Yesterday about 60 people spent 3 hours workshopping the Future of Online Video at the Amplifed 09 session. A few people have already
blogged their
sessions, and the Twitterfeed is
available here.
There is a strong temptation to say that because X happened in industry A, it will by extension happen in industry B. Nowhere is this more tempting than in the evolution of online media, to say that because user generated content (in its widest sense) it is disembowelling newsprint and music, the same will happen in online video.
Readers of earlier posts of ours on this blog will know we are a bit pessimistic about how User Generated Video will survive - by and large its low quality, hard to find, has poor metadata - and the infinite sea of it means finding anything is very hard. As Sarah Lacy notes in a recent post, YouTube
has a problem:
.....so many of the videos are user generated content in which I don't have any interest. They're like the thousand of listings for "Buy It Now!" socks on eBay. But Hulu isn't just better because it can have professional content: It's the technology. The last three times I've looked for a video clip, I've spent half an hour scouring Google and YouTube only to get a flood of inaccurate results. Each time, I've tried Hulu as a last result, and found the clips within minutes. Hulu has better fields, parameters and user interface for searching videos than Google, which still appears to search for video the way it would for text.
.
Research we have undertaken on the future of online video implies that User Generated Video has 3 major issues vs higher quality content:
(i) The economics of video is different, and favours the large players more than text and audio based media does
(ii) The user dynamic is different - the customer has a higher preference for high quality video than audio or text (ie tolerance for poor quality UGC is lower)
(iii) Advertising works differently in video - text and audio models are not directly transferable
Video Economics is different
In a nutshell, it costs much more to create, curate, and distribute a piece of high quality video media:
Creation - a 30 min TV show, or a movie, or especially a 13 episode series is a very resource intensive thing to do. This requires more organisation and structure in the industry - finding the the single blogger or brilliant MySpace musician is nearly impossible to imagine in long form online Video. A 5 minute good quality Web TV show requires many hours behind the scene even for "head and shoulders" sets like Rocketboom etc - 30 minutes is exponentially more, there are few scale benefits to be had.
Curation - this incorporates functions such as editing, making the right content findable etc - automated systems such as Google search work well for text, less well for video - especially long form video. Tagging etc has not proven to be a scalable approach to date, and the Semantic approach is still always available "in the near future"
Distribution - A 4 minute blog post is c 1 Mb give or take, 4 minutes of a song is about N Mb give or take, 4 minutes of TV quality video is about X Mb - and a 30 minute show is Y Mb - this costs a lot more to distribute to an audience - at say $1 per Gb its $A vs $B per 1000 downloads - this totally changes the burn rate economics for any startup spooling video rather than text or audio
The User Dynamic is different
Despite continual
claims to the contrary by media 2.0 flagwavers, all the emerging evidence is that most people, most of the time, like linear stories that are told well, and preferably in high quality formats. In fact, as this analysis below shows, although UGC video will fill the tubes, there will be very little money in it.
Note YouTube's rush to offer better programming content recently, and note the rapid growth of Hulu, despite it being lambasted at startup by many of the "Web 2.0" leading lights
Advertising Works Differently on Video
The web staples that have driven great wealth for Google et al in "traditional" webpage media (Adwords etc) just don't translate into Video. Nor does trying to make much money from advertising on 3 minute clips of poor grade content, as YouTube has found over the last 2 years.
In a nutshell, the "long tail" advertising revenue that drove text / picture based webpage monetisation apply far less for video. What does provably work so far is:
- Sponsorship of higher quality short form media on shows that repeat
- "Traditional" Ads (the 30 second shot and similar) on Long Form media
- To a lesser extent, product placement (it ages fast)
So, given this, the opportunity to discuss how to solve some of these problems at Amplied was very useful. In the first session We discussed 3 key areas that UGC video needs to solve:
How do you find stuff?
As noted above, its a problem. So - how to solve? The "Folksonomy" has by and large proved to be a chimera as millions of people tag videos in all sorts of ways, the variety just gets too hard to parse easily. James Whatley, blogging in the
Spinvox blog post, notes:
There is one question you might be asking yourself right now and that is:
“Why would SpinVox attend ‘The Future of Online Video’?”. That’s a very good question, that requires a very good answer. All will be revealed very soon.
Very mysterious - but one thing we discussed was to turn the audio stream into text (which Spinvox does) and make that searchable. Seems like quite a good plan.
Even if you've found the Video, how do you find the bit you want?
One of the major issues we discussed was that its not just searchability of video that counts, you need to be able to scan it fast to find wat you want. Few people have the time to sit through video in a linear listening mode, and the view is some form of easy, rapid navigation through the video is required. The common view is that "video search" per se is a way away, so "good enoughs" could emerge in the interim. For example, simple videoscanning (eg a good slider system) and some form of transcript would be useful (even something as simple as this TED video system below helps navigation hugely)
Clearly to do this manually for every video is hard. However, it is not impossible to imagine that, is some form of metadata like the above was made common, people would be motivated to use it because it will make their video more attractive.
In future, making User Generated Video may thereforel require the addition of User Generated Metadata - and lack of it would be picked up by a playout system, which, if for example it asked "show only videos with transcript metadata" would kill videos without it.
What other data can you pick up to help video search
Given that tagging/folksonomies/semantic webs are not really fit for purpose right now, what other options are there? One of the issues discussed was how to cull cr*p video using data like the bit sampling rate, as many people do with music, and that led to discussions about what other metadata could be deduced automatically.
Who is organising say all the great videos on TED, MIT, Oxford etc and curating the "Interesting" channel. Even making it easy to organise YouTube videos into some form of user chosen channels would be ahuge step forward
The "big picture" realisation was that "traditional" video media formats have had 60 years of evolution of the best ways to organise it - just agreeing what the "Electronic Program Guide" (EPG) should have in it is hard enough. never mind what it may look like and how it would work.
The future of Seesmic, Social Video Objects and Copyright
The second session was a discussion on the "videotwitters" like Seesmic - general view is that they were not as yet compelling at all due to all the metadata issues, and the "shorter the better" - like 12secondTV - was seen to be the best way. How will they make money - the view is that they are not sandalone and naturally belong as adjuncts to Telco type services, paid for as part of a connectivity bundle
The discussion sort of sideswayed off to talk about how low the interactivity of conventional media was, and the view that a "use it or lose it" culture may emerge - ie if a piece of MSM is put out but doesn't join the conversation, the conversation may well appropriate the medium (not so much piracy as squatting if you like). thsi of course led to discussions about teh future of copyright, especially for these nuggets of social object content. No conclusions, except at present the mountain of new usage and the Mohammed of traditional copyright dogma will have to come together in the next few years.
So how do we use video socially then?
The last discussion tackled social video - the key observations were that:
- Joost didn't get to define how we socialise while watching video - a horizontal app, Twitter, is now the chat system of choice while people watch TV.
- When it come to enjoying content we are not so much social as tribal - we cleave to people who liek what we like (we duscussed the difficulty if dealing with social text net friends who had lousy taste in music and video - do you have different friends for different mediums?)
- Social Media works best when people react to smaller nuggets of content - its not really great for essay clinics - so video will need to be broken into smaller chunks anyway.
Another aside on this - we noted that a lot of us would fire up say a TED video, and then just listen to it rather than actually watch it. Why was this? Well, to make a video takes a lot more care, the presenters are more structured in general compared to audio (also, as one person noted, you tend to eyeball the video's presenter to take a view about listening more) for so the bang per buck per minute is higher.