Digital relationships: building a platform for future TV
Some time ago, Pavel Cherkashin, an entrepreneur and VC, asked me to comment on his piece regarding the future of TV. I was quick to jump on this opportunity and enjoyed it. Yet, I felt there was much more left to say, so I decided to write a piece on my own.
A look into the future
Having spent last 5 years in the online video industry, I see around lots of clues on how this industry shall and will look like in the next 3 to 5 years. Everyone (myself included) seems to agree that TV and video-on-demand will gradually merge in a universal OTT experience, where users will be able to choose what bundles or individual channels they would like to have, and whether to watch certain show or episode live or on demand. Add easier internationalization / adaptation of content thanks to instant translation and captioning, and this experience easily becomes global (subject to the content rights, of course). Sure, we all see it’s already happening, so the future seems inevitable, and everyone seem to benefit from it.
There is one big problem that is yet to be solved — the problem of choice, that is not going away but in fact is getting worse (study shows that Netflix users could spend up to 18minutes to choose something to watch). Imagine what will happen when you are be able to chose from 3 Netflix-sized catalogues? 5? 10? In our present and future connected lives, this problem is not specific to choosing movies, or any kind of content. Same is true for choosing any goods or services on the Internet. World is getting more and more connected; choices are widening but the problem of actually choosing something is relevant than ever. How do we separate signal from noise and get what we really want, not what manufacturers / content owners / distributors want us to get / see / use?
There seems to be lots of available solutions when it comes specifically to TV and video: there are data- and knowledge bases (IMDB, …), there are all sorts of guides (The TV Guide, BuddyTV, Fan TV, …), media outlets and blogs (THR, Variety, Fandom, …), analysts (Nielsen, …). Not to mention Facebook, Twitter, YouTube, Snapchat that want to become your “single destination” for everything. We post, like, discuss, share, watch, react. We do it every day. We do lots of it. In one way or another, we generate lots of explicit and implicit clues with regards to what we really like (don’t confuse this with “Like!” button) or don’t like and even more — what we are, and what we are not.
Yet this seems not enough, not even close to get our lives easier when we have our evening or weekend TV/movie time. Why?
Relationship? It’s complicated
Obviously, the problem lies in the fact that no single service has (and, fortunately, could have) all information on what I like and what I’ve done. Facebook doesn’t know what I’ve just watched on Netflix, and Netflix doesn’t know what I like to read on Twitter. What are the consequences? Well, we know the drill — almost every time we subscribe to a new content service (take Flipboard or Apple News as example), the first question is — please tell us something about what you like, chose at least 5 topics from the hundreds that we have for you… otherwise we just won’t let you past the front page!
Well, no one seems to have full knowledge about us, and we tend to be much more cautious exposing our digital selves. Still, when it comes to finding something, what do we do? Who do we turn to, and how do we do it?
Wait, aren’t we missing something? What about good old search, that tends to gradually become more and more human (think Alexa, or Siri, or Google Assistant)? Well, replacing keyboard input with a sophisticated voice recognition software doesn’t change the essence of search. Search still… sucks. Well, not entirely. Search works well when you know exactly (or at least approximately) what you are looking for. That’s definitely not the case when it comes to TV/movie time for at least 99.99% of us who doesn’t know exactly what he or her is going to watch next (to tell the truth, I believe that the remaining 0.01% are heavy IMDB users and do their homework in free time). Remember the epic fail of Google TV (not to be confused with Android TV), where TV was meant to start with a search bar and a keyboard?
Let’s look from another angle — what is generally the best working decisive factor when it comes to making choices? Trust. Relationships. Recommendations from someone we know, we trust. Something we know that we most certainly will like. In movies and TV world this could be a certain TV channel, certain program host, certain actor or movie studio, opinion of a certain media outlet or a friend. How can we build a system that makes use of all these and solves our little but important problem — what shall I watch tonight?
Building relationships and trust takes time. This is what we humans do through our entire lives. Can this experience be replicated digitally, but without necessarily putting this into some third party hands (be it Facebook or Google or Apple)? What kind of system / service could handle it? Certainly, there isn’t anything like this yet, and I’m not sure we’ll see one. However, now it’s rather easy to imagine a completely de-centralized system (think blockchain-type one) that has an anonymized instance of your digital self exposed to all kind services you use or may use. This will work well only and only if the majority of the most important services use it. Why would they? Because, once this setup is in place, everyone wins: you don’t sacrifice on digital security while letting all your suppliers get holistic picture of your relationships. Which in turn, help them serve you (and others) better, because they know exactly what you like or may like. Whatever is shared by participating parties makes the whole system more accurate.
Why is it better for you? Because you are in control of your digital self, and no-one else (de-centralized encrypted ledger). What’s more, like in real life, you could build different “circles of trust” around yourself. You could choose who are your primary sources of information and recommendations. Those relationships would exist throughout your physical life and reflect your human relationships (to the extent you deem reasonable). They would strengthen if your sources / suppliers continue to bring value and stay relevant, or would become less relevant or vane altogether if they don’t. Reaching you would also resemble the real-life process — you would trust recommendations from someone within your inner circle to introduce you someone from the “outer rim” (LinkedIn has been doing this for a long while).
Well, this all sounds purely idealistic, and, like I said, may never see life, because of the necessary effort to shift our current state of Internet to this new one. Still, as a concept it shows the gaps in our current digital world and gives ideas where we can go from here. Let’s look back to our initial idea — how to make TV personal and entertaining. How can we truly personalize viewer’s experience? How do we shift paradigm from “tell me what you want and we’ll show it to you” to “since we know and trust each other, this is what I think would be the best choice for you for this moment”?
Bridging the gaps
Let’s talk about recommending and recommendation. What is the basis of any good recommendation? Trust, for starters, and also knowledge of the subject matter. In my opinion, those things are inseparable when it comes to recommending something. Currently, we rarely see true implementations of this, precisely because it’s hard to capture knowledge and even harder to capture relationships. And if the system is limited to one provider (think Netflix or Spotify) it can never be powerful enough. One really interesting company, the Video Genome Project (VGP) has been just snatched by Hulu that apparently is getting very serious about its new TV offering.
To achieve ambitious goal of building “over-the-top” knowledge and recommendation system that could actually solve the users’ problem, we would need to build a graph representing the subject matter (TV and video domain) that goes way beyond usual relations based on genres, cast and crew. We would need to ingest everything that we know about movies, shows, programs, actors, plots, stories, etc. and make it fully-searchable in real-time. We would then need to add and correlate all kinds of engagement data we could find (tweets, likes, posts, discussion, reviews, viewership data), add time and geo-location layers, and then apply deep-learning algorithms to make predictions and recommendations. This will probably create something close to what I talked about before. Were the folks at VGP doing something similar? We’ll probably be able to tell once it gets integrated into Hulu TV and sees the world.
Will building such a recommendation system and integrating it into UX of the premier league of Internet TV players solve the problem “what shall I watch tonight?” Probably not, or not completely. No system, no AI will ever be able to finally “decide” for you. Neither you probably would want it to. We are still humans, and we have some “variables” that are not known to machines. Our mood, for example, that could change rapidly and unpredictably and that can have dramatic impact on the relevance of any recommendation. So ideally the system shall come up with several scenarios for every moment you would use it. Just like good old TV did.
We happen to live in an exciting time when 50+ year old TV industry is changing forever (and hopefully for good), gradually fusing with the Internet. Promises of these changes are endless, and are yet to be understood, captured and realized. And while we watch the battle of giants (Netflix, Amazon, Apple vs. incumbent TV players), where the battlefield is becoming global, lots of additional value is being and will be created by teams around the world.
I’ll be back with more thoughts on the matter.