Evolving technology is like children splashing in puddles: interview with Scott Jenson
Welcome back! This is Wojtek Borowicz from Estimote, happy to present you with another entry in our series of interviews about technology, Internet of Things, and contextual computing.
Just before Google I/O, which brought us several exciting announcement about Android’s push for context, I talked to Scott Jenson: the mastermind behind Google’s Physical Web. We talked about the web making a comeback in the race against native apps and whether this race should even be a thing. We also discussed the problems plaguing consumer IoT. And since Scott is one of the most experienced UX specialist out there, in the business since before the term UX was even in use, I couldn’t miss the opportunity to ask about his thoughts on the latest gold rush in interfaces: the conversational UIs. Enjoy!
Wojtek Borowicz: You lead the Physical Web project at Google. What’s the big vision behind it?
Scott Jenson: As we’re getting more connectivity and more smart devices, more things are going to be capable of talking and interacting with us. But today you have to install an application and go through a multi-step process. We’re trying to reduce that friction. We feel you should be able to talk to something with just a couple of taps. Our catchphrase is: walk up and use anything.
The app marketplace is so saturated people are barely installing new apps. With Physical Web reducing that friction, do you believe we’re nearing the end of the app economy?
That is too strong a way to put it. There are clear advantages to applications and applications are not going to go away. What we’re seeing is that for small applications, the things that you only want to use once or twice, friction is getting in the way. Physical Web isn’t about killing applications, it’s about providing lightweight alternative where an application doesn’t make any sense.
It’s interesting because for past couple of years we’ve often seen people claiming native apps will trump web. Are we going to see mobile web bounce back in popularity and exist alongside native experience?
Many people like to see mobile as web versus native horse race where one must win. Over the last couple of years there was a strong message that web is dead and apps are going to be the only solution. Now we see that both have their pros and cons and that the web is coming back. It’s being appreciated for what it is and it’s coming back in a much, much better way.
Physical Web lives on top of Bluetooth beacons, technology people were hyping up a lot three years ago when it came around. Now we see much more conflicted views, from claims that beacons have matured and are providing a lot of value, to calling them out as a fad. What’s your take?
I always go back to the Gartner Hype Cycle. Every technology goes through this. There’s going to be the peak of inflated expectations, because whenever a new technology comes along, people misunderstand it, overuse it, and overpromise it. And then you go through the trough of disillusionment. And then things go back and they actually build properly. That people were expecting beacons to be this mystical cure-all is totally understandable.
Physical Web is a second generation for beacons. We understand that the only thing to use beacons for is NOT to track user walking through your store but instead to offer user value and interaction on demand. The Physical Web is representing a much more mature and consumer-focused type of solution that will have more value in the long run.
So is Physical Web a way for beacons to climb onto the plateau of productivity?
Exactly! Every time you mention beacons, people jump to precise location. They want millimeter accuracy and that’s just not practical given the physics of the 2.4 GHz spectrum. Making all these magical things happen just isn’t that simple. Physical Web is much simpler and more consumer-focused, so the user has to be proactive. I can’t tell you how many people asked for the Physical Web to buzz in the user’s pocket. And we’re adamant that will not happen because we want users to be in control. Even though it makes the process one step longer, because the user has to actually ask, it also empowers the user.
It makes them feel they get the functionality they wanted. That type of user enablement is critical for the long term acceptance of the product.
Physical Web’s potential seems huge for lightweight, contextual experiences. But Google has only integrated it with Chrome. Why not move forward with Now or Maps? (note: at Google I/O it was confirmed that Android N will support Nearby Notifications for Physical Web on the OS level)
I think that as Physical Web develops, it will be integrated more deeply into other operating systems. It’s likely to happen over time and integrating it into Chrome was an obvious place for us to get started, to get things moving, and get people excited.
You’ve opted for a consumer-first approach of Physical Web. And that’s funny, because when I look at the state of consumer Internet of Things, I can’t help but think something went wrong. We’ve started disguising Wi-Fi modules slapped on home appliances as innovation. Would you agree with that?
The issue with consumer IoT is people got excited by the raw potential and extrapolated too quickly. And now there are many problems that need to be solved. On my personal blog I talked about this quite a bit. Other people are talking about this, too.
Thington is a company in the IoT space and they have a lovely example. They say that when the user says turn my lights on when I get home, what they usually mean is… and there follows a detailed list of lights, a time of day at which it should happen, additional conditions for when it’s sunny or dark outside, and oh, by the way, don’t do any of this if someone is home because it will probably disturb them. So there’s a huge disconnect between what we think we want and what we really want.
There’s certain kind of depth and precision required in consumer IoT. Now what’s happening with, for example, Wi-Fi enabled water heaters, is perfectly understandable. They’re basically doing what Nest did and apply it to a water heater. This is a necessary place to mature, grow, and experiment. But to coordinate everything will require a lot more work.
Other problem you mentioned on your blog is that many IoT manufacturers assume people live very structured, almost algorithmic lives. Are there any devices showing a better understanding of human interaction with technology?
The argument I made was that we’re in love with automation and we’d love for things to happen completely automatically. I’m more a fan of Douglas Engelbart, who talked about augmentation. I would much rather have a series of macros in my house that I invoke or get offered to be invoked for me.
In the lighting example, I might have to specify the lights that turn on when I open my front door. Then when I actually get home we can experiment. When I open the door the system can flash my watch and offer to turn on the lights. We could have a conversational UI that says Welcome home Scott, do you want me to turn your lights on? All of these things can layer on top of each other.
I like the idea of the macro approach which could be automated, but could also be offered, like a domino that I push. I haven’t seen that done yet, but it strikes me as a more mature and understandable model.
Isn’t there even a single consumer IoT product you could single out as doing an exceptional job with UX?
It’s an extremely difficult problem and it’s gonna take us a while to figure out. And we still need more infrastructure, like shared communication protocols and standards, for this to really take off. I don’t wish to be critical of anyone, I just think we’re a year or two early.
Many people seem to think that shared protocols are the holy grail of IoT, that they will solve all problems: an approach you once called out as lazy and old-fashioned. Why?
Because it is not enough. It’s kind of like saying if we figure out TCP/IP, every application in the world will work perfectly. Yes, it is a very good first step. If we figure out how devices can be on the same bus, that’s awesome. Now let’s talk about what we’re sending over that bus. What happens if the internet goes down? Local control versus global control? Where the data is stored? There is so many other problems that we have to work our way up the stack. We’re so preoccupied with the plumbing that we forget that we’re building a house.
All in all, why do we need smart objects? I often find myself among friends not following the latest technology trends and when I mention a smart doorlock, a smart lightbulb, or a smart scale, they just ask yeah, so what? What problems do we, in IoT, solve for people who are not geeks?
That’s exactly it. A lot of time we’re adding connectivity because we can. Smart scale is a really good example of that. Just make a better scale, dammit. Another example people like to make fun of is the Quirky Egg Minder, s smart egg holder in your refrigerator so you know if your eggs are fresh. Clever idea, it’s good that they tried it, but I don’t think it’s a burning issue for most people.
However there are some really compelling examples. Something I found on Twitter the other day: a smart bicycle crossing. As a bicycle approaches the crossing abd a motion sensor triggers lights in the street, so that drivers get a warning a bicycle is coming. It’s smart — but not too smart — and the value is clear. You’re just letting drivers know bicycles are coming. It’s a beautiful example: people get that and don’t feel threatened by it. We will uncover more types of things like that. Simple, predictable, and valuable.
Before we wrap up, I wanted to ask you, as a UX specialist, about another hype cycle: bots and conversational UI. Do you expect it to become a major way of interacting with technology?
Conversational UI feels to me like videophones did in 1980. We thought that video was going to be in every house. It was in every vision of the future and everyone was so excited. But it took thirty years before the technology actually matured enough for us to get things like Skype and Google Hangouts. The conversational UI fad is enthusiastic, it’s interesting, and there are cases where it solves the problem. But it needs a significantly more nuanced vocabulary, interaction, and synonym model to work properly. It will eventually get there but people are naively assuming that it will be in two weeks. It’s coming over time but I believe it will be a long road.
So we’re not there yet with language processing to make it universally applicable?
Absolutely. Just try any voice interface. Yes, it can do certain things really, really well but the vision is that you will talk to your computer at any point of the day and it will just do things for you. That is a lovely vision, but even if you make a pause and go ummm, it completely throws it off and you have to start over again. If you can’t even pause, how do you expect this thing to work with real people?
Amazon’s Echo received outstanding reviews, much better than anyone expected.
Alexa did surprise everyone and for a certain limited subset of tasks it is a lovely product. Asking about the weather, about traffic, turning on lights, and so forth. We should definitely applaud the success of products like that and you could see why the success of Alexa is making everybody go crazy about conversational UIs. It’s a good first step, but still not enough. But maybe I’m being a bit of a UX perfectionist.
Don’t you think this kind of attention that voice and bots get is detrimental to the goal of actually giving deep thought to the problems technology can solve and encourages chasing TechCrunch headlines and VC money instead?
That’s a bit strong. The way we evolve technology is like children splashing in puddles. You just have to explore, stomp around, see what works. There will be many, many things we’re going to try and that’s just how human nature is.
I don’t wish to be critical but I do get frustrated. I have a theory that the internet is so overwhelming and we have so much data that people are desperate for clarity and meaning. So they try too hard to figure things out and jump to solutions as if they’re offering the answer. So these days you get people rushing to the left side of the boat and saying it’s all conversational UI and then to the right to be it’s all about machine learning! In effect, we have these mad dashes of people groping and desperately hoping that this technology will save us. And what will of course happen is each of them will have its own trough of disillusionment. We will learn something, we will keep moving, and next year it’ll be better. It’s a never ending cycle of try-stumble-get up. We will get there, it just takes time.