WEBVTT

1
00:00:00.000 --> 00:00:05.000
Today, I want to start with a question. What if the Map itself became the Web? 

1
00:00:05.000 --> 00:00:10.000
This is a question I have been thinking about for more than 30 years. 

1
00:00:10.000 --> 00:00:14.000
Why did I start asking this question? Why have I kept working on it for so long? 

1
00:00:14.000 --> 00:00:20.000
And, perhaps more importantly, why is it time to ask this question again now? 

1
00:00:20.000 --> 00:00:23.000
That is the story I want to tell you today.

1
00:00:24.000 --> 00:00:28.000
 First, let me briefly introduce myself. I'm Satoru Takagi. 
 
1
00:00:28.000 --> 00:00:38.000
I have spent much of my career at KDDI, a network operator, 
 
1
00:00:32.000 --> 00:00:38.000
working on technology research and development, as well as real-world systems. 

1
00:00:38.000 --> 00:00:43.000
In the early 1990s, I started my career working on technologies 

1
00:00:43.000 --> 00:00:49.000
for underwater robots and the construction and maintenance of submarine cables.

1
00:00:49.000 --> 00:00:54.000
Around that time, the World Wide Web was rapidly emerging. 

1
00:00:54.000 --> 00:01:00.000
I started wondering, “Could we use this new technology for these kinds of applications?”

1
00:01:00.000 --> 00:01:04.000
That led me into Web technology, and eventually into Web GIS. 

1
00:01:04.000 --> 00:01:14.000
Since then, I have worked both on building and operating real systems, and on Web standardization through the W3C. 

1
00:01:14.000 --> 00:01:18.000
And since I came up with the idea of Hyper-Layering in 1995,

1
00:01:18.000 --> 00:01:22.000
 I have spent these 30 years exploring the relationship between the Web and maps,

1
00:01:22.000 --> 00:01:25.000
both through implementation and through standardization.

1
00:01:25.000 --> 00:01:32.000
 It was 1995. The Great Hanshin-Awaji Earthquake had just happened. 

1
00:01:32.000 --> 00:01:37.000
At the time, I was deeply drawn to a new technology called the World Wide Web.

1
00:01:37.000 --> 00:01:49.000
And this disaster sparked a discussion, in a study group I belonged to, about how to connect information from different organizations during a crisis. 

1
00:01:49.000 --> 00:01:57.000
That is when I started thinking that the future Web would be one where many different organizations publish their own information, 

1
00:01:57.000 --> 00:02:00.000
and hyperlinks connect it all together. 

1
00:02:00.000 --> 00:02:08.000
And I thought: maybe a map could work the same way. Not a map owned and managed by a single organization, 

1
00:02:08.000 --> 00:02:17.000
but one where different organizations each provide their own information, and the user freely layers them together. 

1
00:02:17.000 --> 00:02:21.000
That was the idea I called Hyper Layering. 

1
00:02:21.000 --> 00:02:29.000
To me, the Web was never about collecting everything into one place. It was about connecting independent pieces of information.

1
00:02:30.000 --> 00:02:40.000
But the Web took another path. In the 2000s, Web 2.0, mashups, and eventually cloud computing became dominant. 

1
00:02:40.000 --> 00:02:49.000
It was a very powerful model. But something changed along the way. Integration moved from the edge to the server. 

1
00:02:49.000 --> 00:02:58.000
The server collected the data. The server processed it. The server integrated it. And then it delivered the result to the user. 

1
00:02:58.000 --> 00:03:05.000
This was, of course, extremely useful. But it also created new problems. Gatekeepers. Vendor lock-in. 

1
00:03:05.000 --> 00:03:11.000
And increasing concentration of the Web around a small number of major platforms. 

1
00:03:11.000 --> 00:03:18.000
These are no longer just technical concerns. They have become structural challenges of the digital age.

1
00:03:18.000 --> 00:03:27.000
And then something else happened. OpenStreetMap. OSM did more than create a wonderful open dataset. 

1
00:03:27.000 --> 00:03:34.000
It made the map itself open. Instead of one organization deciding, "This is the map of the world," 

1
00:03:34.000 --> 00:03:41.000
people around the world could create, share, and use the map themselves. This was a huge change. 

1
00:03:41.000 --> 00:03:44.000
The map itself became something everyone could build together. 

1
00:03:44.000 --> 00:03:50.000
And that has a deep connection with what I was thinking about the Web back in 1995. 

1
00:03:50.000 --> 00:03:57.000
But it also leaves us with another question. If the map is open, what happens next?

1
00:03:57.000 --> 00:04:07.000
So, once we have an open map like OpenStreetMap, and we start putting all kinds of information on top of it — what happens? 

1
00:04:07.000 --> 00:04:15.000
Weather. Disaster information. Shelters. Traffic. Community information. There are countless kinds of information. 

1
00:04:15.000 --> 00:04:25.000
And very quickly, we think: "Let's bring all of this together." Collect the data. Integrate it. And deliver it as one service — often as a portal. 

1
00:04:25.000 --> 00:04:33.000
This is exactly the familiar approach that Web 2.0 taught us. But here another question comes up. 

1
00:04:33.000 --> 00:04:39.000
Where should that integration actually happen? Does it really have to happen in the center?

1
00:04:39.000 --> 00:04:44.000
I started questioning the idea that we should put the whole world into a database. 

1
00:04:44.000 --> 00:04:50.000
The real world consists of countless actors, countless kinds of information, and countless services. 

1
00:04:50.000 --> 00:04:56.000
And they are constantly changing. After all, the real world is not a database. 

1
00:04:56.000 --> 00:05:03.000
So perhaps we don't need to force everything into one schema. Each piece of information can stay independent. 

1
00:05:03.000 --> 00:05:10.000
And we compose them only when we need to. I came to believe that this is actually the more natural way.

1
00:05:10.000 --> 00:05:18.000
 And this is where the map becomes interesting. The data can be different. The sources can be different. 

1
00:05:18.000 --> 00:05:27.000
The protocols can be different. But they can still share the same space. Different data. Different sources. One spatial context. 

1
00:05:27.000 --> 00:05:30.000
The map doesn't have to own all the information. 

1
00:05:30.000 --> 00:05:36.000
It can simply provide a common context in which different information can be understood and composed. 

1
00:05:36.000 --> 00:05:42.000
This is the basic idea behind what I call Hyper Layering Architecture, or HLA.

1
00:05:42.000 --> 00:05:47.000
Then came 2011. The Great East Japan Earthquake. 

1
00:05:47.000 --> 00:05:55.000
At that point, everything I had been thinking about maps and the Web became a real world problem. 

1
00:05:55.000 --> 00:06:01.000
By then, Web 2.0 style portals were already becoming common. 

1
00:06:01.000 --> 00:06:11.000
But in disaster response, the information you need is extremely diverse, and the combination you need changes depending on the situation.

1
00:06:11.000 --> 00:06:20.000
Weather. Damage information. Roads. Shelters. Aerial imagery. And information from people in the field. 

1
00:06:20.000 --> 00:06:26.000
It is simply not practical to integrate all of this into a single portal in advance. 

1
00:06:26.000 --> 00:06:32.000
So in reality, the information remained published separately, by different organizations. 

1
00:06:32.000 --> 00:06:37.000
The information was available. But it could not be combined, when and how it was needed. 

1
00:06:37.000 --> 00:06:48.000
That was the problem I encountered again in 2011. So I began rebuilding Hyper Layering Architecture, or HLA, into a practical, working system.

1
00:06:48.000 --> 00:06:56.000
Here I realized there was a large gap between the world I had imagined in 1995 and the reality of the Web. 

1
00:06:56.000 --> 00:07:03.000
You could collect links to different map sites. But clicking a link just took you to a different site. 

1
00:07:03.000 --> 00:07:09.000
Layering itself simply didn't work. Why? The reason turned out to be very simple. 

1
00:07:09.000 --> 00:07:15.000
I had originally imagined a world where the Web shared one common map representation. 

1
00:07:15.000 --> 00:07:24.000
If that existed, different organizations' map information could stay independent, and still be layered together right in the browser. 

1
00:07:24.000 --> 00:07:30.000
I had hoped SVG could become that common representation. But that is not what happened. 

1
00:07:30.000 --> 00:07:41.000
The data behind maps came in GeoJSON, in proprietary CSV formats, in PNG tiles — each with its own format, and its own mechanism. 

1
00:07:41.000 --> 00:07:48.000
In other words: the Web had links. But it had no common layer to connect maps together. 

1
00:07:48.000 --> 00:07:58.000
That was the real gap between the world I imagined in 1995 and the reality of Web maps. So I had to rethink my approach.

1
00:07:58.000 --> 00:08:04.000
And I never stopped building. I used it in real systems. When something broke, I fixed it. 

1
00:08:04.000 --> 00:08:10.000
When a new kind of information appeared, I made it work. I did this for about 15 years. 

1
00:08:10.000 --> 00:08:18.000
Over that time, HLA became more than an idea. It became a practical way to combine very different systems. 

1
00:08:18.000 --> 00:08:21.000
And then came another important step.

1
00:08:21.000 --> 00:08:28.000
To close that gap, I took the idea of a layer one step further. What if a layer was not just data? 

1
00:08:28.000 --> 00:08:33.000
What if a layer were a Web App? A weather layer could be a weather Web App. 

1
00:08:33.000 --> 00:08:40.000
A disaster layer could be a disaster Web App. A satellite layer could be a satellite Web App. 

1
00:08:40.000 --> 00:08:44.000
Then it wouldn't matter that the data formats were different. 

1
00:08:44.000 --> 00:08:52.000
Each piece of information could have its own logic and its own behavior, because the Web App itself absorbs those differences. 

1
00:08:52.000 --> 00:09:00.000
A layer can carry not only data, but also logic, UI, and interaction with the user. 

1
00:09:00.000 --> 00:09:09.000
A layer itself becomes a Web App. And each Web App can handle its own information in its own way, while still coexisting on the same map. 

1
00:09:09.000 --> 00:09:13.000
This was the next step I arrived at.

1
00:09:13.000 --> 00:09:19.000
And this brings me back to the idea from 1995. Web pages are connected by hyperlinks. 

1
00:09:19.000 --> 00:09:27.000
So what if map layers were links, too? The user chooses a layer. That triggers the Web App behind the link. 

1
00:09:27.000 --> 00:09:32.000
The Web App runs, right there, and builds the layer on the map.

1
00:09:32.000 --> 00:09:39.000
The user chooses another layer. Another Web App is invoked, and that layer joins the same map. 

1
00:09:39.000 --> 00:09:48.000
In other words: the Web's hyperlink extends all the way to map layers. Just as the Web has hypertext, the map can have hyper layers. 

1
00:09:48.000 --> 00:09:55.000
No central server needs to decide in advance "this piece of information goes with that one." 

1
00:09:55.000 --> 00:10:04.000
Whatever the user chooses gets composed, on the spot. The integration is not decided by a central server. The user decides.

1
00:10:04.000 --> 00:10:13.000
And this changes where integration happens. Not in the server. Not in a database. In the user's browser. 

1
00:10:13.000 --> 00:10:19.000
The browser loads independent Web Apps and composes them within the same spatial context. 

1
00:10:19.000 --> 00:10:25.000
So, the user's browser itself becomes the integration layer.

1
00:10:25.000 --> 00:10:33.000
Now let's actually run HLA 2.0, the idea I've just described. Weather information. 

1
00:10:33.000 --> 00:10:40.000
Disaster information. Traffic information. OpenStreetMap. Satellite data. Community information. 

1
00:10:40.000 --> 00:10:46.000
Each of these is prepared as an independent Web App. The data formats can differ. 

1
00:10:46.000 --> 00:10:54.000
Each can have its own logic. And when the user picks what they need, each Web App starts running, right there on the map. 

1
00:10:54.000 --> 00:11:02.000
Different Web Apps, composed together on one map. Many Web Apps, One Map. Let me show you what this looks like.

1
00:11:02.000 --> 00:11:06.000
Change to DEMO Video

1
00:11:06.000 --> 00:11:13.000
What you just saw is more than a demo. HLA's basic properties: independent, composable, practical. 

1
00:11:13.000 --> 00:11:19.000
And these were refined over about 15 years of real-world development and operation. 

1
00:11:19.000 --> 00:11:28.000
No data warehouse. No schema unification. No central integration server. And yet, very different systems work together. 

1
00:11:28.000 --> 00:11:40.000
I also extracted thematic data from OpenStreetMap into separate layers, using techniques like QuadTree Composite Tiling for fast, large-scale mapping. 

1
00:11:40.000 --> 00:11:44.000
This isn't just an architecture that looks good on a diagram. 

1
00:11:44.000 --> 00:11:52.000
It works. And try doing this at this scale with a centralized server architecture — it gets difficult, fast.

1
00:11:52.000 --> 00:11:57.000
And then, a new technology entered the picture. AI. 

1
00:11:57.000 --> 00:12:07.000
Traditional GIS assumed that for a machine to understand information, the data has to be structured — modeled, and mapped to a schema. 

1
00:12:07.000 --> 00:12:14.000
That's still important today. But AI opens up a slightly different possibility. 

1
00:12:14.000 --> 00:12:25.000
A human can look at a map and understand: this area is flooded, this road is impassable, this shelter is too close to a hazard. 

1
00:12:25.000 --> 00:12:33.000
Now AI is starting to read that same spatial context directly from a map's visual. So a new question comes up: 

1
00:12:33.000 --> 00:12:39.000
does "machine readable" have to mean "readable from a database"? Structured data isn't going away. 

1
00:12:39.000 --> 00:12:46.000
But with AI, the map itself becomes a meaningful source of information for machines, too.

1
00:12:46.000 --> 00:12:52.000
So let's revisit what a map actually gives us. Geospatial data tells you where things are. 

1
00:12:52.000 --> 00:13:00.000
A map tells you more — how things relate to what's around them, what matters, what's happening right now. 

1
00:13:00.000 --> 00:13:05.000
The map itself becomes context, not just data. And that reframes HLA a little. 

1
00:13:05.000 --> 00:13:12.000
HLA has always said: you don't need to unify the contents, you just need to be able to compose them as a map. 

1
00:13:12.000 --> 00:13:16.000
Until now, I mostly saw that as a Web interoperability problem. 

1
00:13:16.000 --> 00:13:25.000
In the age of AI, that same composition — different data, sharing one map context — becomes meaningful to AI as well. 

1
00:13:25.000 --> 00:13:32.000
Data. Map. AI. I believe that's one of the biggest new possibilities ahead for the map.

1
00:13:32.000 --> 00:13:40.000
That covers HLA 2.0.  Since HLA 1.0, I had been trying to turn this into a Web standard — 

1
00:13:40.000 --> 00:13:51.000
working through the W3C to make SVG the common map representation, and even standardizing it in Japan as JIS X 7197. 

1
00:13:51.000 --> 00:13:55.000
But SVG never became a global standard for maps on the Web. 

1
00:13:55.000 --> 00:14:04.000
So with HLA 2.0, I stopped waiting and built it myself: a framework called svgmapjs, and LaWA — Layers as Web Apps — on top of it. 

1
00:14:04.000 --> 00:14:15.000
I put a wide variety of WebGIS systems together, and made them work. HLA 2.0 proved, through implementation, that this was possible. 

1
00:14:15.000 --> 00:14:24.000
But that success raised a new question of its own. Why should this only work inside a special-purpose framework? 

1
00:14:24.000 --> 00:14:27.000
Why can't the Web itself do this?

1
00:14:27.000 --> 00:14:36.000
So it's time to take on Web standardization again. With HLA 1.0, I tried to get there through standardization. 

1
00:14:36.000 --> 00:14:41.000
HLA 2.0 proved it by building the framework and the applications myself. 

1
00:14:41.000 --> 00:14:48.000
The next step is giving this capability to the Web itself. Web pages compose through hyperlinks. 

1
00:14:48.000 --> 00:14:57.000
Maps should work the same way — not through a special-purpose framework, but through the browser itself, treating maps as hyper-layers.

1
00:14:57.000 --> 00:15:03.000
The Browser Should Be Hyper-Layered. This is the direction I am pursuing with HLA 3.0.  

1
00:15:03.000 --> 00:15:08.000
Not another special-purpose framework — a capability of the Web itself. 

1
00:15:08.000 --> 00:15:16.000
Thirty years after I first took on this challenge, I want to try, once again, to get there for real.

1
00:15:16.000 --> 00:15:23.000
Web And now, let's come back to OpenStreetMap once more. OpenStreetMap made the map of the world open. 

1
00:15:23.000 --> 00:15:31.000
That is a truly extraordinary achievement. So what comes next? I believe it is building an open Web on top of that open map. 

1
00:15:31.000 --> 00:15:39.000
Different organizations. Different data. Different services. They don't all have to be collected into one center. 

1
00:15:39.000 --> 00:15:45.000
The user can choose them, link them, and compose them. OpenStreetMap made an open map. 

1
00:15:45.000 --> 00:15:54.000
Now, on top of that achievement, we can build the next Web. An Open Map Web. From an open map, to an open Map Web.

1
00:15:54.000 --> 00:16:00.000
I have long wanted to bring to maps a Web where information from different people and organizations can remain independent, 

1
00:16:00.000 --> 00:16:03.000
while users can connect and combine it themselves. 

1
00:16:03.000 --> 00:16:08.000
Meanwhile, OpenStreetMap independently made the map itself open. 

1
00:16:08.000 --> 00:16:18.000
I have used that open map as a foundation for HLA, while exploring how different information and Web Apps can be combined on the same map. 

1
00:16:18.000 --> 00:16:26.000
And recently, HLA has also begun to move beyond being the work of just myself and KDDI. 

1
00:16:26.000 --> 00:16:39.000
With support from WebDINO Japan, a community has started to grow, with contributions including more LaWAs and the internationalization of our documentation. 

1
00:16:39.000 --> 00:16:43.000
HLA is gradually becoming a community effort. 

1
00:16:43.000 --> 00:16:49.000
And now, I think we are at a point where these two independent paths can come together through the Web. 

1
00:16:49.000 --> 00:16:55.000
So I come back to the question. What if the Map itself became the Web? 

1
00:16:55.000 --> 00:17:05.000
I think it is time to find out. If you want to build with HLA, create a LaWA and share it at hyperlayering.org. 

1
00:17:05.000 --> 00:17:13.000
And if you are interested in making this part of the Web itself, join the W3C Maps for HTML Community Group. 

1
00:17:13.000 --> 00:17:21.000
Let's build the next Web on the open map that OSM has created.