Unbundling Twitter
A decentralized communication protocol could change for good the rules of the social media game.
Hello! Welcome back to Cloud Vertigo, a weekly newsletter about the wildest ideas that keep me up at night. It could be highly contagious. You have been warned.
The world is changing at unprecedented scale and speed. Today’s disruptors are already being disrupted. ChatGPT hit a million users in less than 5 days. Even innovation leaders such as Google are struggling to catch-up. Old paradigms are shifting. New ways of thinking are redesigning the information economy. Today’s issue explores how a simple innovation could redefine social media.
Social media platforms are hardly news. They have been around for more than fifteen years now. Their dominance and impact in the old media economy is undisputed: from crappy start-ups they became technology behemots employing (and now laying off) tens of thousands of people.
In its book “Postjournalism and the death of newspapers” Andrey Mir offers a definitive perspective about the changes the media economy underwent (in Italian, you can read Franco Debenedetti’s excellent review). I will pick up from the metaphor that wrapped up their analysis, that comes originally from McLuhan, who got it from a short-story by Poe. Caught in the middle of the Maelstrom vortex, the sailor survival depends on his curiosity to quickly understand the forces at play in the storm.
The huge vortices of energy created by our media present us with similar possibilities of evasion or consequences of destruction. By studying the patterns of the effects of this huge vortex of energy in which we are involved, it may be possible to program a strategy of evasion and survival (McLuhan).
So keep calm, sit back, relax, and enjoy the storm.
David
Could a Decentralized Protocol Change the Rules of the Social Media Game?
Good ideas are often strikingly simple. They come with a certain feeling of “Why didn’t we think about this before?”. They are like someone you meet for the first time and you feel immediately familiar with.
Nostr - Notes and Other Stuff Transmitted by Relays - is not another app (GitHub). It is the simplest yet extensible open protocol that allows building truly censorship resistant and decentralized social media experiences. Reportedly, the project was recently funded with a 14 BTC grant (around $ 0.35M) by Twitter’s founder Jack Dorsey. Privacy activist Edward Snowden followed suit.
According to its proponents it’s the first protocol with a chance of working. It’s key strenght is that is a minimal protocol. A concise (unlike this article!) common framework or language that makes applications interoperable. In particular, it bridges between two kinds of apps: “clients” (that curate the user’s experience of publishing and reading content) and “relays” apps (that behind the curtains handle the storage and distribution of content). With Nostr Implementation Possibilities, in short NIPs, more functionalities are countinously extended. Client and relays choose freely which NIPs extensions they support.
The end goal? Everyone can build its own social media platform as they like it and by using the same protocol, their data streams become interoperable. Sounds too good to be true? Or a nightmare? Imagine having your Instagram posts on Twitter... It’s the greater unbundling hitting the great unbundlers. Welcome to 2023.
What is fundamentally a social media?
Paradigm shifts occur when a deeper understanding of the phenomena offers a fundamental reframing of the problem. The key thesis I am going to argue today is that the media publishers-subscriber problem which has been solved in a bundled fashion by social media platforms, now can be solved in a decentralized manner by protocol(s). Nostr protocol is a dominant alternative because is both better and cheaper. The question that’s been in every major software engineer’s system design interview has now a new answer, it´s suprising nobody thought of it before.
So, what is fundamentally a social media network? What are the key forces at play in the vortex? We may identify social networks with a pub-sub communication architecture. There are two non-disjoint groups of agents: publishers who produce and share content in exchange for some social affirmation, and subscribers who consume content from a set of publishers and choose to reward them. The two groups are likely overlapping more and more: today most of us are both content publishers and subscribers to others. In Mir’s argument people pay newspapers to validate their worldviews.
Communication is chiefly asynchronous (sorry Clubhouse and Twitch, live communication is for another episode) and mediated, with intermediaries facilitating the exchange. Intermediaries store and distribute content, index the identities of publishers to make them discoverable, arrange the order in which content is served to subscribers, manage rewards. Rewards can be monetary, or not (as “likes”). Platforms today mostly monetize subscribers’ attention to publishers by injecting some paid content (“ads”) from certain publishers to specific subscribers.
Many flavours of social platforms
Social media platforms have played with all the possible implementations of this model. This pub-sub relationship was initially symmetrical, as is the case with Facebook friends. Then it became asymmetrical, as with Twitter and Instagram, where your followers are not necessarily the people you follow. Some platforms, such as LinkedIn, maintained both symmetrical (¨connection¨) and asymmetrical (¨follow¨) options.
Content can be publicly accessible, as with Twitter, where everything you write is accessible to everyone, or permissioned, as with Instagram private accounts where publishers control who can subscribe to their content. Alternatively, content can be fully private, as in direct messages, which are a special kind of symmetric stream that only the intended recipients can read. Some platforms, like Onlyfans, offer publishers options to directly monetize their audience, by charging a subscription fee.
In social networks, agents are likely human, but more often than not, identities are hidden behing a pseudonym. Sometimes, the identity of the agent is verified by the platform. Often, the actions of the agent are the result of some arbitrarily smart program, a “bot”. A platform usually wants to be in control of the autonomous agents publishing/subscribing in it, so by their terms of service they don´t allow any other “bot” except from themselves: all the ads we get are published by the platform’s publishing “bots”. A key platform’s job is to separate real users from bots.
Network value of being a platform
To Zuckerberg’s credit, Facebook was among the first that really understood the value of being a network platform. Facebook was born by letting its users connect and publish and subscribe to content, for free. It thrived by allowing developers to build cheap and thrilling social games on top of it, that made the experience even more engaging. It did not have to do the costly publisher’s job of creating or selecting valuable content. It could just focus on creating the best possible experience for subscribers and publishers. Instagram was built similarly with tools (filters, adjustments) for amateur photographers to make and share their best photographic content and distribute it. Fast user acquistion and sticky retention was worth more than charging users a fee. Network effects followed as a critical mass was on boarded, business defensibility ensued with user and content lock-in. Reaping the benefits of “network orchestration” has been in every venture capitalist playbook ever since.
As an aside, a great collection of internal memos by former Facebook exec Sriram K showcases some example of this ¨platform mentality”. It also explains why Google+ was a failure. Google did not have the platform mentality. It built everything the cool Google way, well integrated internally but not externally. Mark Zuckerberg negotiating with Kevin Systrom over the Instagram acquisition and Andrew Bosworth on the strategy for FB ads are some of my personal favourites.
What do we mean by unbundling?
It is hard to argue with the fact that creative destruction is a part of the tech industry’s DNA. Unbundling has been called the great disruptor force. Ex a16z partner Benedict Evans’ 2021 presentation The Great Unbundling is great showcase of examples from e-commerce to media.
In layman terms, unbundling means repackaging an offered service by decoupling its parts. Social media unbundled the editorial from ads which traditional newspapers tied together and distributed to consumers.
Image credits: The Great Unbundling, Stratchery, 2017.
If you think about it, social media offer is itself made of a big bundle of things.
To subscribers social media give access and discovery of publishers, they offer a native (iOS, Android) and web interface to consume content on computer and mobile; and a recommendation engine that suggests always new and relevant content, designed to be as good and effective as it gets (read, addictive).
To publishers they sell storage of their content, distribution according to different tiers of service (including paid promotions), account verification, engagement reporting.
To both subscribers and publisher, they sell identity management. We are “@such-and-such” handle on Twitter and on Instagram.
Bundling is not bad per se. With homogeneous goods it can benefit buyers and sellers alike, but it can also hurt consumers if it is used by incumbents to exploit their broader catalog to “deter entry” by new competitors. Think Microsoft in the 90s.
The house mental model for unbundling
Unbundling this offer is to ask: What If these functions could be offered standalone, on many separate markets by many different players, while preserving interoperability?
Imagine your internet presence is like a home. You could live in a fully-furnished fully-managed property where every detail of the experience will is curated by your landlord, like in a resort; or you could build your own thing, furnish it and manage it accord to your own needs. On average you are likely to have a combination of both: you may own your apartment, built by some developer on his land, design some tailored furniture for it and buy some more decoration off the shelf from various resellers, while leasing services for the regular upkeep of the property, such as gardening and cleaning.
Unbundling the keys to identity
The biggest lesson crypto brought to the internet is that identity management does not need a trusted third-party, which is to say the concierge service holding the keys to your door is optional. In stronger wording, you don´t need to rent the address of your online presence, you can own it.
While username and passwords identities are granted by some centrally-managed database, you can have key pair digital identities that are verifiable in a trustless and permisionless fashion. Instead of a username you have a public key to share with your contacts, instead of a password you have a secret private key that you need to keep safe.
Key pairs are cheap and easy to generate, you can do it locally on your computer. Nostr leverages the crypto intuition of the keypair identity without having the burdain of a full blockchain. It designs a minimal protocol for broadcasting messages with a signature that attest they come from the author. Deadly simple. Here’s a message:
{
"id": "b9fead6eef87d8400cbc1a5621600b36...",
"kind": 1,
"pubkey": "82341f882b6eabcd2ba7f1ef90aad...",
"created_at": 1676161639,
"content": "this is going to work",
"tags": []
"sig": "76d19889a803236165a29...",
}Different kinds of messages and tags allow for Encrypted Direct Message (NIP-04), Handling Mentions (NIP-08), Reactions (NIP-25) and so on.
One big caveat, is that there is no locksmith that can break your door if you lock yourself out of your apartment. In a self-managed identity, there is no password reset mechanism to fall back to. If you loose the key of your Bitcoins, they are gone forever. You can make a complaint, sue the company, get law-enforcement involved, nothing will bring them back. I know it sounds scary. Independence comes at a price, therefore most times it makes sense to outsource it, but having the option not-to is a key web3 design feature.
Unbundling identity verification
Once you take identity management out of the bundle, why stop there. Identity verification is the next plausible candidate. Verification can mean increasingly more complex things: from you are a real person to you are who you say you are, from you are a public figure, to you are a big deal. With companies offering fake Wikipedia articles and Spotify albums to cheaply trick Instagram to make you into a ¨someone¨, identity verification is big business in the digital age. The much discussed new Twitter product decision to grant the blue check at a 9$ monthly fee is a case in point. All in all, the ¨blue checks¨ on Twitter or Instagram mean different things to different people, so they could naturally benefit from broader market offerings.
Imagine various tiers of service fit different purposes: from certifying that the account is not managed by a bot, to a publicly certified digital identity, to - say - a vanity gold check at 100$ per day, just to show off that you play in Premier League. Maybe there could be some curated invite-only authority vouching. A verification service may not only have different levels of prestige, it’s also crucially representing something ontologically subjective. Authority depends on other authority who’s granting it. Any single centrally-managed identity verification service is ultimately just grotesquely medieval, weak and misleading. The world we live in today is plural: Internet better reflect that.
Nostr user base exponential growth mirrors that. According to some stats its Weekly Active Users base been growing 5x in the last month alone from 100k to 500+k, according to Nostr Directory there are at least 10k verified Twitter users.
Protocol network value proposition
While identity is a common problem for both publishers and subscribers, other pain points may diverge. From publishers platforms naturally compete for content; from subscribers they compete for attention. A protocol-based offering would need to have compelling services about both.
For subscribers, the value propostions that future social media will unbundle concern mainly the user-experience and the interface’s look and feel, as well as the content curation policies. For instance, different quality levels of spam and sensitive-content filtering mechanisms could be built in. Furthermore, you might have many market possibilities of recommendation algorithms with different content-to-ads ratios, free or paid-for. You may empower users with various unique one-to-one messaging capabilities. Finally, discoverability of content (the “Explore” function), browsing / and searching are key: you can build clients with a more or less performant indexing and popular content caching, quality recommendations or a fully sponsored experience of “viral” content.
On the other hand, unique value propositions for publishers would need to solve pain points around content ownership, storage, availability and distribution. Different levels of service agreements could play with the quantity of data to store (text, image, videos), as well as with the acceptable content quality. Most news institutions big enough to bear the costs could host their own content, freeing themselves up from platforms. User content’s relays could be free, charge per-content or on a subscription basis. Yet, it’s not just about the greater pricing elasticity, different levels and definitions of acceptable content by various relays empower Nostr protocol with ultimate and real censorship-resistance as “there will always be some server accepting your messages in exchange for a fee”. Regulators have been arguing to hold platforms accountable for the content they host, but encountered much (fairly justifiable) push-back. No chance of
Finally, there’s the monetization question. What happens to likes? There already are some NIPs experimenting around zaps, micro-payments using the Bitcoin Lightning network. You may ask, what about ads? A complex topic best left for another time.
I think by now you got the idea of what the unbundled experience looks like. In its simplest form is more desirable than the status quo, since it allows having many parallel separate marketplaces. Each will bring efficient pricing to how the individual needs are best addressed. The end result is that there could be many various competing subscriber-publishers equilibria, from “free & ads-based” to “fully paid-for”, in an integrated continuum.
News itself is a very paradoxical commodity. It always ‘needs’ to be read; it is always in some kind of demand from below. But there is always someone from above who wants to pay for certain news to be delivered to the public. And those from above—those in power or advertisers—want to pay to deliver the right news much more than those from below, who are willing and able to pay to receive the news. (Mir, p. 55)
Some of these functions will be vertically integrated or re-bundled. Some of these markets will belong to the disruptors who deeply understand them, some to incumbents. For example, operating systems incumbents are best positioned to offer client applications. A sign in this direction is Apple News that shows strategic awareness to be the place where its users read news content. On the other hand, cloud providers are well positioned to have cost-effective offerings about storage and bandwith. Specialists players could offer niche private networks, and there will always be edgy wild-west providers offering the kind of freedom Bitcoin maximalists expect, playing hide and seek with regulators wanting to ban them.
Now, Elon’s Twitter has put a price tag to the premium social network experience, the 9$ a month every internet business is after, even though justification has been a challenge for their product managers. In this protocol-based economy, you could pay 1$ a month to StorageCo, 2$ to VerificationCo to widely circulate your words, 3$ a month to RecommendationCo to discover and read content with minimal ads and still have some budget left for a reward pool to your favourite publishers. All of this without being tied long-term to any particular platform?
It’s your own plug-and-play, à la carte, social media experience.
Can protocols be cheaper that platforms?
This question still need deeper analysis and thought. Here is a rough sketch of an answer.
You may wonder at first about the technical feasibility of such an infrastructure. No internal regulator setting the standards may sound like a nightmare. A lot of developer effort could turn worthless overnight. Let me introduce extensible protocols. Protocols are for builders. The internet is built on many layers of protocols from Ethernet to the top level, the application layer, which is HTTP. Protocols are a language of exchange, a standard format that enable cooperation. History proves that a good protocol such as the web can set paradigm shifts in motion. Good protocols, like regulation, are minimal. Since every extra detail is possibly contentious, designers need to keep it short.
If a protocol is open you may ask what are the incentives then? Well, sometimes it’s not about what is there, but what isn’t. Developers are flocking to build on Nostr because of the freedom it empowers. Would you build in an authoritarian country controlled by a single owner or in an open free space with minimal intervention? Remember: platform thrives by attracting publishers. Protocols are meta-platforms: they win over any given platform by attracting platform-builders.
Most web3 blockchain projects have been all about the early incentives. The ¨get first here, get rich¨ by holding some otherwise worthless token could make some sense to bootstrap infrastrcture networks. The value of network externalities after all comes in later. But this very risky equity-like bootstrap also favours the (often uncharitable) association to Ponzi schemes. We did not need another privacy first social network, or another blockchain token to trade, what was needed was to break free from the platform model and build a protocol of exchange.
Platforms, like other intermediaries, derive their rewards from the struggle of very illiquid markets, with sparse buyers and sellers. They make the market more liquid, which in turn decrease the risk of trading, hence bring more people to the market; by doing so they tighten their spreads but unlock scale with higher volumes. They are in some sense on a mission to make themselves redundant or achieve monopolistic dominance over a market if they lock-in sufficiently well their counterparties. Their defensibility is a function of how hard the user-adoption is for their competitors. Interoperability makes the marginal value of the walled gardens decrease exponentially. Twitter is already facing the pressure…
It may be great for a while to feel like a prince and live in a resort, yet the lack of choice inevitably comes at a steep, eventually unsustainable, premium. While some may argue that people would not care about the unit-economics since they already have a free product, the assumption here is that without a perverse alliance with regulators platforms cannot compete for ever against the whole market. Once an open protocol exist, no single entity can hold a network monopoly for a long enough timeline. Adapt, or be gone. This is especially true given local properties of network effects: for a given user being in-the-network needs to be more valuable to being out-of the-network. It’s hard to imagine a single entity could own a universal network broader than the global interoperable network. On Nostr apps there will be on offer precisely the level of control, cost and personalisation that are tailored to one’s unique needs, capabilities and values.
On protocols, being first, being fast, being on-point has value. Opportunities are many. The web inventor Tim Bernard Lee did not become a billionaire, yet he achieved financial independence and the impact of his invention will outlast him for many generations. What do you think? What am I missing in this analysis? Disagree? Write me a note!
My Nostr Public Key 🗝️ npub1dval4534hhxwn06yse63k98ry7c76mf4usqqspldaun2rx28etgsefy8f8
📚 If you want to get started on Nostr, check out these resources!
If you read all the way to the end, you may enjoy the next Cloud Vertigo chapters. If you’d like to schedule a conversation on these topics, you should drop me a line. Take care, bye!
Loved this, it feels like edging to a frontier of technology yet to emerge and change our lives