A series of interviews with innovators operating at the intersection of consumer behavior and business transformation: Rob Glaser, RealNetworks founder and CEO.
Bruce Rogers: Is it correct to call this the reinvention of RealNetworks or the evolution?
Rob Glaser: I think that’s a fair thing to ask. I ran the company for 15 years and I was away for about 3 years and then the board asked me to come back. And I’m now in the second phase of the reinvention of the company. The first phase was moderately successful, but we didn’t hit any grand slams. I would say we had some singles and doubles.
When we started with streaming, it was just at the dawn of the commercial internet. The internet was narrowband going to broadband. We focused aggressively on the public internet as a commercial phenomenon, built products that were adequate with narrow band but world class in broadband. At that time, everything we were trying to do had the wind at its back. So we were focusing on fundamental trends that were going to unfurl and the only question was how fast they were going to unfurl.
Another question was what existing incumbents would end up being powerful competitors. But it turned out for us and a lot of first generation internet companies that the internet was so different than what had come before and was such a huge opportunity relative to what had come before that it was a great opportunity for new entrants to come in and have a huge impact.
We had a great run at that. By the time I came back in the early middle part of this decade, most of Real’s core businesses were not fully aligned with the right type of vision I would say. We were no longer on the leading edge of where the world was going.
Rogers: Take us through the next phases of the business.
Glaser: The first effort to reinvent the company was through a product focusing on personal audio, video and personal photos called RealTimes. The idea was to leverage our expertise in photos and video, but instead of focusing on professional content which had been our forte, RealTimes focused primarily on personal content.
RealTimes provided tools and products that would enable people to take their own personal photos and videos and propagate them in lots of better ways than they had been. That particular product was not a consumer success. Not because it wasn’t a good product. It got great reviews. We were not very successful with the product on a consumer basis because we had a premium model at a time where everything was going on consumer side either free and ad supported or built in from the platform provider or some combination of the two.
But we did have good success with it as an OEM product to mobile operators. We got Vodafone to pick it up. We got Verizon to pick it up. We got KDDI in Japan. We got Telefonica in Latin America, and a few other carriers too. So we had carriers on four or five continents around the world using the product, with end users interacting with the product with over 20 million faces.
We didn’t build RealTimes to create a data asset, but it turned out we did just that, with a training set that was very, very good for computer vision and face recognition.
Rogers: How long have you been focusing on facial recognition?
Glaser: Three years ago our CTO came to me and said, “Hey. I think we can build a world class facial recognition platform.” Some of our customers that we had in these carrier channels wanting to do facial recognition. We did not have deep expertise in it but we did have a team that had done some work and suggested that they could do a world class job of it. So I gave Reza, our CTO, about nine months. He gave me a road map on how good an algorithm we could create using our very talented team and the excellent training set that we had.
We managed to build a world class team and platform leveraging this unique global data set of faces. And after about four or five months we had achieved our nine month goals in accuracy. So the team overachieved and convinced me. And then we ended up convincing everyone around here that we had a world class product and technology. We thought we might be getting a little bit early relative to when the commercial market would be ready. But my life experience is when you’re entering a market it’s always better to enter a little early than a little late because then you’re not chasing people. You’re basically inventing the market.
Rogers: How did you productize the technology?
Glaser: We decided to go forward with what ended up being called SAFR. And we had a general sense of what it was great at. We were focusing on not just still images but video. A lot of the facial recognition products were good at stills but they’d fall apart with video because of the sheer volume of data. Ours was designed for video from the get go.
Because the data set was very diverse we have one of the most consistent and bias-free Facial Recognition systems on the market. It turns out a lot of our competitors run into a problem where they’re really good at one demographic or one ethnicity and they’re not nearly as good or nearly as accurate with other subgroups. Because we built SAFR with a geographically diverse data set that spanned all the major ethnicities in the world, being bias-free is one of our competitive advantages.
By the end of 2017 we were getting ready to come to market. We were looking for use cases. And my kids’ school had just put a gate in. The school is a pre-K through fifth grade school in an urban part of Seattle.
And one day in the fall of 2017 I took my kids to school as I did most mornings. And I asked them so how does this new gate work. And they had a gate and a camera and it was all set up and they were just about to send out the note to all the parents because it was just about to become operational.
I said "well, I have some technology for you. You want me to give it to you?" And they said “sure.” And so we helped them deploy what became the first school use of SAFR. And they loved it. No human intervention required. They quickly became so comfortable that it never had false positives, which was part of how we designed SAFR, so for their community of 400 or so people using the system SAFR just worked.
Rogers: Sadly, it seemed the need for the security functionality has never been greater.
Glaser: Society has started to have a much more intense conversation than even previous ones about what can we do around school safety. And we’re sitting on a solution that the parents love, the administrators love. It’s not a panacea, but it definitely enhances security. And it does so without crossing into the sensitive and divisive debate about gun laws.
At this point we were getting ready to launch SAFR and decided to do something good for society that will help put the word out of what we’re doing which is let’s make a version of this software available for free for every K through 12 school in the country.
Rogers: Do people have privacy concerns about facial recognition?
Glaser: Yes, some people do. To address them, we had a very intentional goal of finding a socially positive use case for facial recognition. Because from my experience in doing streaming whenever you get out there in a new market people want to wrap around their minds what is it good for. In the case of streaming we now had all these great uses where you can watch, listen to and later watch shows from all over the world. Anything can be broadcast.
In the case of facial recognition on the other hand, we observed that people had sort of mixed feelings about it. Some people thought it was incredibly cool technology and that’s sort of how they processed it. And it is. It’s amazing how well it works.
On the other hand, some people were worried about potential dystopian risks that personal privacy would be impinged. When we launched SAFR we were explicitly talking about both socially positive use cases and also some of these societal questions head on. We weren’t hiding it from them.
Rogers: SAFR just launched?
Glaser: We’re now in the second phase of launch. The first phase which was last fall was when we launched the platform. Then this past week we launched our first broad scale commercial vertical. We have the educational vertical which we’re still committed to. But the category of use cases that we’re focused on for this next phase of commercial launch is in the security domain. And so last week, we announced our SAFR for Security platform. And we announced specifically the integration of the SAFR platform with about half a dozen of these video management software platforms that are used typically for security professionals who are monitoring dozens if not hundreds of cameras.
The vendors of these traditional systems don’t have expertise in facial recognition technology. So almost all of them have opened up their platforms for partners to integrate facial recognition technology. SAFR is integrated with at this point about half a dozen of the major video management systems. Now anybody that’s interested in deploying facial recognition in a security application has an off-the-shelf solution
Rogers: Do you see a place for RealNetworks and SAFR to play in the personal ID market?
Glaser: Absolutely. In fact, that’s one of the use cases that we have built in our offices; I use SAFR to get into my office every day. We have not announced that as a commercial product yet because there are integrations to do and a few features to add. But that use case around conditional access we think is going to be a very big one.
Rogers: Are there trade-offs for security and civil liberties?
Glaser: We have to be mindful of civil liberties. As a card carrying member of the ACLU, I’m glad that that’s part of the discussion. You can’t get on a plane without everybody knowing exactly who you are and we’re all ok with that. So is that ultimately going to be true for other things besides commercial air travel? Maybe. I was struck by the following. I’ve been involved with major league baseball as a small minority owner of the Seattle Mariners for over 25 years. The Mariners opened their season in Japan this year. And one of the things I loved about baseball games in Japan is that they don’t have metal detectors. It’s a society where there’s a high level of societal trust. In the U.S. we’ve gone away from that. Everyone has to go through a metal detector now. How nice would it be to just have it be like it is in Japan and like it used to be in sports in the US until about ten years ago where you just walk into a building and you’re a trusted person.
Rogers: It’s obviously early days, but how is this new platform gaining market traction?
Glaser: We’re seeing tremendous interest. It’s probably the largest and most diverse and robust commercial pipeline I’ve seen since the early days of streaming in the mid ‘90s.
Rogers: What gets you excited about this? You could walk off and either spend time on your foundation or your other pursuits. So why this? What gets you up in the morning?
Glaser: I’m too old to play second base for the Mariners. So we all have things that are in the sweet spot of things we’re interested in and things we’re good at. And I love learning and I love learning about new technologies and new ways to deploy those technologies. Plus, I want my kids to understand that we need to work hard and you apply yourself and you get lucky in some cases, but you also work for that luck, that you can do things that have an impact.
Rogers: Thank you.
