Interview: John Callahan, CTO of ActiveVideo Networks

ActiveVideo Networks (formerly ICTV) recently appointed interactive TV pioneer, John Callahan, as its CTO. Callahan previously spent 15 years at Time Warner Cable, where he was involved in many of the operator's interactive TV and VOD initiatives, and where he served most recently as SVP of software engineering for its Advanced Technology Group. He was a significant contributor to the creation of the Time Warner Cable Pegasus digital television system architecture, and to such seminal interactive TV technologies and standards as the Interactive Services Architecture (ISA), VOD, OCAP and EBIF/ETV. He was also one of the principal developers of the network PVR service, MystroTV, and subsequently led the transformation of that service into Time Warner Cable's current StartOver offering. In addition, he was closely involved in implementing the operator's caller ID on TV service and its OCAP set-top navigator. Earlier in his career at Time Warner Cable, Callahan was part of the engineering team that created and deployed the pioneering interactive TV service, the Full Service Network.

Callahan recently spoke to [itvt]'s Tracy Swedlow about why he decided to leave Time Warner Cable and join ActiveVideo Networks; about the projects he plans to work on at ActiveVideo, including developing a new programming-navigation model which he terms "never-ending navigation" and which he says will "put content front and center"; about his view of the current status of the US cable industry's OCAP/tru2way and ETV/EBIF standardization initiatives and how they relate to ActiveVideo's technologies and services; and much, much more.

[itvt]: Could you give us a little background on your time at Time Warner Cable? What motivated you to leave Time Warner Cable and to join ActiveVideo Networks?

Callahan: Well, the first thing I would say is that Time Warner Cable was absolutely a fabulous experience for me. It was a company that to some degree I fell into by accident back in 1993: I was asked by folks at what was then US West's Advanced Technologies group if I was interested in working with the Time Warner Cable guys on a crazy project called the Full Service Network.

So, the next day, I went to Time Warner Cable and talked to two guys named Jim Chiddix and Mike Hayashi, and that was how it all began. I really had a very special experience with Time Warner Cable, and so it wasn't an easy thing to leave. However, I think more than anything, I just got to a point where I had to make a change. Let me explain. When I was at Time Warner Cable, I had had a series of five-year plans for myself: the first of those was to take the opportunity to work on the cable industry's analog-to-digital conversion. So, from the mid to the late 1990's, I had a really unique opportunity to have a major role in Time Warner Cable's Pegasus program, which was the company's effort to convert to digital. During that time, we also had a covert plan, which Jim Chiddix used to describe as an attempt to get to "everything-on-demand." So while we were ostensibly just implementing digital broadcast, we were really preparing a platform that would gracefully migrate into video-on-demand. (By the way, in those days, video-on-demand was simply a generic phrase that meant using unicast video streams on a one-to-one basis with the customer. Subsequently--fortunately or unfortunately--it also became a marketing term, and towards the end of the 1990's and into the early 2000's, it began to leak out as the brand name for this new service that cable could provide.)

Anyway, this vision of "everything-on-demand" has been with me ever since. My second five-year plan saw me working on building out the video-on-demand platform at Time Warner Cable, and that eventually resulted in Jim Chiddix asking me to put together the engineering team for an internal start-up that became known as MystroTV. The goal of that project was to make every broadcast program capable of being time-shifted--so, capable of being paused, and started over again, and even recorded and saved for later viewing much like you would do with a DVR set-top box. "Network DVR" was actually a phrase that we used back then to refer colloquially to what we were doing.

That project was technically very successful. We did a trial of the MystroTV service in Green Bay, Wisconsin in which dozens of programmers took part, and we had 200 of Time Warner Cable's employees up there using the service for over a year. The response was very positive: even if they already had a DVR, they loved the convenience and having the ability to watch what they wanted whenever, even if they hadn't remembered to set up their DVR schedule to record it.

Unfortunately, however, it turned out that we were way ahead of where the business was back then: the advertisers and the rights-holders for the programming just hadn't figured out how to make this all make sense from a business perspective--e.g., who would make the money, how would the chain of content copyrights be protected, and how would the advertising models work? And so, because we were so far ahead of the curve, a big part of my third five-year stint at Time Warner Cable was basically about productizing a lightweight version of MystroTV--which ultimately went to market as Start Over.

Start Over has actually become quite successful. Time Warner Cable has indicated that it is available to a sizeable number of their customers and its convenience is being well received by customers. I understand the complexity of content copyright issues so it's encouraging that there seem to be models maturing in that area as well. All in all, that's a vision that's turned out quite well.

The other thing that I was involved in during my third five-year stint at Time Warner Cable was building out the OpenCable platform. When MystroTV was ramped down, along with Louis Williamson and most of the MystroTV team, I returned to Time Warner Cable to work on this big task of actually implementing the OpenCable Application Platform. I spent a couple of years working with our internal engineering and development resources, but also with Scientific-Atlanta, Motorola, Samsung, Alticast, Vidiom Systems and others, to turn the OCAP platform into a reality. As with everything else, when you got down to the details, there was a lot of work to do.

When I got to the point where our implementation of OCAP was actually being launched in the field--so, spring of last year--I started to cast about for what my next big project would be. I came to the conclusion--which many others had too--that I wanted to build a cable platform that would allow the convergence of video from IP sources: so, of Internet-based video or Internet-sourced video (I tend to avoid the term "IPTV," by the way, because it's defined in so many different ways, depending on the constituency). Basically, I was interested in expanding the platform to create a user experience which would allow the viewer to browse IP video streams in a very intuitive and elegant manner--and which would be the realization of that "everything-on-demand" vision we'd had nearly 13 years ago. Actually, the whole time that we were building OCAP, I had harbored a desire to push on this vision for "everything-on-demand." The Internet is an interesting variation of that vision, in that you certainly have a one-to-one experience, as you browse and find things you're interested in. But the question is, how do you replicate that experience on the TV in your living room as you sit in your comfortable couch with a relatively primitive device in your hand--i.e., the remote control? How do you make that kind of experience natural and intuitive? How do you make it so that the end-user can sit down and browse video, and not even notice whether it happens to be coming from a satellite in linear format, or coming from a VOD server, or from a DVR, or even a disk drive on their PC via a home network?

About a year ago, as I became excited by the idea of working on this, I started to think it would be a good time for me to try things from the other side--to go out to the vendor side of the product development community and work with a number of different operators,. That's definitely one of the things I'm looking forward to: the fact that I'll have the opportunity to work with a plethora of operators--even with operators that might not be in the traditional cable world. I was definitely very comfortable at Time Warner Cable. It's a terrific company with great people, and I left on good terms with everybody there and felt that we'd achieved a lot. But, look, I'm getting older! If I don't give this a try now, I may never give it a try.

As for why I decided to join ActiveVideo after I left Time Warner Cable: I got to know Jeff Miller last March, and ActiveVideo's vision of enabling the sourcing of video from the Internet and of using a very thin, lightweight client is one that I share. That idea has actually been at the heart of what I've been working on for years. The last two years of ActiveVideo's efforts have resulted in a very flexible video streaming and interactive applications platform that can run on any number of devices out there, including even some of the oldest, least powerful set-top boxes as well as the newer boxes, and, of course, IP-connected devices such as Web-connected TV's, Blu-ray players, video game players and others. And so, from talking to Jeff, I began to learn a lot about what ActiveVideo is working on, and I found that we had a lot of commonality of vision. We were both excited by the idea of developing robust, but intuitive navigation models for interactive video services. Jeff eventually asked me if I'd be interested in coming on board and actually building that stuff, and I got more and more excited by the idea, and so that's how I ended up here.

[itvt]: Would it be fair to say that one of the reasons you left Time Warner Cable was that they weren't as interested in moving to IP--as interested in the potential for cable of IP technology--as you were? I know that cable is one day expected to move to IP, but was it that you perhaps felt it was moving too slowly?

Callahan: It wasn't about a negative like that. It was really because I personally was looking for an opportunity to develop products that might ultimately have a wider usage than the kinds of products I would have developed within a large cable operator. I really wanted the opportunity to work with a company like ActiveVideo Networks where we can actually develop personalized video streaming.

What this company can do is basically create intelligent streams: i.e., synthesize MPEG video and metadata together, deliver that stream to a very thin client, and allow the viewer to have a personalized, relevant viewing experience. I use the term, "intelligent streaming," to indicate that the (unicast) MPEG-2 stream delivered to the customer is actually composed of several different sources of media--e.g., an MPEG-2 file from a local VOD server, one or more video sources streamed from an Internet-based media server, a set of JPEG images, etc.--that are "stitched" together into this standard MPEG-2 stream at the moment requested and delivered to the set-top box. As the customer interacts with the application--via the expected press/select of UI elements, cursor movements, text entry, etc.--each command is sent back to the ActiveVideo application (running in the network) and, depending on the application logic, causes the various media sources to be modified and the process of composing the MPEG-2 stream to be repeated. This all happens in real time, so the customer has a great video-rich, interactive, multimedia application experience that is unique to him or her. The fact that ActiveVideo can deliver all that over MPEG-2 to existing set-top boxes--and, of course, ultimately it can be delivered over IP to other platforms too--was definitely very interesting to me.

[itvt]: Would it be fair to say that you were somewhat frustrated by the amount of time it's taking for tru2way/OCAP to be deployed, and that you felt ActiveVideo's platform presented a viable alternative? Or do you rather see ActiveVideo's platform as complementary to tru2way and also to EBIF?

Callahan: Well, I would say that we always expect these new technologies to be implemented faster than actually ever happens. When I was working on FSN, back in 1996, I figured, "OK. By next year, this technology is going to be everywhere." I was only off by a few years, as it turned out. Same thing with video-on-demand: we spent what felt like an eternity getting cost-effective video streaming in place, and it still ended up taking another three to five years, depending on how you want to calculate it.

So yes, I would say I'm always somewhat impatient about these things. I would love to see OCAP/tru2way and EBIF--and a whole bunch of other technologies--become ubiquitous much faster than is currently happening. I definitely do think it will take a while before OCAP is universally available. Even though it's basically a Java platform--so based on a tried-and-true technology--there's still a lot of room for more optimization. It needs to be scaled and made robust. I think it will take a few more years for it to become absolutely bulletproof.

But I do think that what ActiveVideo does is actually very complementary to these technology standardization initiatives: I don't see it as competing with OCAP/tru2way. In my opinion, nothing but good can come of there being a common platform that's distributed across multiple operators. The worst thing that could happen would be for this situation to continue where there are literally dozens of different versions of clients from different manufacturers, and even multiple different versions of each manufacturer's software. The amount of time it takes to regression-test against each different version, every time you try to introduce or upgrade a service, is simply cost-prohibitive. So I definitely think that the tru2way and EBIF efforts are very important, and that ActiveVideo complements these efforts: ActiveVideo's client is really just a very thin key-click client; the whole point of the ActiveVideo approach is to do the heavy lifting--the video optimization, the interactive logic of a given application--on the server side, on the network side; and the client, then, is actually very collaborative. On an OCAP platform, it's really just a small Java app; and on the EBIF platform, the EBIF trigger simply fires up the ActiveVideo client. So there's really no conflict at all between ActiveVideo and the cable industry's standardization initiatives. The day when Comcast, Time Warner Cable, Cox, Cablevision, and Charter all have Java platforms that are basically the same--that will be a good day for ActiveVideo, just like it will be a good day for the interactive TV industry as a whole.

[itvt]: Now, although I know neither party is being very public about this, ActiveVideo Networks has been working closely with Time Warner Cable for some time now, and your technology has been deployed on a couple of their systems at least on a trial basis. Is the technology now mature enough to be deployed by Time Warner Cable--and other operators--on a commercial basis? And do you feel that your long relationship with Time Warner Cable will accelerate that process?

Callahan: Well, what I can say is that I would certainly hope that ActiveVideo would become a technology that's widely used by Time Warner Cable, and by other operators as well. As for whether the technology is ready to deploy and ready to scale: it is definitely ready. The core technology for ActiveVideo's server-side processing and its lightweight client are something that it and its predecessor company, ICTV, have been working on for years now. Actually, one of the things I wanted to be sure of before coming on board the company was that its core technology was robust and deployable. And this robustness is not just a result of the technology having been a long time in development: as it happens--and I probably can't talk too much about this--ActiveVideo's core technology has been deployed in a number of systems to many hundreds of thousands of users. So, thanks to these real-world deployments, the company has now had ample opportunity to find out what the problems are with its platform, and to fix them. So the ActiveVideo Networks platform is absolutely ready to go: it's been proven through multiple deployments to date, some of them quite large. I'm now looking forward to the opportunity to take it to whichever operators are interested in using it and getting it deployed in even bigger systems.

Now, I have to acknowledge that the best person to answer your question about whether the technology is ready for mass deployment would be our senior engineer and chief operating officer, Don Gordon. He's someone I'd heard about for years and years--long before I joined ActiveVideo. He was previously with Microsoft, and before that--and I think this is where I first heard of him--at DIVA, which was one of the pioneers of video-on-demand. Anyway, for the past year-and-a-half, Don has been refactoring the core technology of the former ICTV into this intelligent streaming capability of ActiveVideo Networks' platform, and he's someone I have a lot of confidence in. So I absolutely think that our technology is ready for mass deployment, and I think we enable applications that operators are going to find very compelling. And I should add that the technology is deployable by all kinds of different operators: our platform doesn't care whether it's transmitting via MPEG-2 or MPEG-4, or whether it's transmitting over IP.

[itvt]: Can you talk more about some of the things you're interested in using ActiveVideo's technology for, now that you're at the company? You mentioned earlier that your main reason for leaving Time Warner Cable and joining the company was in order to fulfill your vision for the future of television, so presumably you already have in mind some projects you want to work on...

Callahan: Absolutely. One of the things that ActiveVideo is well known for--internally they call it "stitching"--is a process for massaging the MPEG video to only deal with the portions of the application that are modified at any given moment. Basically it means you don't have to re-encode an entire 30-frame-per-second MPEG video frame by frame. You only have to stitch into the existing MPEG stream that part of the UI or that part of the application that has a modification or a change. This is much like the basic premise of MPEG itself, where the I-frame is the complete picture, and the other formats of MPEG are modifications to that baseline, and every few seconds you get a new baseline.

So anyway, the folks at ActiveVideo Networks have developed this technology that allows them to basically have very good compression--and thus very good scalability, because that's always been an issue with server-side architectures. And the way they've done this is to really only send changes in the MPEG stream where they're really necessary. Of course, this isn't a mind-boggling concept: lots of people have a good understanding of the principle behind it, and, as I mentioned, it's part of the premise of MPEG itself. In addition to this, ActiveVideo has enabled the stream with lots of other metadata, in order to render a user interface, in order to prompt for keystroke input--things like that.

So one of the things I'd like to experiment with that is enabled by these capabilities of ActiveVideo's platform is something that I call "never-ending navigation"--and, by the way, EBIF could be used to initiate the viewing session for what I have in mind. Let me explain what I'm talking about by giving you a scenario: you're watching a program you really love--for example, "The Sopranos"--and you're interested in finding out about other programs HBO has that are similar to it. Currently, in order to find that content, you'd have to bring up full-screen application that would cover most of the video and that would have a somewhat limited interface, even with the more recent guides. So, instead of that, let's have a very subtle user interface. Let's have what is essentially a navigation toolbar--with the metadata that populates that toolbar being delivered in real time with the stream that you're watching. What this means is that the content on the toolbar can be packaged in such a way that, when you hit, say, the blue key on your remote control, a nice little menu drops down and says, "Here are other episodes of 'The Sopranos'; here's other content that stars James Gandolfini; here's other content that's mob-related," and so on.

With complete flexibility, the metadata architect for HBO or any other programmer could create a database of its content that could be accessed through this toolbar. The toolbar would also have some global options, allowing the viewer who was done with HBO to go to another channel or whatever. One way to conceive of this is that it would be somewhat akin to the right-click function you have on Windows computers--which is one of the things that Windows does really well, in my opinion. When you right-click, that drops down a menu containing content and options related to the application you're currently working in. However, although some of the operations listed on that menu are unique to Photoshop or whatever application you're in, there are also more global options that are always listed regardless of the application--such as the ability to navigate through your hard drive.

Anyway, the goal here is to create an interface that, rather than always taking you away from the video you're currently watching, is much more non-intrusive and elegant, and would basically enable the viewer to navigate from stream to stream very quickly and easily--so without having to bring up the VOD application, without having to drill down through a series of tree-structured menus, and so on. Basically, it would allow the viewer to effortlessly browse, snack on content, and find the things that interest them.

Jeff Miller came up with a very good analogy for what I have in mind here: it's really not dissimilar in its concept to the experience of using YouTube--though the specific interface YouTube has would likely change for television. With YouTube, once you find the video you’re looking for, you're presented with a whole bunch of videos that are on related topics; and the typical user generally views two, three or four other videos that they would never have watched, had the service not subtly and unobtrusively guided them to those videos. The whole process is driven by affinities in the content, and not by hierarchies of menus.

So anyway, I think that ActiveVideo has a platform and a technology which--along with some of the things I've been working on with Time Warner Cable and the rest of the cable industry for the past 10 years--can blow the doors off the traditional program guidance model. We can develop a very interesting and compelling user interface for the television.

Now, the content that this user interface would present to the viewer wouldn't necessarily just be content promoted by the content owners. The interface would also feature some kind of relevance engine that would present you with other programs you might be interested in, based on your specific viewing behavior. Or it could present you with other content based on Web 2.0 scenarios--so on recommendations from community networks. For example, if you were watching "The Sopranos," and you brought up the navigation toolbar, you might find that the community had commented on the various other pieces of content it brought up, and rated them--perhaps with simple emoticons that could be entered with the remote control.

Anyway, the point of all this is to have a navigation model that puts the content front and center, that has an elegant interface, and that binds the video and navigation into an intuitive user experience, wherever the video comes from--whether it's the Internet, VOD servers, or broadcast. Those are some of the objectives I want to work on with ActiveVideo Networks, and I'm really looking forward to having the opportunity to do so.

[itvt]: Your idea for "never-ending navigation" seems somewhat similar to what other people in the industry are calling "video navigation." I know that a lot of players in the industry--GuideWorks and OpenTV, for example--are looking to develop navigation that's content-driven and avoids the classic EPG grid-format...

Callahan: You're absolutely right. I don't claim to be the first with any of these ideas: I think a lot of people are aiming in more or less the same direction, and implementing this kind of thing is something people have been talking about for some time. I've known GuideWorks' Gerard Kunkel for years now, and we've had some really exhilarating discussions about all this. Of course, what it will come down to is the specifics of each company's implementations of new, more intuitive forms of navigation.

I should also point out that the kind of navigation system I was just describing isn't by any means the only thing ActiveVideo is working on, or that I'll be working on. We'll be developing a whole bouquet of different types of services.

[itvt]: Will ActiveVideo target this "never-ending navigation" capability at cable MSO's? If so, don't the operators already have fairly well-established plans in place for next-generation TV guidance/operating systems--whether they're developing them in-house, like Comcast is with its GuideWorks unit, or with third parties, like Cox is doing with NDS? Or is this something you're thinking more of targeting at telcos and at smaller cable operators?

Callahan: "Never-ending navigation" is really a conceptual or philosophical baseline for a user experience that is video-rich, personalized, relevant, and intuitive to use without a complicated input/control device. It's not a specific product for a specific customer. As you note, cable operators have in-house development groups and are investing in their own product roadmaps--as are telephone companies. My objective is that ActiveVideo, working with these operators, will greatly add to their ability to present a variety of compelling, interactive media-streaming services to their customers via their respective navigation platforms. ActiveVideo takes advantage of something that the operators do better than almost anyone else: i.e., unicast video streaming with very high quality of service. As ActiveVideo's presence in the set- top box itself is very simple, it complements the rest of the navigation functionality that the operator develops.

[itvt]: Obviously, a lot of industry players are interested in the idea of intuitive, content-driven navigation. What do you think will differentiate ActiveVideo's implementation of this idea?

Callahan: There are a few major differentiators for ActiveVideo's platform. First, as all the complicated application logic and media rendering is done on network servers, a media-rich user experience can be developed, which is often difficult to do on set-top boxes that do not have powerful processors and lots of memory. Second, this media-rich user experience can be delivered to all devices with an MPEG-2 decoder (and this is all digital set-tops boxes deployed in the field for the past 10+ years--from the oldest SD to the newest HD DVR, and everything in between). Third, ActiveVideo's platform does not just deliver a stored MPEG-2 video asset, it locates the appropriate content--using standard URI/URL Web standards--for the application, whether it's video files, images, etc., and composites these sources in real time into a standard MPEG-2 video stream. If we combine these capabilities in the context of "never-ending navigation," we can envision a video stream containing within it the necessary metadata for rendering a compelling, personalized user experience--picture a screen with a combination of video and graphics that represents the customer's "favorites" as defined by profile or behavior--and, based on the customer's selections via the remote control, rendering a new video stream in real-time. As I was describing earlier, the metadata driving what the customer watches and how the navigation is displayed is easily enabled by the ActiveVideo platform--again to all MPEG-2 devices. This fits our "Sopranos" scenario or YouTube-like scenario very well. The fourth point is that ActiveVideo applications will be displayed just as the author designed them, no matter what set-top box they are deployed to--thanks to ActiveVideo's "Studio" which features standard AVML and DHTML technologies. This is a solution to the "lowest common denominator" problem of client-side application platforms--i.e., where an application will have to be developed to look and act differently depending on the class of set-top box it is running on (which is both technically difficult and, from the content-developer point of view, not an ideal situation).

[itvt]: What other projects are you planning to work on at ActiveVideo, in addition to "never-ending navigation"?

Callahan: I cannot go into specifics, but I can say that we are very interested in assisting the efforts to enable an advertising platform that meets the needs of operators and advertisers. ActiveVideo does not see itself developing campaign management platforms, but we will be working closely with the industry to support the relevant platforms as they emerge. The ability to execute on all set-top boxes is something that needs to be married to a business platform that will take advantage of that.

[itvt]: ActiveVideo just hired a high-profile advertising-industry figure, Todd Nisbet, to lead its interactive and targeted advertising efforts. Could you talk a little about the kinds of things you'll be working with him on?

Callahan: A key virtue of ActiveVideo is that it enables advertising to be targeted, interactive, accountable and actionable, just like on the Internet. Campaigns can be developed once and made available to viewers on any set-top box, giving advertisers the scale that they desire. We think this ability to deliver Internet-style advertising with scale will be very valuable to advertisers. Todd is working to create partnerships with some carefully chosen brands to validate the efficacy of the kind of advertising ActiveVideo enables.

[itvt]: Can you talk about how ActiveVideo's technology could be used to enable user-generated content or social TV services?

Callahan: I'm not ready to talk about all this at this stage, but I will say that ActiveVideo has the platform to bring UGC and Web 2.0 services to the operators' television platforms without need of DOCSIS, MOCA, or other emerging technologies.

[itvt]: ActiveVideo recently announced that it had enhanced its platform to address TV on multiple kinds of network-connected devices. Could you explain the significance of this announcement and how it plays into the company's strategy going forward?

Callahan: The same advantage of the ActiveVideo platform that makes it a compelling choice for thin-client set-top boxes also works well when considering the plethora of other platforms that are in the market: e.g., broadband connected devices. In general, the ability to author and deploy a compelling application--whether it is a "traditional" interactive television application or a complex Web 2.0 service--to thin-client devices is what ActiveVideo’s strategy is all about.

[itvt]: Presumably, due to the nature of ActiveVideo's technology, the company frequently finds itself having to think up ways of repurposing Web content for television. Do you have any general thoughts on the design principles for successfully effecting this kind of translation?

Callahan: I don't believe this is an issue "due to the nature of ActiveVideo's technology." Rather, I believe it's the realization, on the part of passionate and professional content producers and application developers, that the strengths and weaknesses of a large screen display, a simple remote control, and the user's expectations of application behavior all contribute to the need to design a different presentation and experience on the television--even when using the same content that is delivered via the Web. This is not unique to television; it is true for all presentation platforms--PC's, mobile phones, PDA's, etc. I personally believe the television experience is expected to be, above all else, simple. With respect to ActiveVideo's technology, it allows Web developers to repurpose a cool Web application using Web authoring tools that they are likely familiar with, and--when they get a user experience laid out to their satisfaction in all respects--they know that it will be executed with exactly that presentation on all ActiveVideo-connected platforms. Of course, this Web developer could leave the user experience exactly the same as it is on the Web: the results will not likely be as compelling to the user but the ActiveVideo technology will faithfully deliver it!

[itvt]: Now that you've joined ActiveVideo Networks, will you be moving from Denver to Silicon Valley?

Callahan: No, not at this point. I've lived in Denver for almost 20 years—other than for one year I spent in Connecticut--and my son is still at high school, so we're not going to be moving at this point. I'm not opposed to the idea of moving, but we'll be staying in Denver for the time being.

ActiveVideo Networks