The leader of new ‘content-centric’ software start-up Eluvio shares her thoughts on evolving distribution models, and the merits of perceiving AI and blockchain as ‘tools’ to ease system implementation.
Michelle Munson spoke at IBC2018 in her new guise as co-founder and CEO of Eluvio, a California-based start-up tasked with creating software technologies geared towards the requirements of a ‘content-centric’ internet. But in fact, she has been a familiar name to show attendees for some years, not least through her long association with file transfer software giant Aspera.
Having co-founded that company in 2004, Munson led Aspera as CEO until May 2017, including through its acquisition by IBM in 2014. A holder of multiple patents, Munson’s achievements include the co-creation, with Serban Simu, of the Aspera FASP transport technology. Now a ubiquitous presence throughout the industry, Aspera FASP can be used throughout the digital media supply chain to supersede satellite and tape-based delivery.
Now, having left Aspera last year, she is at the helm of a new start-up that has been established to address some very specific issues, principally “the cost of, and the economy for, digital media management distribution”.
“What we are trying to do is solve some of the big problems that have existed for a long time around video over the internet,” she explains.
Given that it is “very early days” for the venture, Munson – who took part in the future-facing ‘New. Now. Next: The Definitive Roadmap’ session during the IBC conference – is understandably reluctant to divulge too many specifics about the ways in which the ongoing Eluvio research will ultimately manifest itself. But she is more than happy to enlarge upon the pressure points affecting content distribution as volume and quality requirements continue to escalate.
“We have been looking at what has been holding back the internet for video, as consumer demand has grown,” says Munson. This has led to the identification of two primary areas of concern: “the overall cost of distribution” resulting from the continuing use of legacy distribution technologies despite recent cloud infrastructure adoption; and “the fact that it can be very difficult to get at that problem and monetise meaningfully; to add internet scale and extra value out of the media.”
The answer, believes Munson, is for the industry to invest more energy into the concept of content centric networking (CCN), which aims to emphasise content by making it directly addressable and routable.
“[CCN is] a key idea, and hence we are building the [Eluvio software] in a decentralised and open platform,” she says. “But what’s fundamental is how different to traditional cloud, content management and CDN networks this approach will be. The ability to introduce a fully decentralised software will allow us to achieve efficiencies that have not been done before.”
In terms of how a decentralised approach may facilitate greater monetisation, Munson pinpoints a number of primary benefits.
‘We have been looking at what has been holding back the internet for video, as consumer demand has grown’
“[These include] the ability to monetise each stage of the media access control supply chain, from publishing the media in the first place, to the contributor, to the purchaser and the owner and aggregator – and the approval and transfer to more partners,” she explains.
In order to both enhance workflow efficiencies and the longevity of the media, Munson favours an idea termed ‘media master and reuse’. In this approach the programmable architecture of the fabric “stores the media and metadata in a scalable way and has programmable bitcodes. These bitcodes allow the metadata and core content or essence to be composed and then [become available for] any rendering you might want to undertake.”
As a result, the bitcode “becomes essentially a dynamic pipeline for generating media, and can be very powerful in terms of efficiency, personalisation, reuse of the same core content, and so on.”
One by-product of this new approach to content distribution will be, envisages Munson, a reshaping of the current internet business landscape. “Right now we have a huge economic shift, putting the power in media value to the big companies that control the internet audience. They build distribution as large internet companies [in such a way] that they can leverage their position over the internet and are now ploughing the profits back into making content. [Consequently] we see a complete shift taking place where making content is concerned.
“There is the possibility that a decentralised approach could change that,” she continues. “First of all, because it cuts so much of the advantage down in terms of cost of distribution; and also because it can allow for contribution and infrastructure resources, in the same way that we see with services like Uber.”
‘With machine learning and AI there is more maturity in the market than with blockchain, hence we are seeing some mainstream services for video classification and image recognition’
Munson was also expecting to keep a keen eye on all developments AI-related at this year’s IBC. “There is so much discussion around the possibility of using Artificial Intelligence to basically ‘smarten’ the media supply chain in different ways,” she says. “In some senses it is an area that has been overhyped [to date], but at the same time I have found it very useful for solving larger-scale problems that were previously very difficult to solve theoretically in storage and distribution.”
Would she agree, however, that although there is a great deal of excitement around AI, there remains significant confusion around its long-term application? “Oh, of course, and blockchain is even worse! But with machine learning and AI there is more maturity in the market than with blockchain, hence we are seeing some mainstream services for video classification and image recognition.
“As a technologist the main point I want to make is that if you use these approaches right they offer a tremendous opportunity to solve problems,” she states. “In the past these problems might have required highly theoretical approaches, at least with regard to networking and optimisation, that in practice meant they wouldn’t really get implemented. But now with learning systems they become almost ‘bottom up’; you have the chance to greatly simplify the solution, so they hold a lot of promise if they are looked at as tools.”