1300-1600 words, 6 1/2 to 8 1/2 minutes.
An Etching of Samson by Julius Schnorr von Carolsfeld, from an 1882 German Bible.
“He grasped two pillars of the temple and bowed himself with all his might” - Judges 16:30.
Introduction
This post is about online services which rely partly, or wholly, on user-contributed content. It’s about what happens to that content if those services close.
I’ll examine the constraints and motivations of the parties involved. I’ll then present a solution to the problem of what I call content custodianship. Finally I’ll explain how we might harness human nature to turn that solution into a successful, practical implementation.
If I’m right then what I propose will help all user-contributed content companies get started quicker and run leaner. If the worst happens and their business falters, it increases the chances of salvaging value. It may even accelerate the next hopeful content platform. It’s probably worth considering because a lot of user-generated content is far too valuable to waste. Even if the temple it built falls.
SoundCloud
Skip to “The Samson Option” if you already know about SoundCloud.
SoundCloud, the online audio distribution platform is in some difficulty. Reports suggest they have been trying to raise a round of funding since 2016 and had discussions with potential acquirers over the last year, which didn’t reach agreement. As a result of this, approximately 40% of the staff were let-go earlier this month. The San Francisco and London offices closed. Berlin and New York remain.
The company raised a $70 million debt in January 2017. This means creditors are confidant enough of seeing that money again. You don’t bridge a company out over a chasm, knowing you can’t reach the other side.
Maybe the company’s $700 million valuation (Jan 2014) is a problem. Maybe it’s the (dis)agreements with the music industry. Maybe the strategy is simply to align costs with income until the corner into profitability is turned. SoundCloud is registered in the UK and files accounts at Companies House. Whatever the outcome, we’ll be able to understand what happened and why. Mark your diary to check their 2017 accounts.
Investors and management are taking steps to buy the company time and increase the chances of a positive outcome. While it’s sad for those who have been let-go, I don’t doubt for a moment that it was the right thing to do. Having closed several funding rounds the company has many backers. Among these are some of the most respected VC firms in the business. Index, Doughty, and Perkins. All of which have years of experience in maximising the chances of the best possible result, and 700 million reasons to try.
SoundCloud is a platform into which many independent, aspiring artists put their own original content over the last decade. Some will have taken care to store duplicate copies others won’t. Users are now contemplating the possibility that the service might not be around, at least not in its current form. Some who enjoy SoundCloud content but don’t contribute, are also wondering if they will lose access to the library in the future. An inevitable content-grab and predictable blocking by the company followed. (Updated 23-07-2017)
I don’t really want to talk about any of that though. SoundCloud is just one example of something much larger. It’s an example of what happens when users worry about loss of access to content. It’s a learned behaviour.
The Samson Option
What should happen if a service which relies on user-contributed content needs to close or change in a way which results in loss of access to that content?
Part of the reason for making this a self-hosted blog, is control. Last time I checked, LinkedIn didn’t offered me the option of downloading all my posts in some convenient bundle. Self-hosting the content means I’ll never lose access and can “get out” any time. It’s not a big deal. These articles could all be downloaded in less than a minute. The same can’t be said for user-generated music/videos. Just 1 hour of 720p mobile phone video is 2.2GB of data, 10-15 minutes to download on an average UK connection. Multiply by 4 if you are in a rural location.
- What if you were a user making an hour of video/audio every week?
- What if you had built up an archive of many years work?
- What if a thousand users decide they were concerned enough about losing access to retrieve their content?
- What if non-contributing users try and do the same with the whole library?
- What impact would such an exodus have on the quality of service provided to other users?
- Could that stampede even accelerate the demise of the site?
- Due to a spike in bandwidth.
- Or collapse in performance.
- Or the potential for a rival site to grab the library.
Returning to SoundCloud for a moment, their service attracts over 175 million active monthly users. The numbers get big.
The Content Custodianship Problem
Should sites with large amounts of user-generated content have some kind of “escape module” or “insurance policy” for users content?
Do users even care sufficiently before trouble strikes? Everyone is wise after the fact naturally. If users do care, what would reassurance look like for them? Technically how would one implement an escape module or insurance policy that safeguards the availability of content in such situations? How can we reconcile that mechanism with the site operator’s need to monetise the content right up to the point at which it becomes non-viable? Who would bear the cost of the escape module? What would be the right way to charge for such a thing? Anything has got to be better than either:
- Losing access to that content forever or…
- Uncontrolled, uncoordinated “looting” of the library by panicking users.
Who’s Paying For It?
The site operator doesn’t want the added cost of an insurance policy. End users aren’t going to pay for something they may never use. It’s worse than that. For most users it will be hard to hold in their head two separate and conflicting ideas:
- Putting content onto a single site is good because it makes that site more valuable and likely to succeed.
- Doing so increases their exposure to risk of loss of 100% of that content because it’s all on one site.
It’s a bit like selling life insurance at the airport departure gate. It’s something nobody wants to hear about. That doesn’t make it a bad idea though.
I’m sure there are some good studies on the psychology of insurance and perception of risk. I can tell you from my work in Cyber Security, that most peoples abilities to identify and evaluate risk are better suited to the savannah circa 15,000BC than the modern digital economy of today. Users aren’t going to pay for “content insurance” for mostly-free content, especially since the cost of providing it increases with the volume of material. At this point I’d normally call it a day. This must be what the “interesting but not-profitable” quadrant feels like.
Or is it?
A Solution For Content Custodianship
What if users content wasn’t stored centrally? What if the content was encrypted and distributed across the systems of users/subscribers? The same way that large files (e.g. movies) are stored in whole or in part on a peer to peer file sharing network.
This reduces cost for the service/platform operator, they no longer have to store or serve the content library. They may still wish to keep a copy, to ensure there are enough “seeders” when they are still building their user community. Other than that, they’d hold the user interface (web site), the metadata, and the decryption keys for the users content. That’s where the real value is.
End users would lose a bit of disk space (which they could limit and reclaim at will) to hold content files (or blocks) when they become a user of the service. They could set the disk space to be zero (and would in the case of mobile devices), but humans are naturally altruistic. They would understand giving up a little of something they probably didn’t need, in order to buy an insurance policy. They can see it isn’t a bad deal. No money changes hands so it doesn’t feel like they are losing anything. If the service or site goes bust, the operator can release the content keys. If executives or backers think they still have a chance of monetising the content, they retain that option by holding the metadata.
If the site operator goes bust and there is no further need to limit access to the content then they can release the master key. From then on it is just a P2P file sharing network. If they were feeling generous then they might want to put the file hashes and some metadata in pastebin. Cost: zero.
- Users need never worry about losing access to content, in exchange for giving up a little disk space and upstream bandwidth they probably would never miss.
- The site operator has exclusive use of the content for as long as they wish, and holds on to the valuable metadata about each file.
- The site operator has longer to monetise the metadata and content in the event of a company shutdown, because their costs are lower.
- The cost of bulk content delivery declines as users of the service increase.
- Access to the content archive (and potentially the metadata) will make life easier for the next similar company (who might pay for that benefit).
Now all we need is for someone to design and build this control framework on-top of existing peer-to-peer protocols.
Technical Challenges
Delayed start to playback (particularly video) is a significant technical challenge. Content must stream from multiple peers and be assembled in the browser in the right order before it can be viewed or heard. I don’t think this is insurmountable. Once streaming any kind of video was an outrageous fantasy. Now thousands use Periscope every minute of every day on their phones. Perhaps a hybrid architecture would suffice. Perhaps P2P for the totality of the library, and traditional centralised (and/or CDN) streaming for live playout.
Either way, cost of storage of the content library and access to it “post-mortem” is pushed off to end users. Is a 5-10 second delay an unacceptable price to pay for a truly decentalised libray and delivery model? How might a total collapse in bandwidth/CDN charges improve the viability of user-generated content sites?
It’s probably worth figuring out because everyone wins and nobody loses.
