Background

Why did you make this and why is it so complex? What's the big idea?

What’s the motivation for making this?

mova follows on from the arrival in the last few years of decentralized social media tech like Activity Pub, Matrix and Secure Scuttlebut – and micropayment and streaming payment systems like Web Monetization and Interledger. Together these support a decentralized media ecosystem where the discovery and production/hosting layers for pay-per-view, subscription and donation-based video funding are decoupled. This has long been discussed in the Open Video movement, but it’s only now that there’s payment technologies that can support it.

This decoupling is much like how offline film distribution works, with a separate producer, distributor and exhibitor/TV channel/video store. It’s also similar to how most of the web works, separating server, search enginge/social network/email server and browser. But given the complexities around video - not only for copyright but legal and safeguarding issues, the web today has centralised both non advertising video hosting and discovery around a few services.

How does mova fit in?

Just as the decoupled web uses the Domain Name System (DNS) to match a URL with the correct server – and lets anyone link to that, a decentralised video ecosystem needs some kind of a metadata registry. There are currently

mova sits somewhere in between, in that it uses a unique identifier (like the ISAN or EIDR code) but that this is generated from the media itself (like CAI). Because mova uses as its identifer a new open source media fingerprinting protocol, the International Standard Content Code or ISCC - anyone with a copy of the same video will generate the same ISCC identifier. Most signficiantly, similar video files (ie those which differ only by size, colour space, resolution or file format) will generate simiilar ISCC codes and be able to match up. So unlike CAI if someone transcodes a video file (ie from H264 to WebM), the link with the metadata needn’t be lost. This is useful online because if you upload the same photo to three different websites, they will transcode/compress it differently so normal hashing (like that used by NFTs) would suggest there is a different file.

So the ISCC is like a DNS nameserver?

On its own the ISCC is just a 52 character long string (or more accurately, four 13-character strings describing four different attributes of a file) – and any text, PDF, image or video file can produce one. But because all media files can produce an ISCC, and recognisie ‘nearest neighbors’, ie similar media files - the ISCC as an identifier can be used in a decentralised way to provide further information about that video. This could be copyright & license info, credits, subtitle files, legal notices, or a list of academic citations, etc.

We should also not be naive that repressive regimes and giant corporate legal departments can use fingerprints of media to create watchlists of activist or unfavourable content to block. The creators of mova strongly oppose such use.

So Whois in control of video DNS then?

Very funny (eh?).

Our starting point is that no single company or country should or can be in charge of it. However we anticipate a number of companies and blockchain projects may try to be, leading to the emergence of two types of registries:

  • centralised media databases
  • large open decentralised databases

So what kind of registry is mova?

In this too, mova tries to take a middle ‘Goldilocks;’ path. We use a private decentralised database shared between everyone who runs a copy of the mova app: it’s neither sitting on one company’s computer or left to run wild across the entire Internet.

Sounds a lot like the Blockchain?

No - what’s the same as the Blockchain is mova uses a ‘distributed hash table’ (DHT) or ‘public ledger’ – which is a shared public database.

But unlike the Blockchain, this database is only writable to by the mova application and its protocols. It’s a private database, which lets all of the app’s users write to it. But you could also say it’s a public database, closed to anyone not using the mova app.

What’s this public ledger called?

Holochain.

I knew you were trying to sell me Blockchain!

It’s a confusing name, but Holochain is NOT a Blockchain. A Blockchain like Bitcoin or Ethereum is writable to by every client and app using that system - it’s one giant multi-terabyte database of all the diffeerent applications and services and transaction using that protocol. This can make transactions slow, because nodes have to process terabytes of data, and it normally has a big environmental footprint.

Holochain is a peer-to-peer database, with a closed network of peers unique to each Holochain app. So if only 10 people donwload and use mova the data is shared only between 10 people.

The main challenge with distributed databases is establishing trust because any node or group of nodes has the potential to manipulate the data that they are hosting. Blockchains use either Proof-of-Work or Proof-of-Stake to establish trust (Proof of Work being the reason Bitcoin uses so much energy). Holochain sets out to use a ‘gossip protocol’ where shards of data are shared and verified with random peers on the network. This is a much lower energy form of validation.

OK so it’s not Blockchain, but why does mova use a decentralised database?

The initial motivation was simple: we needed a desktop app to avoid people having to upload videos to a server to generate ISCC codes and waste a lot of bandwidth, and when we considered where the database should then live, Holochain’s model of sharing it between all the app’s seemed quite elegant. It also overlapped with our belief that no single entity should control film metadata.

But this was quite a superficial assessment, and on building the app it became clear how the many issues around moderation, governance and legal compliance are so much more difficult on a decentralised system. Everything is written to a ledger; so nothing can be deleted, only appended. Upgrades to mova break the link with old distributed databases, potentially allowing those who don’t upgrade to continue using old datasets; in other words, every update has the potential to fork the entire database.

But maybe - if we don’t want to spend our life running from lawsuits or flaming torch Twitter mobs - then this forces much stronger governance practices than the centralised web. The ability to delete someething on a cetnralised service gives a gives of legitimacy, but in reality digital doesn’t work like that. You can delete a Tweet but anyone can screengrab it first; someoneon can download a YouTube video before it’s been taken down; while servers, logs and index store in their caches old versions of content for long after it appears deleted. To quote Arthu Miller “You can quicker get back a million dollars that was stole than a word you gave away”.

We’re on a journey to discover if a distributed hash table is the answer to decentralised and open media governance; that if data is the oil of the 21st century, then distributed databases are like public utitlities.

What does that acutally mean?

Because its generated from each video, the ISCC allows multiple authorities to add layers of meaning and provenange information to ISCC codes – and we think this is the best solution to the risk of a single chief censor for the world. Such a censor would adopt the political views of whichever country they’re headquartered in, and whichever government is in power there at the time. It would end up in an eternal battle between the divides of the political spectrum; such an registry cannot work if it’s only serving half of a population.

Some in the Web3 and Blockchain world see as an alternative to centralization an automated, leaderless structure in place of this, where digital authorities and DAOs folllowing Smart Contracts decide everything, with power centred amongst those who’ve learnt how to best master the new tools, or even simply those who own the most tokens and are the richest.

We picture something somewhat different, where existing central authorities coulld play a role alongside user-led communities to create overlapping layers of meaning and metadata around media. For instance, the BBFC or MPAA might have a centralised registry of film age certification data; while a film journal might build a publicly generated and approved list of films that pass the Bechdel test. The IMDB or TMDB might offer comprehensive film metadata, while sales agents and film rights blockchains might aggregate license and payee info. Amazon Web Services or Google Cloud, GreenHost or Ben and Jerry’s Bicycle Powered Server Farm might provide hosting and CDN, but none of those systems need to monopolise the whole system (let alone track you so they can sell your personal data to the highest bidder).

Nothing too ambitious then?

We’ve planted some seeds with mova for a new kind of metadata registry, but this will only grow into a healthy project with enough demand and a community of support. The next step of development, however long it takes, is something we will want to do as part of a much broader community - we’re too small (and English-speaking) as a team to take too many big decisions. Any system that wants to scale up for media governance, would need to be shaped by the community that needs and uses it.

There also needs to be a model to support the time put into moderating, classifying, verifying and keeping the database in good shape - otherwise it would only be strong in areas with lots of volunteers available. This is just an idea of how a payment model for moderation could work, but without media or a community - or proof of demands, it’s got a long way before getting to version 1.

So what next?

mova is first step in a possibly long journey. It works as a decentralised app for individual filmmakers to register and share their film’s metadata with an ISCC code and payment wallet, but moderation is still centralised and very basic; while the Holochain architecture is so new (it’s technically still pre-launch) we can’t guarantee it won’t break along the way.

For that reason we’re taking a move slow and don’t break anything else approach, with a slow roll-out of beta test accounts to trusted parties. We expect to open this out more widely in the coming months so join our mailing list to be the first to learn when

Edit this page on GitHub

← Web Monetization
How to update →