Judging from a number of conversations I've had lately, content recommendation is a hot industry topic right now. How do broadcasters and OTT service providers like Lovefilm and Netflix currently recommend shows to viewers? What innovations are we seeing in the market? And what does the future of content recommendation look like?

I put some of these questions to Richard Kirk, Red Bee Media's Head of Development for Metadata, last week. To give you some context, Kirk recently wrote about the richness and complexity of TV metadata and how it's powering increasingly advanced experiences. He pictured a future where information about individual viewers is taken (with their consent) from social network profiles, online search histories, retail accounts and smartphones and is used to provide super-targeted, personalised television and advertising:

"We can truly revolutionise the way that information and choices are presented to those viewers (including ads of course). Content can now be highly targeted to that viewer and more importantly, highly targeted in time, location and context.

Let's imagine that I'm that viewer. I'm about to leave the house but I have 15 minutes to kill. What can I watch to pass the time? Don't suggest a movie I might like, there's only time for cats on skateboards. Offer me a chance to buy a related product? Not now, you know I only shop on-line late at night. Sync some appropriate content to my car's entertainment system? Yes, perfect. How did you know I was going out for lunch?"

KANJI: Hi Richard, let's start with the humble TV guide. Have you seen any applications that marry TV listings with user's own data - either that you're developing yourselves at Red Bee or from other companies?

KIRK:  Well, there's two parts to this. The first is around the topic of "recommendations". Last year Red Bee acquired a search and recommendations software company called 'TV Genius' and they analyse and look for relationships between a number of pieces of data: TV metadata, how individuals navigate through television schedules, different programmes that are available etc. And from this they identify trends: for example, people who like one type of show may generally favour another kind of show too. This knowledge can then be used to enhance on-screen TV guides. Now in and of itself this isn't a particularly new thing and we're not the only company doing it but we like to think we're one of the best.

New content needs to be provided using one method of 'browse, navigate, discover'

The other thing we're seeing is the bringing together of the traditional world of linear viewing with VOD services. Over-the-Top (OTT) services to me haven't been done that well up until now. They've been fairly stand-alone and disorganised and lacking in an integrated experience for the viewer.  Right now, consumers have to go one TV guide for linear programming and several others for VOD viewing. So you search for programmes inside the BBC iPlayer app and then quit that to look through the ITV Player app and the 4oD app and so on. TV platforms are now recognising this. New content needs to be provided using one method of 'browse, navigate, discover' and, as a company, we're being asked to solve this problem and bring it all together to provide one consolidated experience.

KANJI: But this isn't available yet is it? I mean there are 'Social TV' services like zeebox and KIT digital which has recently released a white-label product which incorporates a backwards-EPG and social recommendation but no one broadcaster has launched the kind of service you're describing which amalgamates linear and VOD?

KIRK: Not quite. There are two types of VOD. So, if you take the Virgin TiVo service and some of the services provided by European cable providers then yes, there are platforms out there that are starting to blend linear and VOD. But if you look at the OTT service providers which are entirely delivered by app and not via set-top box then no. Probably the one platform that is going to successfully do that and be a fundamental differentiator is YouView. If my understanding is correct, every piece of content on YouView will use the same metadata standards and so all content will be as discoverable as everything else.

KANJI: In your article you gave an example of only having 15 minutes to watch something and being recommended specific content by an intelligent system that recognised this. That type of service doesn't exist yet either does it and, if you think it might happen one day, in what time-frame?

For me, as far as the recommendations piece is concerned, we're only at stage 2 out of 10. In a few years' time everything will be real-time, connected and responsive.

KIRK: Again, not yet. But, for example, Google has recently acquired patents from Motorola and if they combine this smartphone data with all the other data they have and integrate it into a VOD platform then it could happen soon. To make content recommendation really useful you can't just say we think you'll like one show because you liked this other show. You need to put viewers' choices into context: what can you glean from other sources like social networks? What kind of content do people watch on different days of the week? What are the different patterns of behaviour? For me, as far as the recommendations piece is concerned, we're only at stage 2 out of 10. In a few years' time everything will be real-time, connected and responsive.

KANJI: If we are at Stage 2 right now then what are we missing to get to Stage 10? Is it just a question of having the will to do it or are we missing fundamental data sets and technologies?

KIRK: I think there are three things here. First, we are still missing the will to do it but the second thing we need is the consent of individual viewers to be allowed to use their data in this way. That in itself is an interesting point. The current trend is that people are becoming more and more comfortable in giving away their privacy. We basically sign our lives away every time we agree to get cookies or set up a Facebook or Google + account! Is that trend going to continue or is there going to be a backlash of some sort? My guess is that we will carry on giving away our privacy because we're getting useful content and services in return.

The third thing - which I covered in my article - was how do we do this? We're talking about integrating many different sources of data and in huge volumes.

KANJI: And talking of privacy and how this might be done, can you see a future in which we, as consumers, are happy to get advanced personalised recommendations as long it comes from a gateway device - possibly a replacement for a set-top box - or a gateway service that we feel secure with?

KIRK: Well, I'm not an expert in security so I'm not qualified to say how this might pan out but if you're implying that there might be a single place that would "authorise" - so maybe something like Facebook becomes a central point of your interactions with just about everything including TV, media and everything else - then yes, I can imagine that would be your "trusted partner".

KANJI: Final question. You mentioned a "Cloud PVR" in your piece which would function like a regular PVR except all your programmes would be saved to the Cloud. Are you aware of any Cloud PVRs?

KIRK: No. I was at CES earlier in the year and this was a very strong theme. The question about physical devices and set-top boxes only has one answer and that is that everything will end up in the Cloud. If you look at the economies of scale for Cloud services presented by companies like Google and Microsoft I can't see that manufacturing bits of hardware to sit under your TV is ever going to be economically viable again.

KANJI: Agreed -

KIRK: Everything is turning into software and equipment is becoming commoditised. It will be really interesting to see how this develops.