a16z Podcast - a16z Podcast: Beyond CES: Connected Home Devices, Voice, and More

Episode Date: January 22, 2018

In this hallway conversation of the a16z Podcast, Benedict Evans and Steven Sinofsky discuss CES 2018 and share insight on what they took from this year's show. How much can you discern each company's "big picture" strategy out of the slew of new products, press releases, and announcements that flood the floor? How do you sort the wheat from the chaff? And beyond the event of CES itself, Evans and Sinofsky analyze the experimentation we're beginning to see in connected consumer electronics for the home. When it comes to the smart home, it seems as though more is better -- more devices, more connectivity, and a single network to rule them all -- but that isn't the case. How (and when) will these devices and appliances -- some of which you only buy new every 10-20 years -- all connect into one system, and what will that platform look like? Which devices will we actually need to be "smart", and what will be today's equivalent of the electric carving knife? Where will new kinds of UI come in; when will a simple GO button be the better option? All this and more in this episode.

Hi, and welcome to the A16Z podcast. We're here today with Benedict Evans and Steven Snovsky in another of their hallway style conversations. They share takeaways from the recent annual Consumer Electronics Show, CES 2018. More broadly, they consider, what's the future of the smart home and who are the key players, especially given the rush to bundle smart devices together under a single system? And then, on the design side of things, when is simpler better, and which devices require more complex or simpler UI? So, the two of us separately went to CES, in Las Vegas with another 180,000. We didn't even see each other, which is sort of funny. I did bump into like 10 people that I knew at random places around the show, but we didn't

bump into each other. And we each walked for about 10 miles. And we saw a bunch of stuff. And we saw a bunch of press releases and we saw a bunch of news coverage. And now that we are a week and a half on, we're sitting and thinking, okay, have you digested this? How do you think about what was actually important as opposed to what were the 300 press releases and the exciting coverage. And so I think, you know, there's just, there's, there's so much there. And I think the first thing that's kind of interesting is a lot of people go and they want to see the grand master plan for every big company. And so the question is, can you really discern the grand master plan for every, every big company? Yeah. I mean, so you see clearly there's a whole Alexa story. And then

you've got people going around saying, well, Alexa and voices and your platform, and this is going to replace smartphones. And Amazon's got off to a great start. And Google's coming up behind. And it starts sounding like kind of NASCAR commentary as to who's in front. Right, right. And, of course, oh, Apple are completely screwed because the home pod missed Christmas and all of these. And Cortana is, like, doing press releases the whole time. Like, they're still going to try and they're going to, you know, take a slow pace.

And so the thing that I think you see here is, particularly when you look at kind of a super big company that's got lots of stuff going on, is there were some things that are actually this is the board level fundamental strategy of what they think is crucial to the existence of the company in the next five years. and then there's stuff that are kind of applications that flow out of that some of which may or may not work and then there's things that happen because you've got 10 or 15,000 people and you've got an SVP for widgets and taps and the SVP for widgets and taps is going to try and get into bus tubs and that doesn't mean that anyone in the board knows that you're trying to get into bus tubs it's just because you've got somebody whose job it is to do that so they're doing it and I think particularly when you see these kind of blizzard of announcements in these huge stands you have to kind of take a step back and think well which of these things are tactical and which of them are strategic

Jake, and which of them are existential, and which of them, well, they're just doing that because they've got people to do it. I think that that's really important, especially in the startup world, because there's this tendency, you know, if you're a startup and you work in a particular space, and then you just hear an announcement from a big company that they're working on something, you apply like, oh my God, it must be important. And also, I'm going to get squashed. Right, right.

It's important to me, so it must be super important. And it forgets that dynamic at a big company where, you know, you have end groups working on their vertical little things. and then the new thing appears in its N plus one, the likelihood is that three, four, or five of those groups are each going to go try to do it. And to your analogy, it's like if bathtubs became like a thing, it's likely that all the groups are going to go find a way to go do bathtubs.

And we joke about this all the time about, like, adding photos or adding messaging. And so a lot of what you see at CES is... It's the internal bureaucratic dynamics. It's the org chart. It's just the org chart of the company. So, like, one of the ones I saw was like a whiteboard. and you know Samsung has a TV group and they have a phone group

and they have an industrial group and Microsoft has a collaboration so they have to have a whiteboard and Epsilon has a whiteboard because they make displays right right and printers and their whiteboard had a printer built in and so the first thing when I saw the whiteboard which was announced

at a very low price and it was a very nice product was I wonder what operating system it has and like which part of Samsung made it because that would dictate like does it have an OLED screen or not an OLED screen Is it touch or not touch? And it turns out it came from like the industrial equipment side of the house, which meant it didn't even really have like a whole OS in it.

It had only just enough software to like make a bulletproof whiteboard, which turns out to be a great product. And you all say, no, that there are people in three other groups in Samsung who looked at that press what he's in school. And they want to have a meeting and they're like, we got to get in the room with Mr. Kim and figure out like why I'm not responsible for whiteboards anymore. But then the interesting thing is what do we say about it for the pure, you know, the companies we think of as tech companies, you know, because it's really, like, you could divine the master's strategy for, like, a Samsung or an LG. Yeah, so, well, this is sort of the thing

we were talking about earlier, that everyone goes there and says, what's the Google strategy, what's the Facebook strategy, what's the Amazon strategy? When you go to the Google, the LG stand or the Samsung stand, it's actually kind of interesting. You look at the tech press coverage, and they kind of, like, why is there a washing machine here? Why is there a dishwasher here? For like five years now they've been doing this. And the answer is, these companies make everything, and they want to make everything electrical in your home. They want to, to make the air. They make the air conditioner and the washing machine and the TV and the PC and the phone. And they have an org chart around that. And so they want you to buy

your washing machine from them. And clearly in some way the washing machine is going to be digital or connected at some point in some way over the next 20 years. And so they want to do that because they've got connected people. And they want it to be, have some software in it or some interface that means you have to buy the washing machine from LG and not from Samsung and not from Electrolux. And so they've got their kind of leverage structure. strategy of connecting all of these things together, which is also what Sony were doing 20 years ago. So what Sony did. And interesting, like, Sony was very impressive, but, like, massively impressive, but didn't make nearly the breadth of things that are currently made by the Korean

companies. I mean, it's mind-boggling just how much stuff, unique products they make, and get distribution. Like, you could go to Best Buy and see a lot of stuff from Korea. Yeah, I mean, there are these sort of stories. If you live in Korea, you could spend your entire life, only consuming Samsung products from the Samsung house to the Samsung car to the Samsung bank and everything else. I mean, one of the things that comes out of that there, when you look at like, back to the washing machine point, okay, the washing machine has a screen and there's a bunch of interesting things you can do. Like, you can pour a whole bottle of washing liquid into it, and then you can tell you it what load it's got and it will automatically

put the right washing liquid in so you don't have to load it every single time. So there's kind of interesting, useful stuff you can do. But then you think, okay, does this need to have Android on it? Does it need to have a little tablet on the front? Does it need to have a screen at all? does it need to have just have a kind of have a go button? And this is a question that applies to all of these sort of non, these computer devices that are sort of smart and sort of a computer but aren't going to be your main device. They're not a PC, they're not a phone, they're like a fridge.

Yeah. So how much UI does that have? How much computing does that have? How much value do you have on that device versus how much is it like a single purpose thing? And I think that's a really important point for folks to understand because that's one of the ways you can sort of discern how much is like a corporate existential strategy from like a local, like I make the plumbing-based appliances, which might be how the org chart is organized. Because like if you looked at Sony way back in the day, like all

of a sudden everything had a memory stick in it. And a joke dial. And a dial. And like a memory stick in a radio seemed like a very dubious kind of thing. But believe me, the radio person went to a board meeting and was told memory sticks will go in all of your products. And it was absolutely a board level kind of thing. Whereas it, Samsung, just like the whiteboard showed, you know, it's not clear yet that, like, Android has to be in everything. But it is clear if you went to both of them that, wow, like Bixby was a board meeting. Yes, force.

And I picked Samsung's Bixby and Chloe at LG. Those were board meetings, or at least, like, an exec staff meeting where they all were told you've got to go participate. And then you can sort of see the seams because way out on the edge, a home security camera, they just do like the checkbox minimal API because he's like, look, I got to compete with ADT and ring and a whole bunch of companies for distribution in the US. I don't have time to be all corporate. Plus, if I want to actually get into Best Buy, they're going to tell me, well, you've got to support Alexa and Google HomeKit. Exactly. So you have these devices

that have, it's to your point, it's as though you've got a bunch of Sony devices that had the toaster in it that had a memory stick and an SD card and a compact flash card. And so you have the security camera that has Alexa and Bixby and Google Home and HomeKit. Right, which is for startups a super interesting point because it means like if you take the lead from the big companies in what to do, you'll see that all of them, they have the resources to like do everything five times. And in fact, like even some of them will, they'll literally just have two products. They'll have the for the Korean market, which is like all up Samsung with no seams. And then the best buy one where they put in like API access for other stuff and

it's a different model number, might be a different color. And the thing is this is particularly for voice, this is just another cross-platform kind of thing. We'll get to whether voice is a platform itself, but like, you know, the idea that you're just going to go do all of these presumes that we've reached a steady state. And I think that's not where we are. It's weird because, I mean, there's sort of two or three connected home questions. There is the, how many of these things are actually going to end up being smart or connected at all? Will lights work? Will the doorlock work? Will the fridge work? Will the cooker work? Will the cooker hood work? Will you say, hey Alexa, turn on the extractor fan.

How many of those things will happen? That's an exhaust fan, just letting folks know. The fan over the stove. How many of those things will happen and how many of them won't? You can kind of look around your kitchen and you think, okay, well, toasters worked. Microwaves mostly worked. Electric carving eyes mostly didn't work. You probably wouldn't have been able to tell that in advance.

And that's sort of where we are with all the smart stuff. Yes, there'll be a bunch of stuff. Some of the stuff that's being proposed will work and some of it won't. And that's one of the reasons you see so many of these different things because they're all riding on the smartphone supply chain. There was the era, as you used to talk about, like, of putting a DC motor in every appliance. And there was a time, and in fact, in the U.S., electrical codes are actually built around this notion that you're going to have, like, a lot of plug-in appliances in your kitchen.

So there needs to be an outlet, like, every four feet at counter height. And, you know, there were, like, carving knives. Like, nobody really has those anymore. Electric can openers. How many blenders and mixers and other things? And so there was a time, you know, mostly in the 60s and 70s when, like, everything got a DC motor in the kitchen. And then we worked out which ones made sense.

And then we realized that it didn't happen. And it's also different by country. So, like, everyone in Japan has a rice cooker. Everyone in the UK has a kettle. In America, maybe neither of those, but everyone has a coffee machine. Exactly. So you kind of can't predict which of these toys or these gadgets will turn into just something that everyone expects to have. So there's that, like, well, what will be smart and what won't be smart?

Right. Then there's, will they all be one system or not? And will you be able to say to Alexa or to your phone or to a panel on the wall or something, do this thing? and the lights come on in this room they go off in that room the stove turns on the water turns on the sprinklers do this

so like it's all one thing or at any way you can just say do this and it can happen as through one interface or do you have like point solutions like the door lock is the door lock and it unlocks automatically when you walk up to the front door

and this is sort of nothing else he doesn't talk to anything else right and this is where you can get really sort of confused as a as both a consumer and like looking at strategy and what to build because like it is this sort of tech geek nirvana

you know like it's also a big company And a big company strategy. Because they wanted all to be one system because then they can get you to buy it all from them. Right, right. The difference, I think, is that the big companies, especially the huge ones, are actually, they can have the strategy, but they also need to sell things individually. Like, they can't, like, you can't go to Best Buy and say, I need a new washer because mine broke. And then they'll say, well, not unless you buy an air conditioner. Yeah, and it only works if you've got a Samsung fridge. Right, right. And so they, you can see that in their, in their boots where, you know, like the fridge could be the hub or a speaker could be the hub, you know, and then you have to ask yourself, do I want the fridge?

And we'll get to that in a second. But the big thing is that the thing that you have to realize is that the geek scenario of, like, is no different than the old one of, I want to just relax and listen to music and push one button on a remote that then, like, turns the music on, turns the TV off, switches the AV input. You know, you call this like the Barry White mode. Yeah, yeah. You come home and the door recognizes you're with a date.

And so it turns the music, turns, puts the right music on and sets the lights to the right level and does, like, the whole thing. Right, right. The Austin Power scenario. Exactly. And, well, Austin Powers and Barry White in one. But it is, I mean, and that, like, the thing is, that's been a thing since, like, infrared programable remotes all through X10 and wiring in the lights with special ILS. And, like, that stuff is just so much more complexity than actually where.

I mean, you, you know, you try to do those scenarios and you spend all your time realizing, wow, I can't tell if the three-way light switch, which way the lights are. So it just won't ever correct itself. And then I have to go through the house and, like, reset everything. So there's almost like there's two levels to the problems. One is getting the technology to work. And the other is whether anyone actually wants to go through and make 10 different choices about what the settings are. And so you could argue the bearish scenario would be all of the connected home stuff means that what was a hobbyist scenario becomes much easier to build, but it might still be a hobbyist scenario. Well, and because there are real things that matter.

You know, we've both talked about, like, you know, people visiting your house and, like, how do they operate anything is a very odd thing. Like, they didn't even know if you have Alexa, Siri, or Google, so they don't even know how to start. And so you can imagine this scenario where somebody is just like the babysitter is literally Google, Alexa, Siri, turn the lights on, I can't see anything. You know, and it's sort of absurd. But the idea that you'll get all of these, you know, that's the thing about this market of consumer electronics is it's generally been highly fragmented because of the different ways people live globally and even, you know, urban, rural, you know, suburban. And you have like a 10 or 15 year replacement cycle. Crazy. I mean like 20 year replacement cycle on a fridge.

So you're not going to buy a new fridge just because there's a new one that's got a color screen on it. And then even then you go, you know, you go to, you know, you're going to do a fancy kitchen remodel or something. There's a good chance you have your heart set on like one element of it. Like I really want a gas range that looks cool and professional. Like, God, that means that all of these companies are going to have to make all of those parts. And there are only two companies that can do that, which is Samsung and LG. And even when you've done your kitchen.

that doesn't mean that you replace all of the locks in the house and swap out the hole and rebuild every light switch in the house. Exactly. And so I think that it's a great strategy and it drives a lot of interesting things. But the odds of this coalescing are very low. And that's why this notion, and the lesson from Sony or RCA before it or Whirlpool, they're all there. Like, nobody ever came to globally dominate the home like this.

Yeah, the interoperability standard is AC power. Right. And I think that that sort of means that it's interesting when you get to the now our tech company world, like how much of this, like what, is it really a platform or what are we talking about here? So this kind of gets us to voice, I think. And I wrote a blog post about voice, sort of 18 months ago. And I said there's a whole bunch of reasons why people want voice to be a thing, whether

that's in a box in your home or on your phone or on your, whatever it is, people want voice to be a thing. And one reason is because machine learning means that some parts of voice now work. So you can turn audio into text pretty reliably. you can turn text into a structured query pretty reliably. So you can work out this person who's asking you to play music and you can do it. The second thing is that anyone who isn't Apple or Google has made your platform envy and they want to find a way of going around that and not being worrying about Apple or Google

cutting them off at the platform level on the smartphone. And they want it not be a commodity and whatever it is they currently do. And so that's the, that's the, there's a massive Amazon Facebook motivation here which drove the Kindlefire and then now has driven Alexa and drove a bunch of stuff at Facebook. And so there's a bunch of reasons why people want it to be the thing and a bunch of reasons why people think it should be the thing. I think there's a sort of a fundamental misconception I see over and over again about voice,

which is people think, well, it's almost there, but it'll be better, and it'll get better, and it'll be fixed, and in a year or two, it'll just be, all existing problems will be ironed out. And I think the misunderstanding there is there's actually two challenges here. One is that you say, Alexa, hey, how much longer is there on the timer? And it says there's three minutes left on the timer, instead of saying the Times newspaper has been printed for 175 years, which is what Google Home did to me the other day.

And that's like the voice recognition problem. And that does get fixed. That will absolutely get fixed because of machine learning. The other problem, however, is what happens if you ask it a question that it doesn't know how to answer? Because everything that you say to a voice assistant is basically getting plugged into an IVR. It is now accurately able to recognize what question you've asked, but there are only 20 questions they can answer. And each of those questions is somebody sat down and wrote a thing to be able to answer that question. And at a certain point, you stop being able to scale that.

you can't answer 50,000 questions, or 100 questions, or 5,000 questions. And Amazon has tried to solve this with skills by having an API. And so now there's 5,000 skills or 10,000 skills on Alexa. But the problem with that is you still don't know what you can ask. Right. So it gets you to a command. And the thing people always say this is a command line. Yes, the command line has a million commands, but you don't actually,

you have to remember which ones you can use. Right. And I think a lot of this comes from the spectacular work that Google has done over, you know, the past 10 or 15 years of just you know, like being able to type into a one-line edit control, you know, vast numbers of things. And what they have done, what I think is an interesting lesson is, is that it started off by being this pure, like, we will match your question to the right website kind of thing. And then what they realized is that could only take you so far. And then it turned into, like, they just have different teams. They have specialized teams that say, well, if we detect this as a travel-related thing, if we detect this is, you know, a film or movie-related thing, if this is a knowledge-related thing.

And then even just the appearance of Wikipedia change their ability to answer directly. They can pull the structure. Right, because now you can ask how old anyone is and if they have a Wikipedia thing or where were they born, those are easy things to do. But there's a basic fallacy here, I think I forget who it is it pointed out. Basic fallacy in sort of non-technical people

looking at AI in any sense is you see the machine do something and you think, well, for a human being to be able to do that, they would have to be able to do all these other things. Therefore, the machine must have that capability. Right. And you forget, actually, looking at a washing machine. A washing machine can wash clothes, but it can't wash dishes and it can't cook you dinner. And what we have here, you know, back to our talking, we were talking

about connected home here. What you have on Alexa is somebody built a washing machine and somebody built a dishwasher and somebody built a toaster and somebody built a fridge. But each one of the things that Alexa can do, somebody had to build. And that doesn't scale. And so the idea this becomes like a universal platform or it's like it's the next smartphone is a complete blind alley. What does get interesting is back to our connected home, you actually probably, can remember how many lights you have or how many rooms you have and like the door, the turn on the lights, turn up the heating, turn, preheat the oven, you can remember those kinds of queries. And so voice might be like your universal UI for those. Yeah. But that

presumes that that that actually is something you want to do. Well, and that, that's why I was actually looking forward. Last year at the hotel I stayed at for CES, they announced that they're working on having Alexa for the rooms. Now, this is a relatively new hotel, which all of them in Vegas are. But this one, they already had like what I think is probably the best central automated switch that I've ever seen in a hotel. Like it has six buttons on it and it's like lights, drapes and front door like privacy and clean. And it's wired and it's very reliable because it's a tablet. It's not even a tablet. It's like literally a resistive screen on this little tabletop thing with a USB charger in it. And oh, and the other is the

thermostat. And so you push a button like drapes, and it says, do you want the shear on or off, and do you want the drapes on or off? And notice that's one of the first things it has is they're on or off, and you can push on or off, not a toggle. Because one of the notorious home automation things is someone gets up and manually closes the drapes, and then the automated system doesn't even know which direction they are, which is really common with lights. And the same thing on lights. It has all on, all off, and then like dim. That's it. And I thought, oh, that is going to work really well for Alexa. Because there's a narrow domain and you know exactly what you can ask.

And you don't have to notice very much. And so then I got there last week and then the lights are, and I spent like 10 minutes trying to get the thermostat down because the room, of course, was Vegas cold. And I could not get it. Like, you know, Alexa would come back, I'm sorry, there's not a thermostat named thermostat. And, you know, I would say, I'd like to lower,

I said, Alexa, lower the temperature to 70 degrees or raise the temperature to 70 degrees. And it says, to auto 61 degrees. And I was like, did it tell me what it's set to? And I just went in this loop. And then I gave up. And then I just started in Alexa, open the drapes. Alexa, close the drapes.

And it was flawless. And I'm reading a two-page index card, a two-page cheat card. Well, this was kind of my point, that there's two problems. There's the one problem is the thing has to recognize you. And that will get fixed. You know, it will get to the point that it will always know that you were talking about the thermostat, whatever you say. The problem is, if you then say to it, hey, what's a good restaurant that I

could have dinner at night. Somebody's going to have to have written that. Well, and that for hotel is actually super interesting because it might use Yelp and then send me to Cesar's. And so, like, and it turns out the queries that aren't built in the hotel skills that they made, like the thermostat and lights and the drapes, are just going to Alexa. And it turns like, anything like that, like, you know, what's the weather? It's not going to the weather that the hotel provides, which might come from any particular source. And so the idea that, that hotels would turn us all into their own, like, huge system is pretty complex. I think there's a sort of kind of, the thing we kind of keep circling back to is there's the vision of the unified system.

I mean, it's like when you watch a sci-fi movie about the far future. Somehow there's no applications. There's just one interface on the screen. Right, right. Set aside the fact that there wouldn't even be a screen and you wouldn't have pilots and you wouldn't be flying. There wouldn't even be any people on the shit. Yeah, well, that was the, I did actually see, like, the automated motorcycle driven by like a robot, which I just thought was the weird. It was like a Cylon ship.

made me sense. Why have you put a droid in the back of the X-ring? Why isn't the droid just flying the Star Wars for everybody? We've all lived through that this summer. But there's a sort of, there's a sort of a separate thing. There's always a fantasy that it'll be unified system. The reality is you'll buy one thing now, and in five years time, you'll buy another thing, and 15 years after that, so you'll have all the... And so this kind of brings us back to the stuff we were talking about at the beginning, which is you have the big company strategies, and you have the desire to create unified systems because that will get you to buy all the stuff, but the reality is it will always be kind of more

messy and more complicated. Right, and that's why I do think, like, the real sort of advice, like, if you're making things, it boils out to two things, which is, one, which we need to go into a little bit, is just how much intelligence to put in your device versus the rest of the network. And then the second is, which one of these sort of systems do you want to plug into? So let's just talk about it for a second, because one of the things is that's driving me crazy is just, like, yeah, you know, Samsung makes screens and they make arm chips, so they seem to decide that every little device has to have a screen with an arm chip in it and that just seems like

overkill yeah well this is the ui question you know should the i mean if there's a the razor i always use here is how many questions should the computer ask and actually so half of the stuff of this stuff means now there's more questions that you have to answer and the other half is now there's less questions that you have to answer so you could argue like the problem with quantified self and fitness trackers and so it was actually more stuff you had to do rather than less stuff so to the washing machine Should the washing machine have a screen that asks you what you've put in and what load you want and what settings you want and all this? Should they take the dials and put them into the screen? Or should they just be a go button and nothing else?

And actually the ideal for the washing machine is this just a go button. The ideal for a dishwasher where you don't have to worry about like different fabrics is absolutely you should just shut the thing and press go and that should be it. And there shouldn't be any settings at all. And the rest of it should be the chip problem. Right. Which the example for me was just all the connected microwaves. Because for 25 years, all the microwaves you could buy have had, like, this ridiculous grid of buttons, you know, chicken, steak, frozen, reheat, you know, and popcorn. They always have a popcorn button.

And, like, it just was ridiculous because, like, the food you want to do is never quite there. And so the question was, like, did this really help? Because the directions on the food always just say heat on high for two minutes. Yeah. Like, which turns out on most microwaves to be tricky to do now. Like, because they want you to use all of these things that make their. microwave look unique. And so I do think that the conclusion is, like, less UI, less intelligence is the, or more intelligence is the right way to approach building this. It also reduces the

cogs and the maintenance and extends the longevity of the device. So the example I've used, which is I'm sort of writing about something about this now, is, so there's two parts to this. One is just an observation. You shut your door. When you leave home, you shut the door, the door locks by itself, and that's like a 2,000-year-old technology. When you come home, in 10 years time the door will also open by itself and that's based on smartphone chips and computer vision like cars work today but from the user's point of view

it's a door that locks by itself and unlocks by itself so one of those is super advanced technology and the other isn't but the user experience should actually be completed the same point of those and it probably shouldn't have a screen to unlock any more than it should have a screen to lock itself right and so there's sort of that thing the other thing I was thinking about was the kettle so certainly everyone in the UK has an automatic electric kettle

You plug it in, you press the button, it heats it up until it's boiled, and then the heating turns off. And you look at this, and I feel like channeling these people who look at CES and say, look at all these stupid devices, look at a kettle, you're an idiot, you could buy a saucepan, put the saucepan on the stove, that will boil water, why have you got a kettle? Right, right. Which is true, but it's actually better to just press a button, go away and come back, and the water's boiled, and it's turned the heat off. I push this even further, the vegetable peeler, not like an electric vegetable peel, just like a kind of a $5 vegetable peeler. Yes, you could peel fruit with a knife. The 15th time that half of the apple has gone into the sink And a small piece of your thumb has gone into the sink

You go and you buy a vegetable peel it Because you're better And I think that's the way to think about a lot of these things That they remove friction And they remove a question Or they remove something that's taking a tiny piece out of your life Every single day And like, yeah, you could boil water on the stove

Yeah But why have you not just got a kettle? Why does the door lock have a screen? It should just unlock Right Why does it come? We have a kettle here that has 15 different temperature settings Based on what kind of tea you're making

Like just really? Yeah. Press the boil button and then pour the water. If you're making things, you want to have the screen. You want to have the screen, and you can actually be different. And this is something, of course, that Apple tends to do better than anyone, which is be minimal about things. Like, you know, if you're making a Windows laptop, it's, like, cool to continue to have all the legacy plugs and all of the different variants of adapters and plugs built into the machine. And Apple makes these bold choices that really do move things forward.

I mean, it was just the anniversary of the MacBook Air. and I was at that event and I sat there with all the famous reviewers and everybody talking about there's no Ethernet what are we going to do? Because back then you know basically you and you know Apple at that demo Steve Jobs went through a lot of length to talk about how to use a DVD player

or actually a CD ROM at the time and it was like air CD or air drive They sold an external drive as well but yeah air disk yeah air disc or something it was called and it was like trying to comfort everybody and I just love that it's exactly the same as them shipping a headphone adapter with the iPhone Right, right. And it just doesn't, these things, life does move forward. And I think that that's really important for everybody to sort of recognize as a key takeaway. Finally, just really want to touch on this point about being cross-platform and whether you should embed the Alexa runtime or the Siri runtime or the, which you can't embed, or the Google runtime, which was Google's, hey, Google's strategy, which was everywhere. Like, they spent a lot of money to do, hey, Google, everywhere. But part of what they did is, it's the Android thing where it's just, it's free to

make your commodity device be different. You buy a small amount of chips and you get the free runtime. Like, does that make sense to you that everything should also embed, like be basically the Alexa microphone and endpoint? Yeah, it's sort of weird. I mean, it comes back to this, you know, you want to make it the thing. Google want everything to be an endpoint for Google Home. Amazon want there to be Amazon Home everywhere. It doesn't necessarily make sense to a person that everything should be that end.

point. I mean, this is, you know, the joke that you'll ask a question and you'll have Google and Alexa arguing with each other on your kitchen counter as to which of those is the thing. I mean, there's something we haven't mentioned. We talked a little bit about voice earlier. There's a lot of experimentation here. There's experimentation around what devices will have smart and what kind of smart and what UI and what system they'll be connected to and is it voice and what kind of voice and voicing what devices. I think there's also just, you know, going back a level to the sort of the actually to what actually is a super high level strategy, certainly for Google will probably for Amazon as well, which is, what do we do with AI?

What do we do with machine learning? What's the product? We've got this kind of fundamental capability that computers can answer these new kinds of questions. What's the product? Well, the most obvious immediate products are right now you can do image consumer image recognition. So you have Google Photos, which is fantastic. And we can do voice. Okay, well, what could we do if we recognize voice?

Well, maybe we could answer questions. Let's try that. Let's see if that works. And you can see in this, I think, an echo of Google Now, which is Google Now. which is Google Now is sort of, well, we've got all this information. What other product can we create out of that? And Google Now, for a bunch of reasons, didn't work. And they kind of shifted that to be, okay, well, now it's Google Assistant.

So instead of it being passive, it'll be active and you'll ask it a question. But there's a lot of sort of almost kind of top of the S curve, of one S curve, beginning of the next S curve, of like, we've got these primary technologies and this primary set of information, whether that's the arm chips or the screens or the microphone or voice recognition. and we're sort of experimenting, well, what would the actual consumer proposition for that look like? What would it be? Right. Is it a voice box? Is it glasses? Is it a text assistant?

Is it a bot? And we don't quite kind of have a resolution on what that would be. And I think that the key thing for people to understand when they're making things is that where at the very beginning there's a lot of experiments, but you can't have your product rely on like an experimental approach. Because especially if you're making it. something in business that needs to have an answer or in the consumer space that the lights do need to actually turn on, you know, like, the thing about the Google search experience is that it was highly resilient to being wrong. Like, if you typed a query and you didn't like the answer, you typed another one.

And you didn't lose anything. Google now, the problem is, is that if you start to rely on it to tell you when to leave for a meeting and what traffic, it has to know a lot of other stuff in order to get that right all the time. And the cost of being wrong just once is very high, like it could be the job interview or something like that that you miss. And this is like literally the biggest lesson from Microsoft Clippy, which was like automatic features and automation are all cool, except they kind of have to be right literally all of the time. Or you shift your expectation. So the thing about Google search is they don't give you one result, they give you 10.

So it's actually manually created. Whereas when you're saying turn the lights on, they can't give you a list of 10 things that they think you might have meant. Right. And also, even again with Google Search, like, they give you a list of 10, and if you don't like it, you can click the two at the bottom of the screen or type a different thing in, and you're, you don't lose out anywhere. Because if you typed in, buy 100 shares of, which is one of the first voice things I did on Tell Me, I asked for, tell me the stock quote for Cisco. And I'm sitting there on the 520 bridge, you know, on my analog phone, and it's giving me the stock quote for Cisco, the company that makes, like, kitchen supplies. and I could not get the stock quote for Cisco. And, like, if I were telling us buy or sell, that would be really bad. So there's a lot here for makers to really ingest.

And a lot of sort of early experimentation. I mean, one of the ways I've sort of talked about IoT is, like, the smartphone supply chain, on the one hand, means there are all these cheap commodity components, and suddenly you can make any connected thing you want really much, much easier than you would have been able to in the past. On the other hand, ubiquitous wireless and cloud means you can kind of light them up in ways that you would never.

been able to in the past. What we don't have is like the really clear answer as to, well, what would you create? So it's almost come in the opposite direction. Previously, you'd have the vision of the thing you wanted, and there'd be like a 10-year march to get the technology to build it. Now, all the technology is there for almost anything you can think of, and people are trying to work out, well, what is the thing you would create? And also, you can make it very inexpensively, you know, there's 100 3D printers to choose from. And so I think that for me, the big lessons to walk away from is just that there's still just as rich an environment for making the right products and the great products, and they are all changing our lives every day.

But it's not happening as fast as people can write blog posts about, like, this is the future, be here. And that's frustrating people. And I think as, you know, as people building companies, the key is to really focus on what it is that the problem you're solving and not worry about, like, trying to discern a galactic strategy across all of these things. That there's still, it's all about building great products that improve lives. Exactly. There's like specific problems you solve, specific questions that somebody gets asked every day that you can take away. Well, thanks, everybody. Thank you.