Amazon Echo: Always listening

Hey, David Pearce here. Before we get into the show, just a quick reminder. Version History podcast. That's our YouTube channel, that's our TikTok channel, that's our Instagram page. Go subscribe to all of those places if you want. All of the clips, the full episodes we're doing on YouTube and honestly, I think the show looks fabulous. You should really check it out. Everything we're doing is on those channels. Keep it locked. We have big plans. Let's get into the show. Imagine a world in which you don't have a phone, you don't have a laptop, you don't have any screens at all. When you want to get something done, all you do is just shout it to the universe and it happens. This is the future Amazon thinks is coming true. And with this little tennis ball can of a speaker that it's about to launch, it thinks it can pull it off. From the Verge in Vox Media, this is version History, a show about the best and worst and most important products in tech history. I'm David Pearce and today on the show we're talking about the Amazon Echo.

0:00

I'm not quite sure how to help you with that. Refreshing Wild Cherry cola meets smooth cream, the treat you deserve. Pepsi Wild Cherry and cream. Treat yourself.

0:52

Last night you spent two hours deciding what to wear to the party this morning. It'll take you two minutes to list it on Depop and make your money back. Just grab your phone, snap a few photos and we'll take care of the rest. The sheer dress and platform heels you'll never wear again, There's a birthday girl searching for them right now. Your one and done look is about to pay for your next night out,

1:20

or at least the ride home.

1:42

Your style can make you cash. Start selling on Depop, where taste recognizes taste. All right, we're back. It's time to talk about the Amazon Echo, which if you are listening to this, I would like you to imagine a very tall and somewhat futuristic looking can of tennis balls. Even inside of Amazon. This thing was called the Pringles can and it is a tall cylindrical speaker. There's a little dial on top and a bunch of speaker grills around it. It is super duper boring looking and actually that turns out to be kind of the point. Psa. Before we start, this is the most important thing I will say this entire episode. If you have Amazon Echo devices anywhere in earshot, mute them, all of them, right now, because we are going to say the word that is going to wake them up so many times that it's going to ruin everything, including this one, which we're going to have to mute at some point in the very near future. So go mute those. We're going to get started. Joining me in the studio today, Jen Tuohy. Welcome, Jen.

1:44

Hi, David. Happy to be here.

2:42

So you are the Verge's smart home guru, czar, expert. Were you covering smart home stuff when the Echo first came out in 2014?

2:43

I had just started, yes. I mean, I've been in journalism for a little longer than that, but I had sort of pivoted to tech journalism around 2012, 2013. So I was right there at the beginning with the Echo device.

2:53

Did you get one early on?

3:08

Yes, this is mine.

3:10

This is your original one?

3:11

Yes, I got it. I actually got it as a Christmas present. I wanted one. I was intrigued by this concept because when it launched, it wasn't necessarily a smart home device. When it launched, no one really knew what it was, and I asked for one. I put it on my Christmas list, and my husband, who is not techie, bought it for me. And on Christmas Day, when I opened it, he said, I hope that's what you want because I have no idea what this thing is.

3:13

Is that right?

3:39

Yep. And I. I gave him a similar present, a Kreg jig.

3:39

A. A whatnot.

3:45

Exactly. So we both gave each other pieces of technology that neither of us knew what they were and what they were designed to be.

3:46

Well, when we do that, EP have him on, it'll be great.

3:52

But yes, it was. So I've had this in my house for over a decade now, so it's a little dusty. Sorry.

3:54

It is a little dusty. And you've replaced it with one or two Echo devices.

4:00

I have a couple others in there. This one's actually been in my husband's garage for about five years, but it still works. So.

4:04

Very impressive.

4:10

It's pretty impressive.

4:11

Yeah. Hayden Field, also here. Hi.

4:12

Hey. Excited to be here with this relic from the past.

4:14

So you are an AI reporter, and I think I wanted you to do this with us because there is very much a smart home story here and about how we interact with technology. But I also, like, there is a case to be made that there is some beginnings of an AI revolution inside of the story we're about to talk about and inside the device we're about to talk about. There's a lot going on in this device, which is kind of why I wanted to talk about it. Right. It's the beginning of the Amazon Echo, which is the device in front of us. Here it's also the beginning of the, the Amazon Alexa assistant. And I think it's fascinating. I'm going to just.

4:18

There it goes.

4:52

That was one. We're one for one. And we're going to mute this thing. Uh, this, this thing like blared at me many, many, many times in my testing of this in 2014. But anyway, these things are very much the same product at first and then diverge in these really fascinating ways over time. Um, but for our purposes here, just to lay the groundwork here a little bit, we are going to try and avoid the whole Alexa story for now, if that makes sense. Like, I want to do our best to talk about the history of this device because I think the story as we're going to get into is very much about this device and sort of built backwards from it into this thing that became Alexa, that ultimately became much more important and I think in a lot of ways much more sort of prescient about the future than this Pringles can of a speaker. So let me back up all the way to the year 2000. And in the year 2000 I found this video of Jeff Bezos on stage with Charlie Rose, which is a whole thing we won't get into. This is a very weird setup. He's like on stage at a very like intimate table with Charlie Rose, but then there's an audience of people. It's all very strange, but he's asked, what do you think the next big thing is going to be? And let me just play you his answer. I think if you take the next

4:52

big step, I believe that for mobile

6:07

commerce, the thing that's going to be the biggest part of that is voice. And I think there's a lot that

6:10

can happen where when you're on the

6:17

run and you need to get something

6:19

done on your to do list, you should be able to talk to Amazon.com

6:20

so, you know, people should be able to call Amazon.com and talk to an automated voice.

6:25

And not in the short term, it'll

6:30

be kind of a stilted special purpose

6:33

language for talking to Amazon.com but in the long term it could even be natural language processing. I think that is a real mind bender.

6:35

If you're not watching, there's a woman in the audience that they cut to for like a half a second in this clip who is just like, oh God, it's great.

6:43

So, yeah, he was into this from the beginning. This is his passion project.

6:51

Right, right. Like that is he just described how the thing unfolds.

6:55

Yeah, Especially natural Language like it's, you know. And he said it's crazy. The juxtaposition. He's saying like call Amazon.com, and then he's saying oh, talk to it in natural language. It's like he's getting the old vibes of how still to would be and then the long term vision.

6:58

Totally. I mean you can hear all the ways he's saying this in the year 2000. Right. He still calls it Amazon.com because back then it was still called Amazon.com. yeah. He talks about calling it. This is, this is an era in which a phone. Right. Yeah. The idea of it was that's how you did voice was you called a phone number and things happened. But yeah, so I think it's very clear that he understood this was where things were headed and had been. Had been focused on this for a long time.

7:11

The voice computer was. Yeah. Was very much his vision from day one. I mean the. And the law is that it was Star Trek's computer was what he was trying to emulate eventually. Like the idea that you had an ambient all knowing omniscient computer that could respond to every request. I think he really hit a nerve that people could kind of understand because when this device launched, no one really knew what it was for. But everyone was like, I can see this potential. Yeah.

7:36

And Hayden, call me crazy, but that description I think you can map exactly onto everything the AI industry is in the middle of trying to build right now.

8:06

Absolutely. I mean like truly one for one. That is their vision. And same with. I always hear when I'm talking to sources or AI leaders. Jarvis, same thing, you know, from the Marvel.

8:14

That's the next Generation.

8:25

Right. That's. So it's like they are, they're always referencing one or the other Star Trek

8:26

computer with better CGI is Jarvis and

8:30

able to do a lot of research and pull together like reports and summarize things and also be like your executive assistant in certain other ways that are more like admin y. Yeah, that's exactly what they're always referencing. So yeah, I mean I think that it's. A lot of these people watch this stuff and then they're like wait, why can't we make that a reality? We have the bones. Let's try to, you know, push things forward. And I think that was their inspiration. A lot of times I hear that from the people building this technology. And also, let's not forget Smart House, the Disney Channel original movie.

8:33

I've never seen it.

9:04

You need to see it. You of all people, more than any

9:06

smart house, might have gotten everything more correct than any other piece of media we've ever had. So all of this is the background of this, right? And then I think we get to 2011 when this story really kind of kicks off. And there are two things happening in 2011 that feel important to mention now. One is that Apple debuts Siri on the iPhone, and it is the first time a voice system is becoming sort of a mainstream way to use technology. If you remember Siri from 2011, it wasn't very good. It couldn't do very much, but it was the same sort of thing you're describing. Like, the fact that it did anything at all felt so cool and just. I think it's a useful frame. I just want to play you like a short clip from Apple's debut of this, because you'll see that Apple describes the thing that it's doing almost exactly the way Amazon will as well. And so this is from, I believe, October of 2011. What we really want to do is just talk to our device, ask a simple question, what's the weather going to be like today? And get a response. In fact, we don't want to be told how to talk to it. We want to talk to it any way we'd like. Someone else might ask, will it rain in Cupertino? Or is the weather going to get worse today? Or do I need an umbrella today? And your device, in this case your phone, will figure out what you mean and help you get what you want done. That's a feature in the iPhone 4S we call Siri. Siri is your intelligent assistant that helps you get things done just by asking. This is the dream. It is the dream. And everybody has the exact same one. Everybody still today has this exact same dream. My favorite thing about that Apple announcement, which I went back to, is they. They demo the weather bit like over and over and over again because I think it's the only thing that worked properly. But, like, you could ask it about the weather any way you wanted and it would work. And that was neat and people were into it. So this is happening. And what I think seems to have happened was that a bunch of people inside of Amazon, all the way up to Jeff Bezos, looked at this and said, okay, this is. This is now a thing that is happening. This is. This has been validated. This is out here. We can do this. So Jeff Bezos writes this now sort of mythological email to his team that he says, and I believe this is a direct quote, we should build a $20 computer whose brains are in the cloud that is completely controlled by your voice. He just sends that to his team, which is like a very funny CEO email move.

9:09

Like, typical, let's do this.

11:40

Email habit to CEOs is one of my favorite things. Like you hear the stories about Tim Cook, like forwarding random customer emails to people at all hours of the night. And Jeff Bezos sending like, question mark forwards to people. Unbelievable. And Jeff Bezos, in the process of sending this thing, also makes a sketch of what he thinks this thing might look like. And I just want to show it to you real fast. So this is Jeff Bezos original sketch of what this smart speaker might look like. And it's basically.

11:44

He wasn't an artist, was he?

12:10

He was not an artist.

12:11

It looks like a kid's drawing.

12:12

At least he color coded it, you know.

12:13

Yeah. He has three sides of a box and then some labels for speaker, WiFi,

12:15

Mike, kill Mike M I K E.

12:20

His friend Mike is going to live inside of it. And then he does my favorite part of this whole thing, which is he writes in a box. He says, invention, type in WI fi, password, first connect issue. Like, what a perfect summation of why this thing was going to be vastly complicated. That he was like, oh, God, how are we going to connect it? Uh, so this is, this is immediately the idea. I want to build a $20 computer whose brains are in the cloud that are completely controlled by your voice. And at this moment, it's. It's useful to remember Amazon was basically a company with two things going on. There was Amazon.com, where you bought things, and there was AWS. And this is the time Jeff Bezos is running around all of his teams basically going, what are you doing to support aws? Because this is. This is where the money is. They've turned on this incredible money faucet that is web services. And he's saying, how do we use everything else that we're doing to prop this up so it makes perfect sense?

12:24

Because it's going to be in the cloud, Right?

13:14

We're going to build an AWS box is essentially what he's saying. There are a bunch of people who get this pitch, they run it down, they say, okay, this might be impossible. Let's try it. Because Jeff Bezos wants to. And the thing that I've learned about Jeff Bezos, we did a fire phone episode a while ago. And a thing that is really interesting about Bezos that is not true of a lot of CEOs, is he has these products and projects that he gets really involved in. And he gets so involved, like to the point of driving people absolutely insane. And he has lots of piece for ages.

13:16

One that he was like. I mean, from what I've read, he was in the meetings, like every day. He was on this. This was his main focus outside of, I guess, the Fire phone, which was around the same time. But yeah.

13:50

So that's the other thing going on here is Amazon is starting its own project to develop a smartphone. Y of the Fire phone ends up actually changing the echo story in a bunch of really interesting ways.

14:02

And I think, as you pointed out on the Fire Phone episode, the two teams didn't seem to know what each other was doing.

14:13

No. Yeah. This is one very strange thing, is like, that Jeff Bezos sees Siri and doesn't say, oh, we should put voice on the fire phone. This thing that we're really excited about and think is going to be huge computer, let's do it. And it's like, what if you. What if those were. Yeah. Instead of having bad ideas about 3D screens, what if you did voice? Who. Who. Maybe that would be cool. Anyway, so they kick this thing off, they give it a code name. The code name was Doppler, which I think is very cool. I love a good code name. And Jeff Bezos tells Greg Hart, who is the person put in charge of this product, that it should cost $20 and it should be ready in 6 to 12 months.

14:20

Love it.

14:56

Love the ambition.

14:56

You gotta love the ambition.

14:57

Yeah. He's like, that's not a problem.

14:59

2011.

15:01

Yes.

15:02

Yes.

15:02

Okay. Super, super normal, plausible timeline on which to build a device like this.

15:03

Well, because they were going for the. I mean, the phone voice assistant was a lot easier than this.

15:08

Explain why.

15:14

Because of the microphones, the environment, the way you are talking to a device like this compared to the way you're talking to a phone. And that was the key technology that they had to develop.

15:15

No, that's exactly right. And that is the next thing that happens. So this team goes off basically to solve two technical problems. One is the one you just described.

15:26

It's how difficult to talk to a

15:33

device, especially from far away. And especially. They even knew this from the beginning, which I thought was really impressive, that they're going to have to make it so that it can produce sound and hear sound simultaneously. One of the very first things they identified this would be useful for is as a music player. And as soon as they started talking to people and testing this idea with people, everybody was like, oh, yeah, I'll use it to play music. And Bezos in particular didn't actually love the idea of this being mostly a music player. He was like, it can be a music player, but we have to make people understand that this is much more than that. But it was also a music player. And, like, spoiler alert, it was a music player. But so they. They understand that this thing has to be able to both produce and hear audio at the same time, which is unbelievably hard. Technical problem. And the second thing that they went out to try and solve was latency. And this I actually give Bezos a lot of credit for. And he's. Again, the story of this thing is him giving people insane tasks to accomplish. And at the time, if you remember, voice latency was like two or three seconds. Between you said something, it would process it, it would go run the query in the cloud, it would bring it back, it would process it back to speech and then it would say back. That was like two or three seconds. Latency felt bad, but it was still new enough that it was magical that it worked at all. But Bezos was like, one second. It has to be one second. And so they went out and they were like, well, that's impossible. The state of the art of technology suggests that we can't do that. He's like, I don't care. Do it anyway.

15:36

Yeah. And apparently, I think when he was testing a mod at one point and he said he was so frustrated with how slow it was, I think he said something like, can you shoot yourself in the head? And I was like, I've heard, I've. I've said similar things to my Echo devices over the years, because when. When you're dealing with something on a computer or on your phone, there is more resiliency to latency because it's a. It's a connection that you are interacting with as opposed to if you're walking around your house or you're expecting something to happen and instantly from a device like this, you expect it to do a lot more because, I don't know, the environment that you're in, there's something about it being a communal device as well, rather than a personal device. It just has to be fast and work and either of those things don't happen and it becomes a useless device. Whereas your phone and a computer, you use them for so many different things. But that's those parts, those two elements were so key to the success of

17:01

a product like this.

17:57

It is the whole interaction model, right? Like, without it, it doesn't matter what it can do. It'll be annoying to use, and people won't want to use. So to speed this thing up, they go out and buy two companies. One is called Eevee and one is called Evona. And basically they are working on opposite ends of the speech to text spectrum. One is working on synthesized speech. One is working on speech to text. And together they're like, okay, we've built some of this technology, and we can start to really make this work. So Bezos has a bunch of, I would say, very good ideas. This red and blue ring at the top of Alexa that would sort of alert you to what it was doing. That was apparently a Bezos idea. The name Alexa evidently came from Bezos. It wasn't his favorite name, but it was apparently his idea.

17:57

Do you remember what his favorite name was?

18:38

Well, I do, because I looked it up. I did not remember until I looked it up. Hayden, do you know what this is?

18:39

No.

18:44

Jen, what was it?

18:44

Amazon.

18:45

He loved calling it Amazon.

18:46

Oh, my gosh. Of course.

18:48

It was like he thought it was such a cool idea to call it Amazon. And a bunch of people who worked there were like, well, isn't it weird to talk to a company? And mercifully, they ended up winning that fight.

18:51

Do you know why they chose Alexa? Speaking of, we have to comment on my shirt that my children gave me. Alexa, bring me wine. Because when you heard me say that, Alexa, bring me wine, what consonant stood out the most?

19:01

The X.

19:16

Yes. And so that was like, they needed something that was gonna be very key so that the microphones could pick it up. And that made. That was really. It had a cute idea behind it that it was connected to Alexandria, the library of all knowledge. But really it was because it was a good word. Just like when you choose a name for your dog, you should have an S in it. Yeah.

19:17

Or an E at the end. I always like. They pick up on the E sound at the end of dog names.

19:39

Interesting. I also learned that Bezos liked the name Finch as a wake word that they ruled out. He also really liked Samantha as a wake word, which they ruled out, but they eventually came around to. It was until the very end, it was potentially going to be Amazon.

19:45

Amazon. And you can still. You can call it Amazon. I think that's one of the wake word options.

20:01

At first it was only those two fetched. It was just, you could do Alexa or Amazon, and then they've added a bunch. But I think Amazon is still an option. Yeah, I Don't know anyone who don't choose to do that, but, you know, knock yourself out.

20:07

I think Google did it.

20:16

That's true. Amazon is just too many syllables.

20:18

Yeah, it just wasn't. I think that was really what it came down to. Like technically it needed to be a more, A more powerful word that kind of came. Came through different.

20:20

They were really worried about. This was also. Amazon was really starting to grow as an advertiser and they were terrified about the idea that Amazon. You might hear the word Amazon on television and your smart speaker would just start buying things.

20:30

They were terrified. Really?

20:43

Well, fair. There were people who were terrified. Bezos might have liked this idea. Yeah. So there are all of these complicated things going around and they eventually land on Alexa, which pisses off everyone named Alexa in the entire world. As far as I can tell, the

20:44

name Alexa as a baby name just like plummeted after the device came out.

20:57

I mean, of course, it seems cruel to name your child Alexa now.

21:01

And I think it's important to also point out that there was a difference between the name of the voice assistant and the name of the device. And that's sort of always been a complication, I think since the device launched. Like, I think most people just call it an Alexa, even though it's. It is an echo. Although apparently it was almost a Flash.

21:04

It was almost a flash to the

21:26

point where packaging with Flash on it. Amazon, Flash. I kind of like that.

21:28

I like, I like Flash better too, you know, but it is funny that like throughout the Echo, its history, I feel like any noun that you can think of, they tried putting it with echo. Like the echo. Show the echo. Like there are so many words that they put with echo. And it's like with Eko, the second word they always added. And then Alexa. There's just a lot of, a lot of words floating around about how to talk about this thing.

21:33

Totally. So this is where we get to the data collection piece of the story. And they're figuring out the hardware, they're figuring out what it's useful for, and they have to make a technical decision. And this is very much a sort of inflection point in the history of a lot of this AI work. So you can basically do one of two things. You can either take a huge knowledge graph of sort of if then statements, answers to questions, and point Alexa at that. Right. And so it'll say, okay, they're asking about the weather. Here is how you deliver the weather. It's very much like a call and response system. And you can make these fairly sophisticated. Right. But it goes in and it's a very simple system that says they're asking a question about the weather. Here's how you deliver an answer about the weather. And that is what a lot of people want them to do. This is like relatively the state of the art technology at this point is like just point a bunch of really complicated if then equations at speech and you're off and running. The other option is this growing idea of machine learning and all of the words that we now call AI that are actually a vast vocabulary of different ways of training a computer to understand things in more deep ways. This is much harder technology, especially in 2012, 2013. It's much newer technology, but it is clearly more powerful. If you can do, honest to God, natural language processing, neural networks, true sort of deep machine learning, you can teach this thing to do much more. This is sort of the time that this technology is becoming the state of the art everywhere. Right. Like deep learning became a thing that we all started talking about professionally kind of right around now, and is like very much leads directly into all of the AI stuff we're talking about now.

21:55

Absolutely, yeah. It is crazy. And it's funny that like even a couple years later with routines and Alexa, everyone kept saying, oh, like, you know, echo is way better at routines than any of the other smart speakers that came later. And I think it's just because of the if this then that rules that you were just talking about. I mean, it was just good at that.

23:43

Yeah.

24:02

So what's funny is that's the kind of stuff you can program to do very well. Right? The very simple. You say these words, it executes this action. To do a bunch of natural language processing and neural network work is actually vast overkill for that. But again, if you go back to that thing that Jeff Bezos said in 2000, it's like what you should be able to do is you should be able to interact with this like it's a person and it should feel normal and human and you can talk to it however you want. And again, this is the dream. So, okay, so this Amazon goes through this whole debate picks essentially investing in natural language processing. They're like, we're going to do the hard thing and then runs headlong into, oh God, we need more data. And we're going to get into how

24:02

lots of words, lots of words.

24:45

We need so, so many words. And they solve it in some really fascinating ways that we're going to get into. But first, let's take a break. We'll be right back.

24:46

Wallet feeling extra light after the holidays. Yeah, same. But recovery starts right now with TikTok slash and free.

24:59

Here's how it works.

25:07

Pick the products you want in TikTok, shop, share the link and watch the price drop all the way to zero. No tricks, no catch. Just free stuff with free shipping. Download TikTok search, slash free and start slashing. Today,

25:09

K Pop Demon Hunters, Haja Boy's Breakfast Meal and Hunt Tricks Meal have just dropped at McDonald's. They're calling this a battle for the fans. What do you say to that, Rumi? It's not a battle.

25:26

So glad the Saja boys could take

25:36

breakfast and give our meal the rest of the day.

25:38

It is an honor to share. No, it's our honor. It is our larger honor.

25:40

No, really stop.

25:46

You can really feel the respect in this battle. Pick a meal to pick a side and participate in McDonald's while supplies last. All right, we're back. So we've left off with Amazon knows what it wants to build it knows how to build it, and it discovers that the way to do it is to just throw an unbelievable amount of data at the problem and they don't know how to do that. Hayden, this, this particular problem has essentially taken over the tech industry over the last decade. Can you just describe very quickly the, the data problem with AI training and machine learning and this whole thing? Like, why is this a giant data problem for everybody?

25:48

Totally. I mean, I think it's just because like these things don't work unless you have the right data. And they realize it's not just good data, it's very niche specific data for literally everything because they don't have common sense, they're not that good at transmuting things over or like, you know, expanding context and being like, oh, if this, then that in a data sense. So basically like, you know, now we're seeing today like a lot of environments or gyms for agents to learn very, very specific tools and words and how to work very, very specific platforms. But even if you they were already trained on another pretty specific similar platform, it won transmute over. So I think like, where a lot of these models are really jagged and so they're really good at like complex things and really bad at some simple things. And they were starting to see this way back then. Like, you know, now we're seeing like improv actors being hired to train these models just to create data that's very, very specific or like create really intense, like rare situations that they can train on. So I think, yeah, way back then they were starting to realize, oh, it's not just getting a lot of data, it's getting a lot of data that is going to be exactly relevant to the types of situations that these people are going to be in. And that's why, you know, that's so important for natural language processing. Because, yeah, it's like if you're mumbling or if you're kind of distracted or if you're speaking in a different accent, like, you know, it's just these systems have to like have a perfect example in order to really understand you back then.

26:26

And the way tech companies would normally deal with like beta or alpha products is get lots of their employees to use them. But employees, lots of employees in Silicon Valley or lots of employees in Washington State are going to give you a very small slice of life, not the huge slice of life that the Echo eventually ended up being in everyone's living rooms. So, yeah, there was a challenge that they had to overcome and it was very ingenious how they did it.

28:03

So, yeah, so Amazon decides to solve this in arguably the funniest possible way. Basically the other thing you could do if you want to just catch up quickly, is license this data from somebody else. And at the time the company to license from and lots of companies were doing so was Nuance. Nuance is like an OG of the speech to text space. They made Dragon naturally speaking a million years ago. A bunch of the people who founded Nuance were like at Xerox park working on this stuff in the 80s. That's the company at the time if you wanted to do this. But Amazon decides not to do this because they really want to control their own destiny. They're like this, we think this is a long term project. We don't want to rely on some third party contractor to make this work for us. So they decide, okay, we need to catch up as quickly as we possibly can. And they go to Bezos, a bunch of the leaders of this team, they go to Bezos and they say, okay, give us more people on our team and we can speed it up. And there's this very famous moment where Bezos goes, you're telling me you've turned a 40 year problem into a 20 year problem and that's good. And they're like, well no, that's not. And he's like, do better. So they go out and they just, they design this amazing experiment. And Hayden, I just want you to imagine you're a subject in this experiment and the way that it works is you are led into a room, a room, honestly, that probably looks something like this one. It's, it's filled with gadgets. Lots and lots and lots of different kinds of gadgets. And a bunch of the gadgets are covered in cloth. You can't see them, but you, you can tell there's something under there. These are prototype Amazon devices, but there are also a million decoys around. So it's designed to not be exactly clear what's going on. But you come in, you sit down, you're given, I believe, some kind of tablet. I think it might have been an iPad.

28:33

It sounds like an escape room,

30:14

but in the most dystopian possible way. Because the thing that happens next is you are asked to spend your whole day, like hours and hours and hours and hours, saying out loud the things that are on your screen, just reading prompts to what were prototype Alexa devices. And what's happening on the other side is they're, they're collecting all of this data. They're, they're sort of running two simultaneous experiments. One is they just want a ton of voice input. And again, like you're talking about Jen, this, this far field input is really important. Hearing it from across the room is really valuable. And that's not a thing anybody has. Right? Like, nobody has done this kind of thing where how do I recognize speech from 10 or 12 or 20ft away? So they're getting all this data, they're getting lots of different voices saying the same things, and they're just brute forcing their way into people, quote, unquote, normally using these products. And then on the other side, they're running this experiment that they call the wizard of Oz, in which people are testing a device, but when they give a request, they're not getting answers back from Alexa. There's actually a person on the other side typing furiously as fast as they can the answer to your response, which then comes out via Alexa. So it's, they're doing the sort of text to speech out, and then the person on the other side is asked to rate the response. And so what they would try is they would try lots of different cadences and different word choices and different ways of answering the same kind of question to see how people reacted. So Amazon is as aggressively as it can, testing sort of every single part of this pipeline, while also collecting every bit of data it possibly can.

30:17

Calling it the wizard of Oz is really genius. That's the perfect name for that.

31:57

It's exactly right. It's exactly what it is.

32:00

It's behind the curtain.

32:02

Right. And so they start doing this thing. They end up doing it in a dozen cities around the US with thousands of people, and they collect just a vast amount of data. And one funny story I heard is that there were apparently so many people coming and going from these rooms all day that people started calling the cops because they thought either brothel. They either thought it was a brothel or a drug den, basically. They were like, why are all these people just coming and going from this apartment?

32:03

Pre Airbnb is Right, Right.

32:30

Yeah.

32:32

Back when that. Now it's normal to see strangers going into the same apartment every single day. But so this. This becomes the way that they catch up, and they start to catch up really fast. And actually this becomes an incredibly successful data collection experiment. And they start feeding it into these natural language processing systems. They start running the neural networks, and you can start to actually see this assistant get better quickly. This is a very, like, I think about this now where like, AI companies are just like, you just download all of YouTube and your problem is solved, which is what they all did. This is such a brute force, manual way of trying to solve the same kind of problem. It's just very funny how dangerous. Much changed in 10 years of technical development.

32:33

Yeah. I love the lo fi vibes of people just like literally saying this stuff and then someone else typing it. Incredible.

33:13

It's great.

33:19

Well, and that was. What was interesting here, was once this was out in the world, it could learn and get better.

33:20

Yes.

33:25

And that's what we now are, you know, becoming very used to. But this, this was. And this was something I think that they kind of presented to Bezos too. It's like, now, this is the capabilities we have. But once it's out there, once we have hundreds and thousands and millions of people using this, it will get better. And it will get better specifically for them as well. Like. And that was kind of a pretty new concept here.

33:26

Yeah. And I think it was hard for Amazon to grapple with because it was. It requires shipping a thing that is not good or not good enough at the very least. Right. Understanding that. Okay, the only way this gets better is to put it in front of lots of people, which is now what

33:50

technology companies do all the time.

34:06

Everybody Beta ChatGPT was a research preview. Like, this is. This is just what we do now. And there's no shame in it anymore. But back then there was. I think Bezos in particular had this idea of. He just kept saying it wasn't good enough.

34:08

Right.

34:19

And he was. And he.

34:20

The.

34:20

The Question he apparently ran around asking his whole team is, when is this thing going to be any good? Which, as a product person, is such a brutal question. Yeah. The answer was both immediately and never simultaneously.

34:21

Yeah.

34:33

But anyway, so they're still going through, doing all this testing, starting to put it in front of lots of people. Like you mentioned, a lot of Amazon employees have this thing at their house now, and the thing they hear over and over and over as they talk to people is, people want music. People look at this thing, and the immediate thing that they think is, this is for playing music. So the team starts to go, okay, well, great. This is our first killer app. It's going to be playing music. The first version that Bezos wanted was like, a hockey puck size, which is actually sort of fun. Jen, you also brought.

34:34

Is this the original echo.it's the original.in the little kid's case, but that is the original dot.

35:07

So I think. I think based on everything that I've read, I think this is originally what they wanted to ship.

35:12

And that does look a bit more like the sketch.

35:19

Yeah.

35:21

And, I mean, it's basically the top of this, isn't it?

35:22

Yes.

35:24

And then they added a big speaker.

35:25

Yeah. It was supposed to be this, and it was supposed to be 20 bucks, but the team looks at it and says, okay, well, people want to play music on this. We should make it good for music.

35:27

So we needed a little bit more of this speaker.

35:35

What you need if you want music is you need space for sound to move around. Right. The reason the Echo looks like a can of tennis balls and not a hockey puck is because it's a music player. Like, very straightforwardly, the problem.

35:37

It did good.

35:48

It was fun.

35:48

We didn't have. What is this, 2014? We had Sonos just. Yeah, but those were really expensive.

35:50

They were.

35:56

So, I mean, for what it did, compared to your phone, it was definitely better.

35:57

Better I will give you. Better and good are not the same thing, but better I will give you. Bezos, like we mentioned, is not into the idea of this being a music player again. He has this big idea about the Star Trek computer, and he wants to build the voice interface of the future. And he thinks that if they ship it and say, this is a really cool way to use your voice to play music, it will be pigeonholed that way. A, I think there's a reasonable theory behind that estimate. But B, a very funny thing that I discovered. There's a great Business Insider story about the history of this, and one anecdote they have is that the whole team thought that maybe the problem was that Bezos wasn't a music guy. And the reason they thought that is because there was a time that Bezos came in to do a demo and he asked for a song from Battlestar Galactica, which he told the team was one of his favorite songs. So the Echo team is just like, oh, okay, this guy just doesn't. He just doesn't like music.

36:05

Died in the wool, nerd.

36:57

Yes. Yeah. They're like, this guy is just not a music person. And that's okay. But anyway, so the other thing that happens here is the way that they changed the speaker also makes the whole thing more expensive. They had already gone from we want this thing to be $20 to the price goal was 50 bucks. And then they make the speaker better and then everything gets a little more expensive and then everything gets a little more expensive and then they need more processing to do the voice stuff. And so the price just keeps kind of ratcheting up. And as they're making all these changes, the date continues to slip. This thing was apparently supposed to be shipped in six months for three years. Every six months. It was six months away for three years.

36:58

Reminds me of the Siri upgrade promise.

37:38

Yes. Oh my God. It's really. It's a. I can really resonate with being six months away for three years. Like that just feels right. They are working on all of these problems. Apparently the biggest ongoing problem was this making it hear sound and play music simultaneously.

37:41

Right. Because once it's playing music, it then still has to be able to hear you. And that sort of juxtaposition was very challenging. I think at one point they just got. When you started to talk, the volume now turns down so that it can hear you. But I'm not sure what the. How they solved that in the first go round.

37:56

The idea was because it understands the audio that's going out, it can just sort of negatively process, which is, it turns out a very hard technical problem that didn't work very well. But okay, so now it is the fall of 2014. This thing is finished. They've finally done it. It is ready, it's ready to ship. But there's a problem which is that a few months earlier, Amazon shipped a phone called the Fire Phone and it sucked and everybody hated it. And you really cannot overstate the extent to which this thing throws the whole Echo project into peril. The Fire Phone almost immediately goes on huge sale. They end up taking a huge write down on is like maybe the first real Spectacular failure of Amazon. And it's a big deal. This is like Jeff Bezos had done this big launch event. They framed it as like the next big thing they were going to win. This is another one that Bezos had been intimately involved with from the very beginning. He had a lot of FE ideas that were not good ones. And so this really just shakes the confidence of this whole company in a, in a very real way. Some of the Fire Phone team leaves, some of the firephone team goes to the Echo team. And all of a sudden there is a feeling that the Echo is not just like a neat thing that we're making. This is now existentially important to our jobs.

38:14

And the only thing, the main thing they were the lab was working on.

39:29

And if they, and if they blew it twice, like there was this idea that maybe this whole project is just dead. Like maybe, maybe we go back to being a Kindle company or maybe we don't even make hardware anymore.

39:32

Do hardware. Yeah. That's an interesting alternate universe, isn't it? If Amazon just remained a software company and never went into hardware.

39:42

And I think it's not implausible, like to take two hits that big right in a row, become pretty hard to go back to. And I will say to Bezos, credit in particular, every indication is that like a thing he always said and is sort of a core tenet of Amazon is they're very happy to fail. And in general, I think the transition from the Fire Phone being a disaster into continuing to bet on a lot of the same people to do the work with Alexa and Echo is like a real credit to Bezos, actually, that he didn't run away, he didn't back off, he didn't get afraid. He kept pushing on this idea. He was like, well, we screwed this one up, let's not screw up the next one. Keep going. But there's a reason that when the Echo launches, it launches the way that

39:49

it does, which was really quite notable in how unremarkable it was. So it was a real stealth launch. There was no big showy keynote, there were no review units, there was no heads up, no embargoes. It was just dropped a press release and here's the product. Although I don't believe the product, as I mentioned earlier, didn't actually launch for a few, few months and you had to sign up to be allowed to order it, which kind of was the first of this, of Amazon's day, day one process where they kind of vet the people that are allowed to own their products. And apparently they did vet quite heavily anyone that ordered it. And they only let certain people get one. So apparently they had 80,000 was their initial kind of run that they were going to allow people, 80,000 people to buy it. And they're huge, I believe. Although there's no numbers have been officially released, there's a lot of speculation it was around 110,000 people actually signed up

40:34

for it like right away.

41:34

Right away. So they were like, oh, okay, there's some interest here.

41:35

Yeah.

41:38

So like you said, this thing gets tons of pre orders. They, they actually have to work pretty hard to catch up to the excitement around this thing. I think Amazon expected this to burn a lot more slowly or was at least prepared for it to burn a lot more slowly than it was.

41:38

It was expensive too.

41:49

Yeah, it did. It was. It was 99 bucks if you pre ordered it.

41:51

And 200.

41:53

Right. Which is a lot.

41:54

Yeah, that's how much I paid.

41:55

That was 10 times what it was supposed to cost. I will remind you, the $20 voice computer was now $200 and a printable

41:57

jig was more expensive.

42:03

What is a Craig Jake?

42:04

No, I don't want to know. Never tell us. Under no circumstances should you tell us what it is. So this thing exceeds all of Amazon's expectations. And Bezos is like, starts feeling himself again and he's like, we, we did it. This is, this is right. We are now, we are officially onto something. They've moved on from the fire phone. It's also important to remember that this is the moment the mobile companies, particularly Apple and Google, are really starting to like, shore up their worlds. Like the App Store is becoming more tightly controlled. Apple is starting to make a lot of money by being the app company. Google is doing the same with Android. And there is this sense of like, we need to invent the next platform. Everybody has been trying to invent the next platform since then, but Amazon is like, okay, we want to have a new thing. So Bezos decides to do a bunch of things. First of all, he decides to pour absolute fuel on the fire, right?

42:07

He, the fire phone.

42:56

Hate that he burns the fire phone and then decides he wants to like really go hard after this thing. So he demands new features every week from the Alexa and Echo team. He immediately decides in an interesting way that actually Alexa and the Echo shouldn't be the same thing. They should be different things things. And that Alexa should be a platform that can work on other hardware and be sort of a platform all its own, which I think again is. It's the thing you do when you also have aws, but it's also just very smart. They're like, we think this can be more than a speaker in your house. We think this can be an ecosystem. And they saw that very quickly in a way that I think was actually pretty impressive. And they decide, we're going to do an app store, and they go chase this idea of skills, which is other things that third parties can bake into Alexa to make it do more stuff than Amazon can itself. And one thing I found very interesting was that all the way up to Bezos, Amazon kind of immediately understood that people were going to have a hard time figuring out what this thing could do. And so Bezos's demand was, you're going to email Alexa users every week with all of the new things that they can do. And I have, like, traumatic memories of all of the emails I used to get from the Echo team describing new features that were coming to Alexa.

42:58

I mean, they had the app, which is still not a great experience, not

44:11

at first, but there was eventually an app.

44:15

There was eventually an app. When did the app arrive? Fairly quickly, yes, because app immediately became a problem that you didn't have an easy way to understand what it was capable of doing and also to interact with it beyond very specific nomenclature and commands. 2014 was also around the time where we had the Nest thermostat, We had Philips Hue. So it was. We had a few big names in the space, but it was still app control. That was the only way you could control your smart home. So you would have your thermostat app, you would have your smart lighting app. And that was really frustrating for users because it was like, every time you want to turn on your smart lights, you get out your phone. That's not great. Voice is like, wow, okay, now I can sit on my couch and say, adjust my thermostat, turn on my lights. And pretty quickly after it launched, this is when we ended up bringing in the Zigbee radio so that they could connect devices directly to the Echo. And it became a hub for your smart home. Now you've got the real Star Trek computer. Now you've got t. Oh, Grey Hot. Potentially in your future.

44:16

Yes. I think that inkling seems to be the thing that convinced Bezos and the whole team that there is something much bigger here.

45:27

Right.

45:35

That, like, not this. This can be the center of an ecosystem and sort of a whole universe of voice computers. And everybody now starts to talk about the ambient computing idea. And again, cannot overstate. This is Everybody's idea. Apple is working on this at this point. Google is starting to work on this at this point. Microsoft gets weird about the ambient computer thing at various times, but, like, this starts to feel like it is going to be the next thing and all of these companies are just going to march towards this. And then if you Fast forward to 2016, 8 million people have either an echo or an echo dot, which is a huge number. They've shipped this, the echo dot, which is much closer to the thing that Bezos wanted at first. It's also a big hit. And then I think basically our story ends in 2016, also at Google. I owe. When Google launches the Google Home. And I should point out that basically the minute the echo comes out, people start being like, well, what's Google gonna do? Google's doing Google Assistant. It is clearly after some of these same things. Google Voice exists like it is. It is pushing in all of these same directions. And everybody's like, where's the Google speaker? And in 2016, Google launches the speaker. Let me just play you. This is a little tiny bit from Google I O where they explain what the home is going to do. Google Home lets you enjoy music and entertainment throughout your entire house, manage everyday tasks more easily, and ask Google what you want to know. With Google Home, we set out to create and design a beautiful product that's warm and inviting and fits naturally in many areas of the home. We think it'll be a beautiful addition to any room in your house.

45:35

Beautiful addition.

47:16

This is like. I mean, he's. He's saying that at Amazon, right. Like all of that. There's a giant Pringles. Can we made a nice one saying,

47:17

yours is so ugly and ours is beautiful.

47:25

Very much. It was. It was a lot of shade being

47:27

thrown in that one. Yeah.

47:29

And to be fair, Google's smart speakers have been. They've looked better from the beginning. They've always had a better design team. The first one was a bit too much like an air freshener. Agreed, Agreed. But it had nice touches. It had that little sort of sloped top. You could sort of see it sitting on your kitchen counter. A little better than this giant black tube.

47:30

So this is actually. This is exactly where I want to leave this story. Right. Because the smart speaker thing is off and running. But I think Hayden here is my question for you is at the same token, right, like, this thing was enormously successful in a certain way. A lot of people had it. It mainstreamed this idea that you can talk to things by voice, but it didn't solve our smart homes. Nor does it seem to me that this is the beginning of the AI revolution. Like, in a weird way, you can draw a straight line from the ideas behind this thing to the ideas of everything that we're doing now. And yet Alexa is not the thing. Why, why didn't this work the way that I think Jeff Bezos thought it was going to in 2014?

47:56

I think for a lot of the reasons that you guys have said. I mean, you know, it was seen as largely a speaker, a music device, you know, despite Bezos's wishes. And also the fact that it wasn't super good at context, it wasn't super reliable. I mean, I think it was more like a symbol, an idea, a way to unlock the imagination of what could be, rather than like in and of itself, an example of what AI could be. You know, I think it was an important milestone, an important step, but it didn't like, push the actual, like technology behind the device forward in a huge way, in my opinion.

48:37

Yeah, it felt like they created an Alexa ecosystem and hundreds of devices. There were so many that came after this one. But the core technology behind Alexa never really seemed to change until recently. I mean, I'm sure that's, I don't mean to discount decades of engineers work, but it just never, it never really fulfilled its early potential because it felt, I don't know what from the business side, but it was like, now Amazon needs to monetize this device. And by. And rather than by making the core product better, it went, it seems to me it went with the tact of let's put it everywhere and hopefully everyone will buy these devices and then spend more money at Amazon. That was kind of, that seemed to be the monetization strategy. Whereas maybe making a better technology from the beginning and iterating on that, but. But was it possible? Would it have been possible to make what we have today with LLMs?

49:18

Hold that thought because that is a perfect segue into the first of the version history questions. Okay, because I have a theory about this and we're going to get to it, but let's take one more break and then we're going to get to the version history questions. We'll be right back. This episode is brought to you by Indeed. Stop waiting around for the perfect candidate.

50:26

Instead, use indicators.

50:46

Indeed Sponsor jobs to find the right people with the right skills fast. It's a simple way to make sure

50:47

your listing is the first candidate.

50:52

C. According to Indeed data, sponsored jobs

50:54

have four times more applicants than non sponsored Jobs. So go build your dream team today with Indeed. Get a $75 sponsored job credit@ Indeed.com podcast. Terms and conditions apply. This episode is brought to you by Welch's Fruit Snacks. Big news for your kids lunchbox. Welch's Fruit snacks are now made without any artificial dyes. A snack parents can feel good about and the same delicious taste kids can't get enough of. All made with no artificial dyes. Try Welch's Fruit Snacks today. All right, we're back. So we ask the same eight questions about every product. These are the version history questions. And Jen, what you were just talking about is essentially the first question. This is the time matrix, a concept that everybody understands perfectly and has no problems with. And this is we map ideas and time and is this the right idea at the right time? Was it the wrong idea at the wrong time, or was it something else? And I think I have a very strong opinion about this, so I'm just going to go first and you guys can tell me what you think. I think this is the right idea at the wrong time. I think this. I think you could make a strong argument that the Echo and Alexa were both too late and too early. That if it comes along before Siri, like, if they managed to make this happen in 2009, that even the very basic call and response technology like you're describing, this very simple, I say Alexa weather and it tells me the weather would have felt unbelievable to people and it would have seemed like a huge step forward and it would have given Amazon this, like, important first mover advantage. That's one way. Or you wait until 2017, you read the attention is all you need paper, and you go, oh my God, this is the technology that's going to let us do it. And then you start from there. Whereas what happened, as far as I can tell, what happened is Amazon was so deep into Alexa that it looked at this new kind of technology that was coming out and it looked at LLMs, which in principle basically solve every problem. They've been like, everything the Echo does. LLMs would have let it do better. And they had this massive amount of tech debt and this massive amount of stuff they had already built that made it very hard for them to catch up to the point where they had a regime change and they had to come in and essentially reboot the whole project from nothing. And so to me, it's like if they had started five years earlier or five years later, I think in either case, the product is probably better today. That's my theory.

50:57

I feel the same way because I think this was like, you know, they really invested in the shell, the wrappers for their technology over and over for years later we saw every noun you could think of was thrown with echo. And you know, we had the so many different devices, but I think they didn't iterate as much on the actual technology as Jen was saying behind it. And I feel like when it came out it was, it made waves for sure. But if it had been earlier or later, it just would have been better. Google waited a couple years, came out with their answer to this in 2016 and it was already better because they were a little more intentional with it. It was better at follow ups later. It was better at context, it was better at understanding you in different situations. So I think it was interesting because Bezos was rushing them so much, but if he had rushed them either more or less, it might have been better.

53:19

What do you think, Jen?

54:10

I mean, I think it's hard to discount what a cultural moment it was. So I think it was, it was definitely a. I think it was the right time in terms of creating a device that brought ambient computing, speaking to your device, hands free interaction to the mass market. Not just, this was not just a tech device. This was not just for tech geeks. This is why I have a T shirt that says Alexa, bring me wine. Because it became a cultural moment. Everyone knows what Alexa is.

54:11

Yeah, but everybody knows what Siri is too. And Siri beat it by three years.

54:48

But not everyone uses Siri.

54:52

No, everybody hates Siri.

54:54

In fact, I mean, I feel like of all the tech products in the smart home space, this is the one that most people I know have or have used. Yes, but in terms of right idea. So I think it was the right time. But as I've said, I think the idea was half baked. Interesting, because they knew they wanted a voice computer, but they didn't really know what they want or they didn't give it the capabilities that perhaps the vision had heralded. Like it's not Star Trek's computer. And that was the vision.

54:55

All right, well, I think you've made a very good case, but I think we're going right idea, wrong time. But we're only going gently. Right idea, wrong time. Hayden and I win 2 to 1. But you made such a good case that we'll move it over. All right, question number two on the version history questions. Was the Amazon Echo Peak anything?

55:31

I think it was Peak like hands free home device marketed to anyone. You know, I mean, at the time I, I Guess not. Still no.

55:47

That's a good one. It was peak voice computer excitement because people immediately started to get burned by these things and the idea of how fun and good they were, it got less magical as it got more real. But at this point, this moment was like the reason they got all those pre orders. It was. Oh. People were like, oh, my God, this is amazing.

55:55

I can control this with my voice. Yes. Yeah, yeah. I mean, yes. Although, I mean, in terms of peak, that means there's a trough. Has there been a trough? I mean, yes, because it didn't. As I've said, I felt like it never really got better. I mean, that's a bit dismissive, but in terms of what I hoped for, it was an exciting time when it came out, everyone. But I mean, peak, I think feels a little too strong because I feel like we've got there's peak to come when it comes to voice computers.

56:12

Okay, can I offer you a couple of other ones? Was this peak Amazon Hardware?

56:44

Oh, God.

56:48

And I can offer you the trough, which was like 10 minutes earlier when the fire phone came out. Has there. But think about it, like, since. Has there been a more exciting Amazon hardware launch since this one?

56:49

Oh, no.

57:01

I think the answer is no.

57:03

Yeah, I think it's peak for sure.

57:04

I think this might be peak Amazon hardware.

57:05

Yeah.

57:07

I mean, like, it's not the best Amazon hardware, but in terms of like Amazon's place in the hardware universe, this was. This was the coolest it ever seems.

57:08

Only really, well, the Kindle, but it's real hardware. Shining moment. Yes. Yeah. Yes.

57:15

And the Kindle was such a slow burner that, like, the Kindle was. Was cool and exciting, but it was, it was enormous and had a keyboard at the beginning, like, it took him a minute. Whereas this just sort of dropped from the heavens and seemed incredible and people were super into it.

57:22

Yes. I would say we talked about this earlier, about how whether Amazon was a software or a hardware company and this was in terms of its lifecycle as a hardware company. Yeah. I would say this is peak Amazon hardware.

57:34

Okay. All right, I like it. Question number three is if you could time travel back, knowing what we know now, and develop this thing yourself, could you make it more successful?

57:45

Well, I would start with a smart home.

57:53

This is what I was going to ask.

57:54

If I could go back in time, I would build local control into this thing from the start, which obviously is antithesis to what aws, but this was. Where so many of the failures came, was the skills. And relying on APIs, rather than building any of these functionalities directly into the device. I know David and I, we've talked about universal remotes in the past and how wonderful and frustrating they are. But this is a universal remote with your voice. But it needed those capabilities from the start to be solid and reliable. Because as to your point, once it doesn't work one, two, three times, you're done. You're not. You're never going back.

57:56

Yeah, I think I could go back in time and direct them to do. To do better, but not. Not do it myself.

58:38

Would you go back in time and be like, listen, Jeff, wait a few years. There's, there's, there's some researchers at Google. They're about to blow your mind to bits with what this thing can do.

58:43

That's what I would say.

58:53

Just hold.

58:53

I mean. Or, yeah, it's like, if you can't do it earlier, I would say, yeah, just hold on and do it a little bit better. I mean, like, you know, it was a big, huge milestone moment with a lot of hype. But I just think from an AI perspective, yeah, I would have loved if he could just wait a little bit.

58:54

Yeah. I think my only answer to this is I would go back and like, grab Jeff Bezos by the lapels and be like, make a music player. It's fine. It's a music player.

59:08

Lean in.

59:16

Yeah, like, let this be music and timers. And it's fine. Don't require it to be anything else. Cause they were like, they went and did Amazon music and there's like, there's a world in which Amazon had a long road of interesting stuff to do in that space that it didn't do because it wanted to be a voice computer. And I think if you. They could have done the music thing long enough to get to voice computer at the right time. Yes, I would have been fired from Amazon long before this happened if I was running this project, but that's what I would have done. Question number four. Will the youth ever make it cool again? This one doesn't really apply because I think the youth still think it's super cool. Like, not the specific echo, but I think, like, no, the youth are not gonna make this echo cool again. This has no chance of a, like, retro nostalgia. Pringles can comeback. No, but it does seem to me at least that kids are still a huge and important market for smart speakers in a way that is really, really valuable.

59:17

I think it depends on what youth you're talking about. Like Gen Z. No, they would never. I think, like, they are not into the surveillance vibes. They're not into the like, AI being everywhere vibes. I mean, maybe like, you know, like Gen Alpha.

1:00:08

I literally have an article I wrote about how I used Alexa as my third parent and my daughter. You know, I had bedtime routines and everything set up. I think that generation will go one of two ways. They will either be traumatized by it and never wanna see one in their house again. Because it would used to tell them things like, you have 10 minutes before bedtime. You know, it's like the evil nanny.

1:00:22

This is why you have to watch Smart House.

1:00:44

And then. Oh, but also, they've also become so attuned and accustomed to just saying play this or do this, rather, you know, they've. And they use their phones for it too. Like that generation, my children's generation, they don't use keyboards. It's all voice input. They're just used to using their voice as a. To control computers in all different forms. So I think it's going to become even more a part of. Of their life going forward.

1:00:48

Question number five. What feature of this thing should every current version have? This is a fun one for Echoes in particular. Is there anything that we lost from the original Echo? Do you think we should put back.

1:01:13

It has a physical volume control and the Echo dot has this, but I think it got lost after that. This was the only generation where you could physically not with buttons, but there's an actual. I'm spinning the top of the device and it's turning it. Turning the volume up and down and

1:01:25

it has the light that goes with you as you turn it. It's such a lovely little interaction.

1:01:42

I love that. And they got rid of that. You have buttons now that you press, but they never seem to respond well. And I just love. I love physical controls on. On tech gadgets. And I would like to see that in all of. I mean, it doesn't really fit with the current design, which is a weird. I'm not even sure what kind of how you would describe the shape of the. The latest Echo dot, but yes, I. I loved the physical controls and they. Every generation of Echoes seem to lose more of those physical controls. The idea of it being voice first. Right. You're supposed to always use Voice to control it, but sometimes you just need that physical control.

1:01:46

I agree. I was going to say the exact same thing. And not just the ring, but the whole set of physical controls here. So we have this original Echo, which is. Imagine a black metal tennis ball can. And that's it. You've done the job. But there's two buttons on top. There's a microphone mute button, which makes this whole ring glow red, which is great. Perfect signal from across the room that the thing is not listening. Not listening to you. That is good and correct. And then there is a button on top that lets you activate Alexa. And that was one of the first things they got rid of because they're like, well, nobody does that. Bad idea. I think there is a. There is a, like, immediate understandability to this set of controls that is awesome and useful and has totally gone away because they're just like, well, everybody knows to talk to it. And I think to some extent that's true. But this thing, like, tells you what's going on with it in a way that new devices don't.

1:02:22

Yeah.

1:03:09

That I really dislike.

1:03:09

And I miss the action button because you could press it and not have to say the wake word, which I think.

1:03:11

Well, ironically, it was here because it couldn't hear you and it was playing music. So you had to press the button.

1:03:15

You had to press the button.

1:03:19

It was a solution to a very real problem. But I agree, there was a thing that it was like, sometimes it doesn't work and having. For any variety of reasons it might not work. And having a thing where I can just press a button. Like a truism of the Verge is we are pro buttons. We love buttons. And that was a good button.

1:03:20

It was, it was.

1:03:37

Hayden, anything else you'd add?

1:03:38

Yeah, I think it's that. I mean, like, it's so compelling. It's fun. Like sometimes you just need to keep it simple. Like keep the fun stuff in there. There. Don't, like, don't strip it away like they always do.

1:03:39

Yeah, agreed. All right, we have three more questions. These are the version History hall of fame questions.

1:03:49

Okay.

1:03:53

And to get into the hall of fame, a product has to pass all three of these tests. Spoiler alert. I don't think the echo is going to, but let's find out. I'm willing to be wrong. Did this product do something truly new?

1:03:53

Yes.

1:04:05

Yeah.

1:04:05

What?

1:04:06

Voice control, Siri. Communal in the house.

1:04:07

That's the one.

1:04:10

I'll give you.

1:04:11

It's the far field microphone that I

1:04:11

think is the answer to this. The voice control from anywhere, which, whilst it did have its issues, also did work.

1:04:13

That part of it was new and

1:04:21

I think hands free control of your music, your smart home, your pizza. You used to be able to order Domino's and fart jokes. I mean, they were great.

1:04:23

There were a Lot of fart jokes. But no, I think it's the ability to do that from across the room that is the thing. Because again, Siri had all of this functionality. Siri could do all these things. There were lots of products that could also do these things. Going way back, the idea of, like, pretty good speech recognition and reasonably useful output had actually been around for a while. But what Amazon did was put it A, in your living room.

1:04:37

Yeah.

1:04:58

And B, across your living room.

1:04:59

Yeah.

1:05:01

And I think those two things were new.

1:05:01

Definitely.

1:05:02

That's new to me. Yeah, exactly.

1:05:03

Okay. All right. So it passes that test. Question number two. Was it either remarkably good or remarkably bad? I will confess this is where I think the Echo loses its chance to be in the hall of fame. I don't think it was either one.

1:05:04

Yeah. I think it was kind of in the middle.

1:05:16

It's like, it's.

1:05:18

It was good at some things, it was bad at some things. It was. And with Ciri coming before it. Yeah. I just think it doesn't fall into either category for me.

1:05:19

If, if the question is, did this thing seem super cool? Obviously, yes, smashing success. But it's like going back and reading my review and other reviews, almost everybody had the same reaction to this thing immediately.

1:05:26

But it did. You knew I was gonna.

1:05:39

You love the Echo so much and this makes me so happy.

1:05:42

It did bring hands free control of music into your home and it did

1:05:46

that well for one song and then it could never hear you ever again for a playlist. Okay, sure.

1:05:53

So I feel like. Yeah. Remarkably good or remarkably bad? It definitely wasn't remarkably bad because it did what it said it could do, but it did not do it remarkably well. Yes. I'll give you guys that one.

1:06:00

It.

1:06:12

It wasn't remarkably good and it wasn't remarkably bad. It was, I just want to say, just kind of. Okay.

1:06:12

It was a bit myth.

1:06:16

On behalf of everyone who loves their Alexa. They all appreciate what you're.

1:06:17

What you're trying to do here and people do.

1:06:21

Oh, yeah. Oh, yeah. Alexa hall of Fame question number three. Did it have a lasting impact? Did. Did the Amazon Echo, like capital M matter? I think the answer to this one's yes, I think it did. I think it. It for all the reasons we've been talking about. Like, I think even though this thing didn't directly lead to all of the stuff that was yet to come, I think you see a lot of what people spent a decade doing on Alexa inside of these other AI products now, like the, the personality and the way that we Talk to them and the way that they talk back. All of that feels born of what, what Alexa was doing back in the day.

1:06:24

Yeah, it was a real, like, signpost, like a, a line in the sand. Like it did something like people, even though it, you know, didn't. It was a little bit mid, it didn't stand up to all its promises. It really showed people what could be and it made, I think it brought a lot of inspiration about, like, oh, how can we do this better? Like, let's see, let's look at all the pitfalls of this. How do we push it forward and make it actually what we had dreamed it would be? So, yeah, I think it mattered for sure.

1:07:00

Great. Jen, you good?

1:07:27

Yeah, I think it mattered. I think it still matters because it really, this technology hasn't moved forward in the way that we had hoped. In fact, some there are, you know, there's the new Alexa, which is, as you mentioned earlier, completely rearchitected. So it has none of the original in it. And people, not, not many, according to Amazon, but many. I have heard of a lot of people who keep, who want to go back to their old Alexa. They don't want to use their new one because they've used this device, you know, maybe years, could be a decade, and it did what you wanted it to do. Well, if you've used it for three or four years, you've got your routines, you've got the things that you want it to do. And then when they changed it, which they just did recently, Amazon pushed out the new Alexa to most everyone. You're like, whoa, what happened? This was my device. It's like when people, when they change ChatGPT.

1:07:28

I was gonna say, is this the first sign of what, you know, people getting mad about OpenAI changing its models and retiring them. My routine, my tone, the vibes I have with it, whatever happens to the

1:08:34

Alexa voice assistant, the Echo device and what it brought to hands free computing to the smart home, all of that really built the foundation for where we are today and where we potentially hopefully will be in the future. I would like to see a future where we have a smart speaker in our home that can control our devices locally, but that can also bring a lot of what we're seeing with LLMs and, you know, more advanced capabilities into our home. And I think we will get that one day. And I think we wouldn't have got that if Bezos hadn't pushed his team to create this device in 2014. So, yeah, I think it definitely matters.

1:08:46

I agree. I'M with you. All right, so it narrowly misses the hall of fame. All right, so that's where we land. Jen, before we get out of here, I've been snooping on your computer, as I am want to do you. Made a list of funny things. Funny things that you like to ask Alexa. And. And some of these are very specifically bringing me back to the early days when everybody was just trying to see what this thing could do. Let's see what this thing can do all these years later. Just give us a couple before we

1:09:30

get into a couple of these. All right, Alexa, tell me what you want. What you really, really want. I really, really, really want a zigzag.

1:09:52

Nice.

1:10:01

Pretty good. I like that one. Okay, Alexa, what's the first rule of Fight Club? Don't talk about Fight Club. Alexa, can you pass the Turing test? I don't need to. I'm not pretending to be a human.

1:10:02

Wow.

1:10:17

We love that.

1:10:17

Okay, and then my favorite Alexa ET Phone home. I can't make intergalactic space calls yet. Love that.

1:10:19

Yet.

1:10:27

Pretty good. Give it time.

1:10:28

This was what made it fun in my house. My kids loved to just. Just ask it crazy things.

1:10:30

That's pretty good. All right, Alexa, it's been a pleasure. Goodbye again.

1:10:35

I'm not quite sure how to help you with that.

1:10:39

There it is.

1:10:42

That holds me.

1:10:43

There it is.

1:10:44

That holds me.

1:10:44

Oh, man.

1:10:45

The perfect note to end on.

1:10:46

All right, we should get out of here. Thank you both for doing this. This has been tremendously fun. I genuinely cannot believe that you found an original echo to bring and that it still works.

1:10:48

Still works.

1:10:56

There's something very impressive.

1:10:58

Impressive about that, yes.

1:10:59

Thank you to all of you, as always, for watching and listening. If you want to support everything that we do, read Hayden's work. Read Jen's work. Subscribe to The Verge. The Verge.com subscribe It is the best way to make all of this happen, including you. Get ad free podcasts on this and all of our other shows. Thank you as always. We'll see you next time. Version History is a production of the Verge and the Vox Media Podcast network. It's produced by Victoria Barrios, River Branson, Eric Gomez, Owen Grove, Brandon Keefer, Travis Larchuk, Andrew Marino, and Alex Parkin. Our editorial director is Kevin McShane. Studio support from Matthew Heffrin and Joe Nebras. Our theme music is composed by Brandon McFarland. You can follow the dedicated Version History podcast feed for all of our episodes as soon as they arrive, and you can watch full episodes on our new YouTube channel VersionHistoryPodcast and to support every everything we do and get access to this and all of our other podcasts ad free. Become a paid subscriber to the Verge. Thanks.

1:11:00