On April this year, we wrote about decentralised alternatives to legacy video platforms. It seems like a decade has gone by, and the market mapping deserved an upgrade. This text describes a few major initiatives trying to disintermediate infrastructure for video distribution, outlines current monetisation models allowed, and points to common denominators.
This is the second piece of a 3-part series on the future of video. Check out the previous one here (a history of P2P content distribution) and subscribe to our newsletter to receive the next one straight into your inbox 📬.
1. Online video in a nutshell
The process of streaming on-demand video can be divided roughly in three major phases. One has to store content; multiplex it (adapt the data to make it suitable for reproduction under varied conditions, which can happen at different points, even before storage); and distribute it, or make sure the video gets from where it is to where it’s going to be played. Digital advertising providers, metadata tracking tools, licensing software and other niche-specific services can be plugged in at basically any point of the value chain, making room for almost infinite verticalisation. A tentative scheme is drawn below.
“Decentralised” video has had different meaning throughout the last couple decades - one can think of decentralising any particular vertical or layer of the chart above, for instance. On some level, the advent of TVs decentralised audiovisual storytelling by taking it out of projection rooms and into family homes. The internet ended up doing something similar, on a much larger scale. Torrents and distributed networks take the concept of decentralisation further, and redefine the way through which resources are allocated and consumed within a network. With peer-to-peer, one can think of applications for sharing videos that are basically uncensorable and have no central points of failure. But it was crypto that allowed us to conceive networks where peer-to-peer goes along with economic scalability.
Even though some of the first ideas (and code) on getting smart contractas to work with video date back to March 2015 (
), it was in 2017 that development in the field has exploded to the point where it’s hard to follow everything going on.At the tip of the iceberg, a handful of ICOs for video-related projects took place. Down below the mass of ice, base-layer technologies matured and stable releases by teams behind some of the most popular DSNs have allowed developers to actually start shipping partially decentralised platforms upon peer-to-peer networks. On a more fundamental level, access to global crypto-capital has led to the emergence of multiple teams working on different layers of the stack, trying to solve some universal and other peculiar problems, forming what’s one of the most exciting sectors of blockchain innovation today.
2. Market Landscape (tech mapping)
The intention of this exercise is to inform aspiring developers seeking for exciting projects to work on, and platforms/businesses willing to integrate/support any of the mentioned technologies. Investors, this is not advice. If you find any projects missing, please drop a line on the comments.
It’s useful to begin by defining the 3 gravity centres in the map:
DSNs: decentralised storage networks are peer-to-peer communication-layer protocols for storing data without central servers, leveraging the spare resources of interconnected nodes. Picture torrents on steroids, with incentives (💰) being assigned to agents/machines according to their contributions in basically storage space and bandwidth.
Bitcoin: the king of cryptocurrencies, bitcoin spawned a lot of interest around the idea of micro payments attached to videos, but the issue of scalability impeded any effort from getting close to practical for some years.
and Lightning are here to change this.Ethereum: the go-to platform for smart contracts and easily deployable tokens, Ethereum has hosted a number of ICOs for video-related projects, while production-level micro transactions solutions over the network are still pretty much incipient.
Initiatives can be plotted “in the orbits” of these base-layer components, for clarification, although this presents some obvious challenges. The tech stack for decentralised applications is not crystallised yet, and classifications are emergent.
We’ll follow here a modified version of
’s proposal on dividing the economy for any given decentralised vertical “into three parts: base protocols, application protocols, and consumers/users”, pretty much equivalently to as consumer markets can be divided into “suppliers, distributors and consumers/users”. First variation: let’s isolate true base-layer protocols from their native subcomponents (built on either the same token or the same network topology), calling these “2nd layer infrastructure” (think Raiden, for Ethereum; or Lightning, for Bitcoin). Second variation: let’s fit all end-user facing interfaces and services that support or enhance them in a broader category named “tools, services, dApps” — all that touches final users or directly reshapes this interaction.Hence, we have “DSNs” (some with surrounding applications already); Bitcoin or Ethereum able to complement as a logic/payment layer; “2nd layer infrastructure”; “application protocols”; then “tools, services, dApps”. Below are one-sentence descriptions of each mapped project:
2nd layer infrastructure
(♢) Swarm: a native layer for distributed storage, along Whisper (for communication), that’s part of the original Ethereum web3 stack.
(♢) Livepeer: infrastructure for distributed transcoding of live video, over Swarm, and counting (currently) on off-chain computation w/ TrueBit for job verification.
(Ƀ) HTLC-Dash: hashed timelock contracts for triggering actions upon micropayments, such as the delivery of a video chunk, under development by Lightning Labs.
Application protocols
: a protocol and standard media player with embedded wallet for tax-less video sharing over IPFS, in the browser.: live streaming and micropayment infrastructure for the adult entertainment market.: a token and platform of dApps for mobile live streaming interactive applications.Services / Tools / dApps
: Brave is an open source, “privacy-focused browser that blocks malvertisements […] and contains a ledger system that anonymously captures user attention to accurately reward publishers”. Basic Attention Token is the tokenised “unit of exchange between publishers, advertisers and users, derived from — or denominated by — user attention”.: a blockchain entertainment studio, focusing on tools for artists to tokenise and distribute their work.Tokit: their first dApplication, basically a crowdfunding platform where artists can issue tokens with embedded revenue share for their fans.
AdEx: a decentralised advertising exchange.
Jaak: a framework for decentralised content licensing and metadata handling.
AdChain: a decentralised crypto-backed whitelist for domains, that aims to tackle rampant fraud in the digital advertising industry.
On Ƀitcoin
PopChest: a video platform with traditional streaming that avoids ads and allows for micropayments over the Bitcoin network between creators & audience.
BitCache: a cloud-based file sharing service for storing data on either central or decentralised platforms and directly monetising content, including video.
On Bitcoin Cash
Yours: a “Medium with a paywall”, where users earn cash for good content, and can embed things in their pieces, including video (built on Bitcoin Cash).
Webtorrent!
WebTorrent is a popular streaming client for torrents, written in JavaScript, using WebRTC for peer-to-peer transport whenever possible. It’s not technically a DSN (as defined here), but can be used for data distribution in combination with extra incentive layers on top. This is exactly what
does, for instance, on its community-targeted video platform. Just as well, WebTorrent can be used in its “pure form”, relying, to some extent, on the altruism of nodes. is an example of such approach.DSNs (decentralised storage networks)
IPFS (by
) a peer-to-peer hypermedia protocol based on content-addressing that aims to replace HTTP.Filecoin: it’s native currency and incentive layer, under development.: Florincoin is scrypt minable bitcoin variant with quick tx processing, that powers the library of Alexandria, aimed at storing and sharing art, history and culture.: a community-run digital marketplace for media and blockchain that runs in a desktop client.: a blockchain with its own attention rewarding mechanisms.
Wildspark: a browser plugin for moving AMPs (a token) between creators and audiences of Youtube videos.
Decent: a blockchain with optimised design for decentralised publishing and focus on subscriptions.
Maidsafe: a decentralised data and communications network with its own browser, desktop mail and web hosting manager.
Sia: private cloud storage on a blockchain whose transactions are simplified in comparison to that of Bitcoin.
: end-to-end encrypted, privacy-focused cloud storage where anyone can “farm” with idle disk space.Steem: a blockchain built upon BitShares’ design, with its own “proof-of-brain” community rewards, that has paid out millions on the
social network.DTube: a decentralised Youtube (even the same colours) on Steem.
Viewly: a web and mobile platform with no ads and direct monetisation between fans and creators, on Steem and IPFS.
EOS: a blockchain designed to achieve high throughput rates and suit use cases that involve media distribution and micropayments, with its own hosting solution under development, in which block producers are responsible for file storage, requesters have to bond their tokens, and scheduled inflation pays part of the cost of the service.
So, the question is: how do all these interact?
Or, put differently: should there be winners and losers among these technologies? Well, it’s early to say. Most are in development, experimental, or, at most, beta phase.
notes that, “unlike networking protocols of the 1970s which were ‘technically so distinct that no mutual communication was allowed’, every set of decentralised protocols built on the same platform, such as Ethereum, are baseline interoperable”. What we see forming is not a single pyramidal stack, where the choice of one piece of tech inevitably “disables” another. Rather, what we have are protocols silos with more or less interoperable layers. In practice, this means seemingly disparate technologies can be combined to extend crypto-incentivised functionalities of a decentralised application (e.g. you will eventually be able to login via to browse on , which may even embed ’s player).
As
, of , puts it, dApps are not vertically integrated monoliths, but rather composed of unbundled layers, each a specialised network with distinct incentives and tokens. A traditional dApp may require a protocol for consensus, a file storage layer, computation, oracles… and different tokens to fuel each. But it makes little sense for users to hold a variety of cryptoassets, some of which they have no vested interest in, to access a specific service. Then comes what Will calls token abstraction. Instead, users can hold on to tokenised fiat, and, at key interactions points, the client passes this fiat along a thick stack of orders into the dApp contract the user’s interacting with, this is picked and executed by relayers, the contract receive back the tokens it needs, and pays for the bundle of it services it wants to, without the user ever noticing the back scene juggling.
Despite the concrete efforts to preserve the compatibility of (at least) ERC20 tokens and also save users from UX nightmare, it’s important to remember that speculative secondary markets for cryptocurrencies cause clear disincentives for interoperability, since teams worth their salt usually want to build their own protocol, and capture more value for developing what whey envision.
has explored the consequences of this behaviour to its extreme consequences. First, he points out that the completion rate and raised amount of ICOs is falling drastically, and should continue to do so for a while (worth noting: this is becoming less and less the principal means of raising capital in the space). Second, Eric proposes that, for an asset that hasn’t yet a large enough user base, poor funding and liquidity can induce a negative feedback loop that transforms a cryptocurrency into a “zombie token”. On a less nihilist perspective, examines some less disastrous potential outcomes, and states scenarios in which crypto M&As may soon occur.3. Monetisation
“It may seem weird, but I’ve been eagerly awaiting the day when I see ads in my viral video”
— Nick Gonzalez, Tech Chrunch writer, 2007.
The quote seems totally out of context 10 years later, with adblockers’ already installed on more than half a billion devices worldwide. Youtube has consolidated a monetisation model inherited from the analog world. Content is distributed by infrastructure providers who intermediate access to the audience, and take revenue cuts to cover operational costs, preserving network effects by isolating user data in privately monetised silos.
Throughout the last years, it’s became evident infrastructure can be decentralised and scaled in ways that are economically self-sustainable. But the open source and the commitment to public data that crypto brings along goes against values in the heart of legacy business models: how can one capitalise on a product that’s free to use and to copy? How can one capitalise on data that’s out in the open for anyone to peek at?
The paradigm shift begins by understanding that it’s not one who’s capitalising on anything anymore, but rather many. To illustrate with an example, let’s take recommendations, something that’s at the heart of modern video platforms and is constantly being optimised.
But optimised for whom? Ad-dependent business models might shape their engines to present users with promotional content highly likely to be watched; platforms seeking engagement will maximise exposure to addictive sugary videos. Whatever the drive is, incentives are generally skewed in favour of businesses, and in disfavour of end-users.
Now imagine a different scenario, in which content is distributed in a p2p fashion and traffic data is not privately held. Instead, user behaviour is tracked client-side, properly anonymised, and then broadcast to peers, being eventually registered in a public ledger where data will live forever, untampered and available for anyone to explore or build predictive models upon. Now why would you build a predictive model upon pieces of “anonymous navigation history”? Because if data is public, so should recommendations be. In this hypothetical paradigm, we could have a market of decentralised engines analysing traffic patterns, applying their own prediction strategies and updating an index that measures quality of outcomes, in a way that any user can choose which recommender to subscribe to, and poll suggestions from. The issue of filter bubbles is tackled at the same time that a new revenue flow is opened up, by replacing the work of monolithic recommendation engines by that of competing algorithms in an open market.
In this case, there’s still a handful of practical constraints, but, a decade ago, the model could hardly be conceived (in fact, a decade ago we were discussing about ads on amateur videos, as you can see below). Along the same lines, the longtime debate comparing “advertising-supported” vs. “subscription-based” vs. “pay-per-view” monetisation models has became anachronistic in an age of customisable smart contracts. Discussion has to be broadened to encompass a wide range of alternative methods being tried out.
For example,
, co-founder @ Yours.org, has stated recently that his team is currently evaluating ~15 varied monetisation models for digital content.Below is a list of 10, to get your ideas warmed up (all are implemented or intended to be implemented by the projects listed above):
- 1. Advertising supported content: the traditional ads-based model. Think 5s pre-rolls before clips.
- 2. Pay-per-view (or micropayments): as in iTunes, users pay for the right to watch something. This can be adapted into a rental model, as in Blockbuster, where users pay to have this right temporarily, or distilled down to a pay-per-chunk model, where users pay for exactly the amount of content they consume.
- 3. Time-based subscription (SVOD): users pay for a month, say, for access to a pre-defined catalogue of content.
- 4. Non-determined subscription (ndSVOD): users commit an amount periodically, as in SVOD, but not in direct relation to a pre-defined catalogue - instead, like Brave does (or Medium, with its claps), this amount is distributed according to some measurement of user engagement or attention expenditure.
- 5. Private channels: sell a stream of private session, or special ephemeral catalogue of content. In looser frameworks or slightly different contexts, it can also be called Patronage or recurrent financing (where you pay a subscription as if it was a freemium).
- 6. Attention measurement: this can be attached or not to non-determined subscription models. The basic idea is to track audience attention non-intrusively, and to distribute some form of compensation to the agents who attract or influence the allocation of such attention.
- 7. Pay-to-comment (per-per-interaction): users pay to comment. It sounds non-inclusive, but if its under the control of the author, acts a filter for how much trolling he’s willing to tolerate vs how much conversation he wants to promote. And a new conversational market opens up if you can tip comments, upvote them, and so on. Comments can be generalized into interactions, and a whole range of specific micropayments can be thought of: imagine you’re playing mime over a one-to-many live streaming channel, and you want random people to bet on the game with tokens; or you’re a teacher running a video-based decentralised Math Olympiads, and willing to take in test responses from any student in the world who stakes an entrance fee, then competes for a decentralised reward).
- 8. Affiliate model (or investment): explicitly stake tokens or reputation in a piece of content/creator (or simply bring more audience to it) in order to earn a share of his future revenues.
- 9. Non-invasive ads (see to earn): a subtle subversion on 1., presumes (A) action by the user to display an ad and (B) the trigger of a small payment to him in exchange for his attention (e.g. “want to earn 5 ETH? Click here and see a suggestion of movie to watch on Friday night”).
- 10. Token curated registries (TCRs): an emerging cryptonomic primitive, a TCR is basically a list with “an intrinsic token to assign curation rights proportional to the relative token weight of entities holding the token”. In the context of video, if content consumers desire high-quality curation, and content producers like to be included in well curated lists, “a market can exist in which the incentives of rational, self-interested token holders are aligned towards curating a list of high quality”.
The underlying dynamics and technical requirements of these vary greatly. In general, it can be stated that each model poses a tradeoff between revenue potential and consumer friendliness. Besides, they can be more or less practical to implement. We can treat these vectors as axes of a cubic matrix, and plot in it models 1. to 10. according to their positions (subjectively assigned) along each spectrum. This exercise yields a visual representation of the distance between alternatives, highlighting benefits and drawbacks, but it has a non-objective nature. The idea is to provide a mental framework for evaluating new postulates.
4. What’s next?
Next is seeing all this come true. Some folks on reddit are already talking about “the dAppening”, a movement of consolidation expected to happen along 2018 with the main net releases of a batch of long-awaited dApps.
However, development still has to go a long road, before we can have fully-decentralised video sharing applications operating on global scale. Some key research fronts are pointed below:
- State channels: micropayments on the blockchain are prohibitively expensive. There’s other sorts of operations, along all layers of the stack, which could instead scale off-chain over safe and cheap state channels. Vynos, a web wallet for interacting through payment channels, is an interesting piece of tech to look at. On a more theoretical side, ’s scheme for probabilistic payments, and multiparty state channels(
1-to-n), as proposed by Velcron’s team, have been emerging as widely regarded solutions. - Anonymization: the public nature of blockchains makes privacy a sensitive issue. How can media buyers (be they advertisers) know who they are targeting without everyone else in the world knowing about user’s private data? Even if we don’t want ads, how can we build content engines aware of the users they’re dealing with, without letting them overexposed? A useful reference here is Brave’s use of Anonize-ZKP to convey ad attributions without leaking private information.
- Cryptoeconomic experimentation is in its infancy. Some sandboxed approaches to rewarding curation activities have been tested and there’s a handful under development. Another heavily discussed issue regards the ability of networks to provide sufficing incentives for audience itself to moderate abusive or inappropriate content. Self-tokenization is also becoming a big deal. If the internet brought self-publishing to a global scale, will web3 make self-tokenisation the next big trend in digital media distribution?
… and these questions lay the ground for the next and last article in this series. Subscribe to our newsletter and receive it straight into your inbox 📬.
Thanks to
, , , and for talks and ideas that contributed to this piece.
