Corporate Innovator: Shailesh Prakash, Chief Information Officer at the Washington Post

Since being purchased by Jeff Bezos in 2013, WaPo has made technology a bigger part of its culture

In recent years, the news space has seen a gigantic shakeup in terms of its influence and viability.

The internet has been both a blessing and a curse for traditional news outlets: while sites now have unlimited space to fill, and the ability to reach just about everyone on the planet, they have also seen their business model, which relied on people who were willing to pay for access to their content, be completely upended.

At the same time, those outlets, which had been able to set the national conversation for over a century, have also suddenly found themselves competing against smaller, niche sites as well as the proliferation of social media. The way people get their news is completely different now than it was even just five years ago. 

In order to survive, it's no longer enough for them to just be in the business of delivering news; in some respect they also have to be technology companies as well. That's not always the easiest thing to ask for centuries old companies just now adjusting to a totally new world.

It shouldn't be a surprise that the Washington Post has been able to achieve this goal in a way others, perhaps, have not. This is due to one big advantage: in 2013 the company was purchased by one of the world's most successful tech pioneers, Amazon founder Jeff Bezos. Since then, it has begun experimenting with various technologies including artificial intelligence, while also developing a multi-pronged app strategy, some of it headed by Bezos himself, to appeal to different kinds of use cases. 

I spoke to Shailesh Prakash, Chief Information Officer at the Washington Post, about how the company views the role of technology in the newsroom, some of the initiatives it is taking to engage audiences, and where this current era will eventually take us in terms of how people consume the news. 

VatorNews: Talk to me about your responsibilities and your role at the Washington Post.

Shailesh Prakash: In terms of my role, officially I am the Chief Information Officer, but I also run product and design and I also run the group that does audience development, so it’s a mixed bag.

In news companies, that alone is an hour long call of what does product mean? For example, if you ask a simple question like, ‘Is the homepage a product?’ if you go to Facebook and ask that question about, ‘Is your homepage a product?’ Mark Zuckerberg will jump up and say, ‘Of course, it’s our most important product. That’s the NewsFeed that determines what you see.’ So is our homepage a product? If you asked the New York Times or the Wall Street Journal or the Washington Post, you know, I don’t think you’ll get a crisp answer. Even if you say, ‘Indeed, it is a product,’ who owns that product? That question is even more confusing because is it news? Is it the actual product team that does it? How does video fit in there? How do we decide how many ads to show? Who makes that call? It’s a tricky question.

Given that I run product, with all its caveats, that means I have my fingers in a lot of jars. Here at the Post we take technology seriously. We truly believe that, and this is an important piece I just want to stress, we truly believe that if you look at a framework where on the y-axis is excellence in journalism, so the higher up you are on that axis, the more, in general, users and people and other journalists believe that, yes, this is brand that’s doing journalism really well. So you take that on the y-axis. And then on the x-axis, if you plot technology, such that if you on the far right on the x-axis everybody says, ‘Yup, this an Amazon or this is a Google or whatever, and they do technology and engineering and product and design really well.’ Here at the Post we want to try our very best to go up on that upper right quadrant, where we are excellent in journalism but we are also excellent in technology. And I submit to you that there are many, many brands that we all could agree do journalism really well, and I think we’re all agreed that there are many companies and brands that do engineering very well, but there aren’t too many that seem to be doing both very well. And I also submit to you that it’s easier, in my opinion, for brands like us, like the Wall Street Journal, like the New York Times, like the Washington Post, who do a really good job in journalism, to become excellent in product and design technology than it is for those technology brands to become excellent in journalism.

You can see that playing out in the market. Facebook, the engineers there are the top notch engineers, but you see how they struggle with, ‘Are you platform or are you not? How do you verify your sources? Is this fake news? What’s going on here?’ Google’s taken quite a few wacks at, ‘How do we do it so that when you search in Google you don’t see all the conspiracy stuff showing up right on top?’ Sometime back, there was this accident here in D.C. where the Republican senators and Republican congressmen were on a train and that train happened to have an accident. An hour after that accident I remember in Google one of the top stories was this absolute conspiracy theory that the deep state engineered that failure. How does that happen? It would be like us putting something completely fake on our homepage. So I think the engineering companies struggle with how to do sourcing and all the stuff that journalists figured out over 150 years on how to use their intuition and their sources and develop a nose for figuring what’s right and what’s wrong. It’s like figuring out fine wine, how do you do that? You need that DNA but, unfortunately, in that process, we got left behind in the product and technology side, and I believe that in the last, I don’t know, five years, seven years, especially after Jeff bought us, that we’ve made some decent progress on moving along the x-axis.

I’m sorry for the long answer but that, in a nutshell, is my role, to basically try to move us to that upper right quadrant for excellence in both.

VN: What are some of the things you’ve done with the technology to make that happen?

SP: First and foremost, the one thing that engineering needs to do is to truly shift from more of an IT, help desk, back office, ‘it’s not me, it’s the vendor’ mindset to really owning the technology stack. So we’ve shifted out mindset from being an IT driven culture to much more of a product engineering culture, where we roll up our sleeves and we are not afraid to build stuff. We are not afraid to sit in the newsroom, truly understand what the process of producing content is all about and then feel the pain that our people in the newsroom were feeling and not then go out in the market and put an RFP and get vendors to tell us what they have and then select it and babysit the system but really to roll up our sleeves and build it and not be afraid to do that.

We started with the biggest issues that we saw; simple things like, ‘I publish my content but then I go to my browser and look for it and it’s not there, but, oh, it’s there again.’ Intermittent caching problems like that, where our newsroom was frustrated and concerned that, 'My story’s out there but I don’t see it, sometimes I do, what’s going on?' Taking a much more stronger view of metrics, where you have the ability to look at not just simple metrics like UVs and PVs but looking deeper: what is the type of content that makes people subscribe? What happens once somebody becomes a subscriber? How do their behaviors change? How can you do it such that you do differentiated pricing? Is it really true that people in India are willing to pay the same amount as somebody in McLean, Virginia or in Manhattan? It doesn’t seem to be obvious but how can we find and build metrics that show what would happen?

We have put together a lot of frameworks that allow us to do testing, and by testing I mean we have tools where the newsroom can change headlines or change images, change images and headlines, the video team can choose to put different thumbnails, or even an animated GIF. They just make those decisions on which are all the different headlines, and the system decides which one should get more and more traffic. So, let’s say you have two stories, each have different images and headlines, and in the beginning both get 50 percent each, and within 30 minutes, depending on which story is doing better, and we can argue about what better means, whichever is doing better, that’s the one that then gets the all the traffic and starves the other one. Now, later on, the newsroom can come back and do an update to that story and, again, have a different image and headline and the test begins again. But it’s the system that takes care of it.

We tried the same thing on the subscription side. Do people, ultimately, end up buying our subscription package if they see something like, ‘Do this for the good of journalism’? Or do they convert better if they see a marketing paywall that says, ‘Just for you today, only 99 cents’? Which one does better? We don’t wait for months to decide that, it’s played out in parallel. Sometimes we have over 10 such things in parallel and the system decides which one should get more and more traffic.

We’ve gotten quite aggressive in being experimental. One of the things Jeff talks to us quite a lot about is: always ask is this a one way door. If this is not a one way door then why are you afraid to go through it, because you can always come back. We’ve taken that culture to heart but there is an engineering framework that’s needed for that culture to flourish, so at any one time we are running 10, 15, sometimes 25 to 30 experiments with some small percentage of our traffic. Different UX designs, different headlines, different images, different page speeds, different ad loads, different types of experiments with ad blockers. We’ve shifted quite a bit to becoming a fairly aggressive experimenter and using those experiments to make data driven decisions. That has, in my opinion, helped us to become much more coherent in terms of what we are doing, as opposed to, ‘We are going to go back into code mines, develop this for six months and then put it out and then hope that it works.’ We have changed quite a bit into, ‘Let’s put out some stuff quickly and then we’ll see how users react to it.’”

VN: Is there any kind of machine learning or AI that goes into how the system choose which headlines and images work best? How are you using those technologies?

SP: That’s a controversial topic and I should be clear about what I’m saying. So far, what we’ve talked about, there isn’t much AI in this. This tool that we have, it’s called Bandito, that decides which image and headline combination should win, in that scenario it’s the humans who have put all the variants in place. So they put variant 1, variant 2, variant 3, it’s created by humans. The only place where the machine comes in is that the machine decides, based on various parameters that you’ve set up upfront, like click through, for example, or scroll depth or however you want to define it, which one of these variants should win. So it’s not the machine creating the variants itself, it’s only deciding which should win based on click through or whatever. So, really, there’s no AI; there’s some fuzzy logic involved in there, but it won’t pass the sniff test of AI.

But we are fiddling in AI and I’ll tell you where we have released some products into production that are AI based.

One is this thing that we call Modbot, and it’s basically trying to help with the moderation of comments. The way we do comment moderation is that we get to the tune of millions of comments, and you know the deal. I mean look at the comments in any large news site. There are some that are basically very well worth reading, they are bringing forth a different viewpoint, etc, but there’s also a lot of just junk. ‘You are an idiot!’ ‘No, you are an idiot!’ You know, come on. And the way most of us do these things is we take the bulk of it and either have an in-house staff, much like the technology companies, or we send it off to some third party and we pay them by the amount that they have to moderate. In our case, we were doing a little bit of moderation on our own but the vast majority was being sent to a third party and then we were paying them, essentially, on the amount of work that they had to do, the number of comments they had to look at. So we built this technology called Modbot, which, essentially, tries to chop off both edges. It will figure out what are obviously the ones that you don’t want and what are obviously the ones that are fine, and then we’ve been working on the AI of it to narrow the zone which has to actually be sent to humans to look at. In the beginning we were able to lop of the top 10 percent each, the obviously bad 10 percent, the obviously good 10 percent, and we had to still send about 80 percent, which itself was a gain. But now we’ve made significant progress where at this point we’re sending about 30 to 40 percent to humans to look at, so about 60 percent the machine takes care of. There it’s tricky; the obvious ones where you’re using the f-bomb and this and that, it’s easy to get rid of. But threats, stuff where you might be disguising what you are trying to say but the tone is obvious that it’s threat, to a human this is somewhat easy to detect but to a machine this requires a learning algorithm.

Another example is something we call Heliograph. That is truly trying to generate content. We have worked with that mostly on structured data, so we first did it for the Olympics because, ‘India played Pakistan in field hockey. In the 33rd minute a goal was scored, this was the first time that he scored in over six years.’ This is something that you can determine from the data itself, you don’t need a human to go in and say what I just said. So structured data, like Olympics, like elections, crime statistics, earnings reports, financial markets, these are types of content that a machine can do a decent job in creating content. And what does that do? That frees up our reporters to really look at the more high end type of stuff as opposed to working on the lower end of where the traffic might be. But the thing you have to realize is that even though you might not get high traffic for that India/Pakistan hockey match, you will still get some if you have that page. Google will pick it up, you can tweet is out, you can put it on Alexa, the machine does all this to push out the content with the correct SEO and then the consumption is a long tail. So, yes, your big stories drive a lot of your traffic but there is a long tail of other pages that, when added up, might drive even more than your top 10 stories. So the more you are able to use the machine to put out content, the more you are able to get digital traffic, and then, of course, that means you get more ad revenue and that could potentially drive further subscriptions.

Heliograph has been very good for us in structured areas like the Olympics, like I said. We used it a lot during the elections; our goal was to pretty much put out a page for every precinct in the United States. We now doing it such that every high school football story in the Maryland/Virginia/D.C. area has a report about it. We are doing it for player cards, in terms of statistics for various players in professional sports. And we are looking at it for finance stories, we are looking at it for real estate, we are looking at it for crime statistics. I’m quite optimistic about using AI in that world.

Then the third piece where we are using AI is the classic recommendation engine. You go to Amazon and you see, ‘People who looked at this also looked at this, people who bought this also bought this. Here are recommended products for you.’ They’ve gotten really good at it. The difference between doing recommendations for products and doing recommendations for news items is that the news item recommendation is a harder problem than products. The reason for that is the news changes so often. You might be really interested in Brazil but that’s only because that’s where the World Cup was going on in soccer, and then you move on. You might be really interested in New Jersey because Chris Christie was making a lot of noise and you were really focused on what he was saying but once he dropped out of the limelight are you still interested in that? So, news changes so dramatically and often the core relations are hard to figure out that a good recommendation engine for news is very, very difficult. We’ve been playing with some secret sauce internally to use AI to try and figure out what is a good news recommendation engine and I’m happy with some of the success in that area.

Those are the really the three main areas we are playing around with AI. We are looking at headline generation, I think that’s a little bit of a ways out. We are also looking at summarization, which I’m quite excited about. I hope that we will soon have some interesting automatic video generation, for example, where you look at an article, you pull out excerpts from that article that you think are good summaries and then you go and find photographs that match those summaries and you stitch it together for a slideshow. Because often times people want to watch a video but they’re at work, let’s say, they can’t turn on the sound or maybe they’re watching it on their mobile phones and they’re in a public space, like say a train, they don’t want to turn on the sound. Facebook has done a great job of doing autoplaying videos with text. I think we can use AI to do a good job of autogenerating slide shows with text, where both the images and the text is pulled out and matched by a machine. You can think about how much scale we would get with that type of approach, rather than a human trying sift through articles and choosing videos for it.

Those are some areas where we are truly playing around with what I would call learning algorithms.

VN: With Heliograph, what percentage of the content for the Washington Post is now being done by a machine?

SP: It’s small. I would not put it at more than 5 percent right now.

VN: When it comes to using AI for recommendation, how do you make sure you’re giving users a variety of news? You go onto Netflix and it shows you things that it thinks you already like, so how do you make sure you’re not only showing your readers what they already like instead of expanding their horizons?

SP: That’s a very, very good question, and I worry about that.

There are a few ways to try and influence that type of broadening of your horizons. And this is not AI but we are doing it, we call it Counterpoint. What we do is, in our opinions articles, which, of course, by definition are opinionated, we have started putting in the recommendation engine, not through AI, but basically with the rule that you will see a counterpoint to that story. So if you have a very left leaning view on who is responsible for the economy doing so well, we, in the recommendation engine, at the bottom of that opinion story, if we do have a right leaning view it will get inserted in the second slot. Not in the first slot but in the second slot.

We are also experimenting right now with what would happen if we took that recommendation, which typically shows up at the end of the story, and put it towards the middle of the story. We have a lot of pull out quotes and tweets and so on in the story, one of those slots would instead have this so-called counterpoint to give you a wider lens into what’s being talked about. And we’ll see if indeed this turns out to be something that works and people click actually through. Early results for that thing that I was talk about down at the recommendation engine, the A/B test is doing well, hence we are exploring putting it up higher. If it continues to do well, I don’t see why we would not be able to expand something like that, not just to our opinion pieces but also to other narratives that are not necessarily opinion related.

VN: What percent of readers read the Washington Post on mobile vs on desktop? What percent of readers read on the mobile web vs an app?

SP: We are now at about 70-30. 70 is mobile, 30 is desktop. And of that 70 percent that’s mobile, about 92 percent of that is mobile web and 8 percent is apps. But, there are several buts here. One 'but' is that in the mobile web versus app scenario, the engagement in apps, or time spent, is significantly higher than mobile web, and that makes sense. When you get an app you have come to the App Store and specifically download the Washington Post app, whereas in mobile web you come from everywhere, social links, links in other sites. So the mobile app, although a small percentage of unique visitors, is a very, very large driver of time spent. The other thing that’s important and interesting to look at is if you look our subscribers, the app generates a significantly higher percentage of subscribers than the mobile web. Again, because I think it’s a little bit of a self selected audience. You took the pains to go to an App Store, download our app, and therefore the chance that you’ll become a subscriber is much higher the random fly by night operator that lands on mobile web.

VN: Since apps are such a big driver of engagement, what initiatives have you put in place to get more people to download it?

SP: To download the app? Well, we’ve done a few things. We’ve had discussions internally about the so-called ‘door slam method.’ Some properties, like LinkedIn, were using it heavily sometime back; every time you click a LinkedIn link it says, ‘Hey! Download the app! Download the app!’ and it slams the door on your face (laughs). ‘You need to download the app!’ We, of course, have not gone that route. We’ve also stayed away from deep linking, which is another way to force downloads, sort of. It’s not slamming the door in your face but let’s say you have the app and you click on a link, then, instead of taking you to the browser, we say, ‘Oh, let me take you into the app.’ We’ve stayed away from that too because, at least in our opinion, while that will certainly increase the usage of our apps, it’s also not something that you asked for. We are forcing it on you.

We have rather taken the approach of investing a lot in our apps and, in fact, we have three apps. Our strategy really has been: don’t force a one sized fits model on our apps, there are different types of users and habits so let’s try to serve a much broader set of them.

One we call the ‘classic app.’ If you go to the App Store it’s the one with the black icon and that’s a more traditional app. It looks like the New York Times app, it looks like the Wall Street Journal app, it’s a very traditional news app. We’ve spent a lot of time and effort to make it fast and beautiful, but it’s traditional looking, hence the word ‘classic’ app.

Then we have another app, which internally we call Rainbow, this is something that, in fact, Jeff helped us design. He was the chief product officer of the rainbow app, and we had an interesting time working with him to design it. At the end of the day, that is much more a very visual app, which, essentially, has a bundle of about 200 stories. So it’s not everything in the Post, it’s about 200 to, at times, 300 stories. It’s addition based, in the sense that at 5 a.m. in the morning the morning edition is downloaded, and at 5 p.m. in the evening another edition is downloaded. It’s highly visual, so as opposed to the more classic headline and image, this has a big image with the headline written on top of it. There isn’t a homepage; you open the app, and boom! You land on two stories. In fact Jeff’s suggestion when we started this was that you should land on one single story. During the course of product development we did not put the homepage but instead of just one story, we land you basically in two stories. Big image, the headline is burned into the image, it’s a flip, flip, flip, you keep going through two stories at a time. There’s the top news section, which has about 20 stories, and then you flip over to the politics to opinions to the sports. We have a section called Backstory. Very different navigation, very different style, and targeted much more to a user who may not be as much of a news junkie as the ones who consume on the classic app.

Then we have a third app which is the other extreme. It’s basically our e-paper. It’s the newspaper itself, and it’s a beautiful experience if you haven’t tried it. It’s like reading the paper, especially on an iPad Pro, it’s a beautiful experience. We’ve taken pains that if you were actually reading the newspaper, today’s newspaper, and you actually touched a story on that newspaper, it would take you to the Web version of that story, so therefore you can enjoy a video or enjoy a photo gallery or leave comments or what have you.

So three very different apps for different audiences, and you can guess the amount of time spent on each. The print app has the maximum time spent, the classic app has the second, and the Rainbow app is third in terms of time spent. The Rainbow app also happens to be in built into every Kindle Fire tablet, so every Kindle Fire tablet that you buy comes with the Rainbow apps pre-installed and it’s one of the most popular apps on that platform. So it drives a lot of usage in our numbers for the apps side of things.

To go back to answer your question about what are you trying to do to increase usage for your apps, we have spent time and effort to build not one but three app so that we cover your tastes and not just give you, ‘Here’s our once size fits all thing.’”

VN: Talk to me about Arc Publishing. Where did the idea for that platform come from? What kind of tools does Arc Publishing provide?

SP: We’ve got a nice site called ArcPublishing.com and if you tap on the products tab on ArcPublishing.com it actually lists all the tools and products we supply as part of Arc, because Arc, as the name implies, is a suite. It’s an arc that goes all the way from content creation to editing to scheduling and planning to actually rendering and distributing to distributed platforms like Apple News, Facebook and Snapchat. So it spans the arc of publishing needs.

How did we come up it? I wish I could tell you that this was part of a grand plan and I put together a business plan and figured out the road map (laughs). That’s not true. Like I said, in the beginning, we just simply started taking ownership of our tech stack and building out the areas where we thought there were lots of problems in publishing in the Washington Post. So we started by trying to solve caching problems, and then we said, ‘Ok, video’s a big area of investment for us, why don’t we try and figure out our own player?’ and soon after that we said, ‘Well, we need the CMS to do the video system so let’s do that,’ and then we went, ‘Ok, subscriptions is a big area of focus, rather than go out and put and RFP what happens if we put our own CMS for that?' Over time, as you will see if you go and look at the products on Arc, we have a wide variety of tools that can work independently; if you are an Arc customer you don’t have to buy the whole suite of tools. You may choose to use your own editor, like WordPress for example, but use our video system and have us build your site. Or you could say, ‘No, I would like your editor (we call it Ellipsis) and that would work as well,’ so it’s a little bit of a mix and match scenario.

Now we power almost 90 sites out there with Arc technology, big known brands. Like the LA Times is completely 100 percent their app and their site on Arc technology. The New York Daily News just went to 100 percent of their traffic. The Chicago Tribune will go on 20 percent of their traffic on Arc with a plan to get to 100 percent in the next few weeks. Toronto Global Mail, their app and video and site is all on Arc systems. The third largest Spanish speaking news site in the world, runs on Arc end to end: site, app, everything. The Boston Globe, we’ve started working on them, they signed an agreement. The Philadelphia Inquirer is working with us. NVME, the largest media company in New Zealand; La Parisian, a big paper in Paris; and there are many more that I can’t tell you about yes because we have some confidentially agreements but by the end of year there will be about 150 to 200 sites and apps running on our technology. Of course, there are the tools that help them create the content to put it on those sites and apps.

So we are very happy with the evolution of Arc. I hope that this becomes a big source of revenue and profits to fund our journalism. That is my goal, to find us a third leg of revenue. Today it’s our advertising side and our subscription side. If I can now bring in substantial, highly profitable source of revenue from Arc to fund the great journalism that Marty Baron and the newsroom does, I would think it would have been a very successful effort.

VN: How have you seen the digital news experience change in the last few years?

SP: The biggest change, and the most, I would say, pleasant and happy thing that has happened in the news industry is I think that people are paying for quality content. It’s so interesting to watch what happens when somebody becomes a subscriber. They begin to pay you, that’s great, the relationship becomes much stronger with that reader, but then you find that they also begin to consume a much wider variety of content than they were when they became a subscriber. It’s like you go to a buffet, and in the beginning you’re only going for the steak or the lobster, but once you become a paying member you also start trying out all the desserts, and you might try the salad and you diversity of content consumed becomes much broader and, of course, you consume much more as well.

So that is really good for us because, like we were talking about, a place like the Post, of course we have a lot of politics and opinions content, but we have movie reviews and we have style and we have food, and there’s so much diversity of our content that we are very pleased that’s being exposed to a much larger audience because of our massive growth in subscriptions. And then, of course, it’s really good for ad revenue too because the more page views you consume, we get a good, healthy stream of ad revenue too. So that’s been the biggest change and whether that’s because things like Netflix and Amazon Prime have made consumers willing to pay for service, or whether it’s just the people realizing that quality content is not free, or whether it’s just more and more news organization tightening their paywalls or becoming paid, I don’t know that it’s any one thing, maybe a combination, but it has been a pleasant surprise and I don’t think that that’s going to change in the near future. If anything, we see that trend accelerating. That’s been the big ‘ah ha’ in the last few years.

VN: Where do you think digital news will be going in the next five years?

SP: That’s a very good question. I had a boss who used to say, ‘It’s not that hard to predict the future, the difficultly is predicting when it will arrive.’ And I think that’s true (laughs).

I’m sure that AI will become more and more dominant. Let’s just put news aside for a second, I happen to be in the camp that in the next 25 years the world will be fundamentally different. And I don’t just mean compare now to 25 years ago when there was no Internet, no iPhone. I mean fundamentally different. Whether it is just the proliferation of self driving cars, the automation of medicine, things like humans having bionic eyes and stuff that cures blindness because you do a direct link to the brain from your glasses. Things like that, I think, will be here, so it will be dramatically different. I happen to be in that camp.

With that backdrop, if you start looking at what might happen in the news world, I fundamentally think that if you watch Harry Potter, the newspaper that’s absolutely live and has an assistant that can help you understand what’s going on, is not impossible. I don’t know if it’s five years, but I wouldn’t be surprised if you have a fantastic, light piece of plastic that’s live and understands you and has a conversation with you about news I don’t think is ridiculous with the advent of materials, technology and artificial intelligence.

The second thing I think is going to be very, very interesting is the whole voice based thing that’s going on with Alexa and Google Home and so on because what that opens us up to is a two-way conversation. Right now Alexa is still one way. ‘Alexa, what’s in the news?’ ‘Blah blah blah blah.’ But it can understand what I’m saying. So let’s say there’s a spelling bee and it’s telling me this was the word in the spelling bee and so and so won it and this is the third time she won it and whatever. It could say, ‘Would you like to try the word?’ And you say, ‘Ok, metamorphosis. M-E-T-A...’ and Alexa says, ‘Not bad, you got it wrong but you are in the 35th percentile,’ or whatever it is. I think that there is a change coming in the next four to five years where these voice-based assistants will give you bidirectional news. We are investing a lot in how we can get ahead of the curve in voice-based stuff. Of course, podcasts are the rage now, but, again, podcasts are one way, it’s more like radio than really the new voice-based assistants that can actually understand what you’re saying.

The third piece, which I’m not sure about but I do think has a play in this world, is all the stuff that’s going on with Oculus and the virtual reality/augmented reality space. At the end of the day, just like when you are playing a video game, when you are consuming news, what is the storyteller trying to do? When you write this piece, you will try to immerse your readers in your point of view and your storyline. If you can can have the experience where that reader is truly immersed, as opposed to just reading the stuff, I think it will be much closer to what your intent is than what is possible today. I do think that that’s something that’s there in the next few years, maybe five years. Then the other stuff, the Harry Potter type of thing, that I think is a little ways out (laughs), I don’t know how far but I think it will come. I tell my kids this all the time. My daughter is 15 and my son is 9, I tell them that the world you will live in will be so, so different than what growing up in. It will be so, so different.

(The Meet the Corporate Innovator series is brought to you by Advsr, a startup advisory firm in the business of starting conversations and sparking big ideas.)