to use these primitives to make ‘self-certifying protocols’. The approach she envisages is not unlike the tokens that the likes of Farcaster add to the Ethereum blockchain. The difference lies in how people view the data stored in either.

With the broader category of self-​certifying protocols, in which Graber lists BitTorrent and the Pretty Good Privacy (PGP) encryption systems for protecting and validating emails, the proof of ownership lies in the private key used to encode messages. The idea is that, assuming you have not let the key be used by someone else, data comes from you because it is signed by you. Blockchains have a stronger claim of ownership as there is a timestamped global ledger that everyone agrees on which contains the claim. Bluesky is working on the basis that these additional properties will not be needed to make its AT Protocol work, though the details of the protocol are still being thrashed out.

Whatever form they ultimately take, proponents see the cryptography behind the Web3 systems, whether they use a blockchain or not, as a way of controlling the greatest problem of permissionless peer-to-peer systems: that acting antisocially is a lot easier when you do not need to obtain permission to do so. Without some form of control in the hands of the community, decentralised social media could simply be yet another vehicle for spam and trolling. It could so easily be like email where billions of emails surge into mail servers every day that no-one wants to see.

One factor in favour of the peer-to-peer systems is that control over what you see is up to you and not an algorithm designed to maximise engagement, often by posting links to videos and posts designed primarily for their outrage potential. But, as with email, short of only receiving messages from senders you whitelist, managing what you want to avoid seeing is a headache.

Developers such as Farcaster cofounder Dan Romero believe content protections can be added efficiently on top of a decentralised system, though details have yet to be thrashed out on most of them. One option that Farcaster is looking at is to treat accounts in much the same way as Google’s Pagerank system, or the blocklists used to help administrators block suspected spammers: user feedback would score posters. The resulting rank might take the form of a token added to the blockchain, or a collection of them. A client would then simply filter messages based on the resulting scores. Other systems might rely on communities building databases of those they trust and those they do not, akin to the lists that some have used to block unwanted followers on Twitter.

One big problem is that with email, spammers just keep creating new accounts that might have no reputation, but they do not pick up a bad reputation for a while. Online trolls follow a similar pattern: if one account is banned, it is generally a trivial process to acquire a new one. One way to combat this problem is to demand real identities, but this clashes with the desire to protect anonymity. A doctor might want to put a point across but does not want to risk their reputation on a potentially damaging public conversation. This may be where the technology of zero-knowledge proofs may provide an answer (see Beyond Blockchain feature). Assuming the vital credentials are available online, these mathematical proofs make it possible to refer to them to confirm status, age, or some other attribute without revealing details.

Were would-be spammers and trolls forced to provide evidence that they are genuinely new users and had not been banned by a service previously – a claim that could be supported by zero-knowledge proofs – their activities could be curtailed severely. They might be able to use other individuals to continue their activities, much as money launderers use ‘smurfs’, but this would likely increase their costs way beyond what they today spend on setting up fake accounts.

The European Parliament will vote on a proposal later this year on incorporating zero-knowledge proofs into the digital-identity scheme that it will mandate for member states. The Tony Blair Institute for Global Change made a similar recommendation to UK politicians in its recent report on innovation.

In principle, the EU’s systems would support an approach known as self-sovereign ID. In this scheme, an individual does not need to provide any third-party documents to demonstrate who they are or what they can do. And they do not need to rely on a third-party service. They can use their own hardware to support any claims, though those claims ultimately rely on data stored in official databases. Notionally, in this environment, your entire online existence, from social media through work to appearances in some shared metaverse, are all controlled directly by systems you control directly, even though pretty much everything you do is handled primarily by a network of cloud servers.

The self-sovereign existence has many attractions at a theoretical level. Will it all work? Facebook and Twitter have massive head-starts. Entrepreneurs like Romero are happy to fly under the radar for a while as the company builds up numbers. Like Facebook in its earliest days, to get a Farcaster account you need to be accepted by Romero himself through a direct message on Twitter. That will change at some point, but now, he claims he’s happy for the user base to grow at five per cent per week.

However, the key to the roaring success of the Web 2.0 platforms has been the way they have inserted themselves between traditional broadcasters and networks of friends through activities of the influencers with high follower counts. Though they are constantly at risk of being booted off the platforms, they were able to build up large audiences without bearing the cost of server and network bandwidth, choosing to take a small royalty instead once they have reached a threshold.

If influencers can build networks of equivalent size on the decentralised systems, that may bring Web 2.0 to an early finish. However, those platforms can do a lot to prevent a kind of disruption they are unlikely to favour. If that is the case, Web3 may remain a niche interest for the average user if they see entertainment as a greater priority than social network control. As it was in the days of Web 1.0, it’s not hard to wind up shuffled off to an unseen corner of the internet; you just have to be someone who does not get promoted by a search engine or social-network provider. Decentralisation may do little to change the power-law dynamics of social networks.