You're watching the new stack makers, a podcast for people who develop, deploy, and manage at-scale software. For more information and articles about at-scale technologies, please visit thenewstack.io. Now, enjoy the show.
Dr. Werner Vogels, thank you for being here. I know I don't think they've ever called you Dr. Werner Vogels.
No, it's actually funny. I never really gave that much thought about my PhD, but when marketing and PR found out that there was a doctor in front of mine, it started to appear everywhere.
Well, it's not nothing to be ashamed of.
No, I'm not ashamed of it, but it was just something you did.
It's true. Now, it's funny because we usually do this every year. I was just telling you ahead of time, it feels like the holidays are coming up when we get to talk about your predictions for the next year. But usually we do this at re:Invent. Yeah, AWS's big conference obviously. What made you decide to switch it up this year and do it a little bit earlier?
Yeah, this is sort of the last time. It's time for the voices to come on, you know. And there are so many brilliant engineers with so many experiences that I think I'm not leaving Amazon or anything like that.
I was just going to ask you what's happening here. Not retiring or anything.
I have tons of great ideas that I would like to pursue, but it is time for Amazon customers to hear some younger voices that have tons of experience building AWS.
Are you saying it's your last keynote at AWS?
Who knows? I hear you. Well, I've gone through this phase where I thought, I mean, I know 21 years at Amazon. You would have asked me 20 years ago if I'd stay with this company for 20 years, I would say no. But it's been such a pleasure, and there's so many things at Amazon. You almost have to go back to the beginning of Amazon. In 1994, the word e-commerce doesn't exist. There is no e-commerce technology. Anything within Amazon always needed to be invented. There's nothing commercially to buy. And if we did buy something commercially, it absolutely didn't work at our scale. So now we have a very long history of good engineering and building things that nobody else has built before. And as an engineer, how can you not love that opportunity? Although I almost didn't go because I was still an academic at that moment. And although academia in the US promotes connections with companies, Amazon came and asked me to give a talk about my work. I thought, Amazon, it's a bookstore, it's a web server and a database, how hard can that be? Until I actually did go, and one glimpse into the Amazon technology kitchen blew me away. These guys were operating at a scale that nobody else in the world was. And just helping them become more reliable, have better control over their systems, removing single points of failure, that was fun and still keeps me going today.
For sure. Now, if we're reminiscing, I remember I think it was the first re:Invent, maybe the second, where you gave the keynote interviewing Jeff Bezos.
Second one. 2012. Yes. Andy gave the first keynote, I gave the second one, and then we had a fireside chat with Jeff.
Yes, that's what it was. Compare the re:Invent over the years. What has changed for you? Because you've always given the closing keynote for the last years.
Well, when we started in 2012, I'm just guessing, we maybe had 15 services and maybe eight different types of compute units. Now we have 1,600 services, 250 different services. Every customer has different requirements that you need to meet. And of course, at the same time, technology is changing radically fast. Over time, we've invented quite a few things like Lambda, the way we built Aurora, the way we built DC Core. All of these things nobody had done before. And at the same time, over the years, we've learned a few technologies ourselves that we now try to teach other people. One of them is cell-based architectures. For example, if you want to make sure everything fails all the time, but if something fails, you want to make sure that the minimum number of customers are impacted. There are certain technologies for that that we have developed over time. So we try to teach that to our customers so they can have the same level of reliability that we have.
Yeah, I think that's why I've always enjoyed the final keynote because it's less about the products but those patterns.
I think the last three or four years, I don't think I made any launches anymore. Really focusing on the bigger picture, how should we do this? Two years ago with the frugal architect, that was absolutely in response to things that I saw with my customers.
They were asking, are we really getting bang for our buck? And is this really the way we should be building these things?
So giving your customers guidance about how to best work together with the business, work together with technologists, and all the things we've learned about how to control cost, or how Amazon retail knows how to control cost. Retail margins are wafer thin. So that one was a real big one, and I'm very happy that we've been able to take all the learnings we had as Amazon and put it in one presentation and allow people to walk away. When I see people leaving the room, I want them to be thinking, 'Oh, can I use that? Can we use that? Oh, are we doing that wrong? Or maybe we should go in that direction.' The goal has always been to educate, to make sure that our customers walk away a bit richer than they were before.
And this year will be fun. Do you want to give us a high-level idea of what we'll talk about?
All right. Well, the other thing you always announce is who plays the afterparty, but that's not going to tell you.
But as always, there will be bands and DJs at different ends of the spectrum, but it will be fun.
Awesome. Well, let's actually talk about your predictions though. You know what we already know, it's coming to re:Invent in just a couple of days from now when this goes out. But your predictions for the next year, and they've changed. I think this year is a little bit different from other years because they're very technical. Last year they were a little bit more societal. This is also societal impact, but they were maybe a little bit more high level. This year, you've gone further down the stack.
I still think there are a few softer ones. I also believe that we as technologists have the responsibility to solve some of the hardest problems that there are. And as such, you can't write about hard problems and not write about technology at the same time. The first prediction, and especially put it up front, is the one about loneliness. If you have lots of friends and colleagues, you probably don't even know what loneliness is. But for a very large group of people, whether they're elderly or younger, it's a big problem. It triggers diseases, makes people unhappy, depression, all these things. If we can build technology to help some of these people get out of their depression, that would be great. I remember almost at the beginning of Alexa, I heard a story of a man who was at the beginning of dementia. He told me, I actually spoke to him, that he could see his caregivers getting irritated because it was the same thing over and over again. Then he would go to Alexa, and he put a sticker on it because he didn't want to forget what it was called. He said, 'I can ask her anything I want, and she never gets angry.' That was not necessarily why we built Alexa, but it's a side effect of technology that helps people. Especially where we can help people overcome depression or communicate better with their caregivers. If we can build devices for that, absolutely. It's amazing that a lot of people treat their devices as pets. They give them names. There's a story about a Roomba from iRobot: about 80% of Roombas have a name. One person actually sent the Roomba back because it was broken, and the company said, 'We'll send you a new one.' He said, 'No, no, no, no. Mervall needs to sweep.'
There's work by Kate Darling at MIT about how people treat devices as pets or living devices. She has very interesting research on the attachment of people to these devices. She did a test where she had two groups, both got a small dinosaur, gave them names, and did fun things with them. Then they were given a hammer and told to kill the dinosaur. Nobody did. However, they were more than happy to kill the dinosaur of the neighbors. These attachments between devices and people are important. Kids with autism have a hard time touching other people, but they have no problem touching a device or talking to it. It's often difficult for children to talk to a doctor because a doctor is a magnificent, important person. But talking to a little funky device next to them, no problem.
I would just push back a little bit though, especially when it comes to loneliness and the elderly. It's great to have those robots, but to some degree it's a crutch for real human interaction. The problem goes deeper. It's maybe a part of a solution, but the problem goes much deeper, doesn't it?
A number of years ago I was in Japan and I talked to the Japanese about exactly this problem. In Japan, there was a long history of taking care of your parents or grandparents. They lived with you. But the younger generation of Japanese want to make a career, they don't really want to take care of their parents. It's fun to see how many Japanese startups are trying to build technology to make this possible. Z Works, for example, is one of them. Some of these devices are simple, like a pressure mat in your bed. If you leave in the middle of the night to go to the toilet and you're not back in 10 minutes, alarms go off. Amazon has built one glass, it's quite funny, a very simple device. It will seek you out during the day and ask you whether you've taken your medicine. Small little things. Would I have asked you, 'Did you take your medicine?' Maybe not so much. The little robot can do these things.
