 Awesome, yeah. Cura, yeah. Adam Moriarty, head of collection information access at Auckland's War Memorial Museum, and today I'm going to talk about voice search. That was a pretty boring introduction to myself and my talk book. I think we've got all the key points in there, right? Who I am, where I work, and what I'm going to talk about. And if you were counting, which I know a few of you would have been, that was 30 words. Now, why on earth would you care how long my introduction is? Well, 30 words is the average response time of a system such as Siri, or the Google Assistant, or Amazon's Alexa. And I heard that fact a couple of months ago. That's the average response of these devices, and it kind of got the gears turning. So I thought about how much time we have spent as a sector working on rich visual content as a way of storytelling and sharing our information. Whether that's our topic pages, or blogs, or research articles online, or digital interactives in galleries, or exhibitions as a whole. We spend a lot of our time making things that are much longer than 30 words. One of the jobs of my team is to help manage inquiries at the museum. I kind of played with the idea of what would it look like if I challenged our curatorial team to only respond to every inquiry in 30 words, and they're not allowed to use any imagery. It sort of starts changing how we think about the content that we have. And so, yeah, I started looking into voice search, and I'm not going to lie, I'm not someone who really had used it before. It's kind of, I've gone iPhone, got the Mac, Siri's turned off. Kind of got some privacy concerns there. And I think it's quickly, right at the start, I want to say, acknowledge there are some privacy concerns when talking about voice assistance. I'm not going to be talking about that today. I really just want to talk about voice as a search. So, acknowledging, parking, moving swiftly on. But we will have a lecture in the background. So, yeah, I kind of didn't really believe that voice was going to be this thing. It feels like it's something from the future, right? We were promised flying cars, and we were going to talk to robots. And voice is probably the one thing that maybe it's actually there. Maybe this is the future that we were promised has come, has arrived. And there's a couple of facts there just to help support that. There are now 500 million Siri-enabled devices on the planet, 400 million Google Assistant. There are 20% of American homes now have an Alexa. So, this isn't some future trend. This isn't something that is coming, the next big thing. I think we have to start acknowledging that this is something here right now, something that we need to start thinking about. It's how people are getting answers to their questions. It's how people are using the Internet nowadays. Another stat that I quite like is I think Comcast reckons that by the end of next year, so by the end of 2020, 50% of all searches on Google will be conducted by voice. So, this is how people are using the Internet nowadays. And so, I kind of wanted to delve into those figures. I wanted to think about how we could start using it as a sector. And so, I bit the bullet for all of us, and I went and bought one of each of these devices, put them in the office and at home and sort of ignored the privacy. I just put that to the back of my head and just started interacting with them to see how they would go. And I'm not going to lie to you guys, I love them. I didn't want to. I thought I bought them into the home and I thought, oh, creepy little Amazon sitting there listening. But actually, it was really great. I mean, because they're really convenient. They're fast, right? So, on average, I think the average typing, so it's 40 words per minute, but you can do 150 when you speak. It's a much more human interaction. I mean, talking is so natural to just speak and get an answer. If you think, again, it's like, oh, hands-free searching. When am I ever going to need that in the kitchen? I need it all the time. And just being able to get that convenience in the office, being able to answer an argument incredibly quickly by just shouting and getting that answer. And so, you're really, really enjoying that and thinking they were cheap. I mean, almost all new phones come out with them pre-installed. And if you want to put one in your home, it's going to cost you under $100 to do it. They're small as well. Small footprints, so you can hide them wherever you want. And it's not going to be taking up in ugly, I guess, where I'm going with that. And they deliver quite personalized results because they know a lot about you. They know where you are. They know the types of queries you've done before. And if you've got the Amazon Alexa, it probably knows everything you've ever purchased. So, they can deliver quite personalized results. So, yeah, I started getting on the bandwagon. This is awesome. We should get more of them. And I started thinking about that. The interactions I was having. And realized that I wasn't dealing with a search engine anymore. And we've got really used to working with search engines. It's what we spent a lot of our time getting our content ready for. Keyword searching. But actually, now we're dealing with answer engines. Because these systems want to deliver you a packaged answer. Because it's convenient. It's easy for you. They don't want you to go searching. I mean, back in the day, when you jumped onto Google, you searched. You got 10 results. 10 blue links. Now, it was up to you to decide whether you chose that promoted link at the top. Or the one the SEO marketer had spent the most time trying to get into the number two place. Or whether, God forbid, you went to page two looking for an answer. But it was up to you. You got to decide what you chose. Now we've got a system. We've got the wealth of the world's information out on the internet. But now we're relying on a couple of devices, a couple of corporations to decide what's relevant. And to package up an answer and deliver it to you. And again, that kind of got me thinking around, well, is this actually something museums, the sector wants to play around with? Is that something we want to get involved? Or is it something one of those fads we can just let slip by? Another shiny new tech that we don't need to get involved in. And so what I want to do today is just go through the three questions that I went through, essentially. So are we going to do this? Are we going to try and work with these systems? How do we do it? So if we decide yes, what are the technical things that we need to do? And should we? Kind of ethical question there. And so, yeah, are we going to do it? Unfortunately, I think if you're online, if you put publishing content on the web, you're already part of it, right? We're already putting our stuff out into the world. It has already been sucked up into the Google Knowledge Graph or into whatever these systems are. And we're putting it out there and it's getting used. And we don't have a choice. That horse has left the station. I don't know. I've mixed metaphors there. But something has happened and it's already gone. And so we kind of have to. Because someone else will be doing it or be putting content out there if we don't. Now, good news, a lot of our content is already out there. Because we spend a lot of time getting things like our opening hours and our prices and our parking directions and our what's on. We spend a lot of time getting that into a format so that it appears highly anyway. And so that information has just kind of been sucked up and it's there and it's ready to use. It's also, if I admit the kind of stuff I don't really care about, because I'm focusing on that collection content. So what do we have to do to get our collections ready for this? And are we wanting to put our collections into these systems? I guess just a quick sidestep. I got the assistance. I brought them into the house. And it took all of three minutes, I think, for my son, who is four, to work out how to use the Google Home. And four minutes and he was just chatting away. It took him probably at the 10 minute mark. He'd worked out how to control all the speakers in the house and get the poor patrol soundtrack blasting at 100% in every room. It was just pumping. He loved it. And there was this amazing sense of power that all of a sudden he had where he was just like, oh, this is awesome. I don't have to use a keyboard. I don't know what a keyboard is. I don't know how to use an ISO. I haven't had to learn anything. I've just shouted at this magic box and poor patrol is playing in the kitchen. And yeah, you could see him kind of like really get excited. And I don't know if anyone else has got young children, four-year-olds in our house right now, but the Coast Guard is king. Like anything, any information about the Coast Guard is just if you've got stuff, share it. Because that's all he wants to know about. And all of a sudden he can start asking these questions and we just would hear him up in the kitchen kind of just asking, well, do the Coast Guard take an oath? What's the biggest Coast Guard boat? How fast can that boat go? And we realised he was just there kind of just asking all these questions. And of course, if he hadn't got that, he'd have to have asked his naively challenged father. I mean, I grew up in Birmingham. We have canals and shopping carts. We don't have a Coast Guard. I know nothing. And so he's able to kind of learn something. And then he was telling me he was giving that information back to me. I haven't retained any of it, but it was an interesting dynamic. And at the other end of the spectrum is my mother-in-law, who's 73, and is the only person I have ever seen use Siri in public. And she just shuts at her phone aggressively. And it's not because I think she wants to, but it's because she can't use the keyboard. Or she's trying to think in case she ever watches the video. She can use the keyboard. She just struggles if she can't find her glasses. And she doesn't know where her glasses are. Glasses in her bag. No one knows where the bag is. And then she has to use the stylus. Now I don't know if you have to find the stylus in a 73-year-old's bag. It's impossible. So it's just a lot easier for her just to shout at the phone. And again, so there's two user groups there using this and using it for reasons that I'd never thought of. And I kind of realised that maybe in the middle, it was just a novelty. I had no need. I much prefer using a keyboard than I thought I did. And maybe actually these two groups, we were in the middle, we were thinking about how we get our content on and how we online and how we use search engines. We're actually these two different audiences who we want to engage. We're using and having this big uptake and kind of, yeah, maybe I had my my blinkers on, hadn't seen it happening. And so, yeah, it's with that kind of make me again think, so are we wanting to do it? Well, yeah, because these users out there, that's how they're getting their information. That's how they're answering some questions. So, yeah, we kind of have to think about how we want to engage with them. And so how are we doing it? So then we spend some time thinking about, well, what's the, how do we start playing with these systems? How do we start seeing how we can try and get our content in there? And luckily one that what we've been doing so far, what we've been working, what we've been talking about here for like the last five years, a lot of us are kind of already on that journey. It's about looking at things like structured data. How do you tag up your content in a way so that both machines and people understand it? We've been talking about this for a long time. If you don't know what the schema.org is, that's your homework. Go and ask Siri. She will tell you. It's the way of tagging up your content on your own website so that machines can understand what you're talking about. They don't have to use keywords. That's something we've been, as a sector, as libraries in particular, have been working on to make sure your content is understandable and findable. Now, you could, if you want, with Alexa, there's a skills store you can go and upload some content to. And it's really easy. It's like a drag-and- drop interface to create these kind of crazy little apps. And you can run them locally, which is kind of cool. So we created a couple in the office that do things that are no use to anyone. But we can run them in the office and we could test and see how they worked. Literally uploading like an FAQ kind of questions, answers and keywords. And we've seen a couple of museums out there using that. In the U.S. they're looking at voice assistants in galleries is where you can ask a question where in a gallery it gives you a response. Could be cool. MoMet and Met have created little apps that are on there, the public, so anyone can download them, which give you a what's on. So in the morning you can ask what's on at the Met and it will tell you, it will give you a response. And unlike the sort of tagged data of using Google, they've able to create a bit of a narrative around that. I mean, you're still creating an app and you're reliant on people downloading it, but it could work. There's 100,000 skills on the Alexa store, most only have one review, so maybe it hasn't quite got there yet. And I have to admit, I don't want to be standing here telling everyone to make the next version of an app. So I'll stop there. When we when we got the devices, we started asking them questions. We asked each one of them 100 questions. And we quickly discovered that if you had an FAQ page, that stuff would come out really come out at the top. Makes sense, right? You've already written it from questions to answers. You've already done all the hard work. But again, a lot of the time we focused on the physical stuff, like how to get here. Where do I park? Can I get gluten-free sandwiches? You know, it's kind of the very much focused on the physical visit and very sort of, I guess, low-hanging questions. It was very hard to find examples where broader collection questions were getting answered in an FAQ format, or maybe it's just one or two, not sort of rich content that could be delivered through these devices. So again, definitely some work that I think we need to do both in our museum, but also as a sector about thinking of the questions that people are going to be asking and providing the answer. So it's easy for people to get it. And I think there's also, sorry, just a side step, Alexa also has, I think it's AlexaAnswers.Amazon. You can go and see all the questions that people are asking. And they're trying to crowdsource the answers. And so it's kind of crazy just to see the types of things people are asking and the types of crazy answers that are being given. The question I searched this morning, the question is Wellington, the world's windier city is still unanswered. So there's an opportunity there if you want to jump in. But it's kind of did the types of questions. The one that we came up with is, can I keep squirrels as a pet? It's like one of the top questions from last week. Yeah, you can't, it's a leave I found out. But it's, again, interesting to see that there are enough people asking that incredibly weird question. And so yeah, that's another way for us to kind of start looking at the types of things people are asking and helping to start building the answer, you know, thinking of the answers for them. And then the other thing, asking 100 questions of these devices, sort of interviewing them. The thing that came out the, you know, at the top, the thing that was the most common thing that we found is 90% of the time it started with, according to Wikipedia, blah, blah, blah. So once again, I'm standing at NDF to tell you about the power and importance of getting your content into Wikipedia, because that's where our people are going to look for information. And it's, it's a powerful way of sharing our collection, sharing our stories and putting our content where people are already looking. So if you're not working with Wikipedia, again, the barrier now is, is quite low for us to get in and create those pages or enrich those pages, add some, add some new sources, references, images. And it's a great thing because when you put the information in there, you're concretely changing what the world knows about our subject of expertise and also increasing how people can find out about our collections. I mean, it's a stat from last year, but we have one collection item in Wikipedia that sees 250,000 page views every month. And it's never been viewed. It's been viewed like 12 times on our own website. This is a great way to amplify your collections and share them. That's it for my Wikipedia kind of sales picture. You should be doing it. And it's also it's an organization whose mission and ideology I think we can get behind. It's not, it's not, we're trying to support Amazon or Google. And so that, that last question around, should we, should we be working on 30 word answers for, for concepts and collections that are far too complex for 30 words? I mean, are we doing it? Well, I think we have to it's, it's already there. The technology has started. It's already been sucked up into the, the knowledge graph. How do we do it? Structured data, things we're already doing. We work with aggregators and third parties to put our collections where people already are. Should we? It's a, it's a tough one, eh? 30 words, it's, it's not a lot. And we spent so much time on these rich deep content. You know, do we, is 30 words really worth it? It's, is it really something we want to be investing our timing? And I, I don't know, I think about that. And so I ask you, I guess, is it acceptable if someone who is searching for Mardi Culture on an Alexa gets hit with a blank end, you know, they don't get anything? Is it acceptable if a school child who is searching for Auckland history on a Google Assistant gets a response from a corporate tour operator who's trying to sell a ticket around historic Auckland? And is it acceptable if someone who has a spark of interest in content or collections that we are the custodians of reaches a dead end? Is it not our role as a sector to ignite that spark? No matter where they're looking, is it not our role, our duty to put our information out in whatever content or context people are looking so they can continue their journey, their journey of discovery, their journey to answer their question? And of course, 30 words, it's not much. It's not a lot of room to play with, but those 30 words may lead to another 30 and another 30. But they definitely shouldn't just stop. If someone has a question about collections that we look after, about information and knowledge that we have, we should be sharing it. Isn't that what we've spent the last 163 or however old your organisation is? Isn't that what we have spent the last 160 years working to create the tools, the expertise, the networks and the trust to put our information out into the world? And isn't it now more important than ever to make sure that someone is putting out trustworthy and reliable information, even if it's only 30 words? I really believe this is something we should be doing. And so to that answer, should we be working with these systems? Should we be putting our content out into the world? Should we be working with aggregators and Wikipedia? Should we be marking up our content in a way that is usable for those people and machines? Should we do it? Yes. Sometimes you don't need 30 words to get a good answer, right? Thank you. Thank you, Adam. That was brilliant. We've got some time for questions. Does anyone have a question for Adam? Adam, thanks for that. There's a substantial difference between voice commands and voice searching. Yeah. So I think the first half of your paper about, you know, wanting to find out more about boats was kind of fun, but not really relevant to what we're looking at in the glam sector, which is actually a philosophically more profound problem. So you have really two approaches to organising information and retrieving it. And one is more structured data and one is better search. And somewhere in between that you've got these new search capabilities like voice search, which have severe limitations at the same time that they open up new ways of thinking about what it is that humans search for and how they do that. So I think I wanted something in there in that kind of middle space. Yeah, that's one of those things where this is quite a broad from both voice interaction and voice as an interface into collection. So thinking of it as a, yeah, how are people going to be asking questions of our collections and the types of questions they're going to be answering, asking and how do we answer them? And yes, that kind of the search, yeah, acknowledging they are two different things, I tried to try to merge them into one kind of into one narrative. Yeah. Any other questions? No. OK, can everyone please join me?