From DJ to CTO: Esri’s Charles Kennelly on the allure, power and still-emerging potential of GIS.
"We’re still seeing organisations wake up to what’s possible."
“My previous career was a DJ”, says Charles Kennelly, UK CTO of GIS specialist Esri.
“I had a kid on the way and needed some structure of a normal life. So I started working in IT support.
“One day my boss came over and said ‘you like new things, don’t you? There’s this thing called GIS that the Highways Department are doing…’ Two weeks later, I knew what I was doing for the rest of my life.
“I was seduced by it and had a vision of how it could be put to work.”
The enthusiasm of Kennelly for the tools he has spent a career refining, is infectious.
And even twenty-odd years ago, when GIS required the computational firepower of a small nation state to run in any meaningful way, important use cases that hinted at future sophistication were not hard to find.
As Kennelly explains: “Highways were planning a road in North Wales (it was never built).
“They were looking at the cost of building it, which basically meant they had an algorithm for finding nice woodlands and trashing them -- because that (immediate build cost) was the only factor. But I knew from the area that not only were they picking the wrong route for the road, but because I worked for the council, and knew about the geology of the area, I knew that there was a very valuable mineral deposit that they could never extract if we put a road on top of it. So the local economy was going to be damaged in ways they didn’t foresee.
“I had a vision of the person designing the road, being able to click their mouse on the map and get feedback on the impact of those decisions; change the way they do things. That, for me, was the part where I sort of fell in love with GIS basically; I’ve been working on that problem ever since. And now have a system that can do that; that can actually give you contextual information, as you plan something, and change your behaviour.”
Esri UK CTO Charles Kennelly: "I was seduced by it"
Esri is a unique company.
Founded in 1969, privately held, still owned by founder Jack Dangermond and his wife, saddled with zero debt, and having survived and evolved through numerous major shifts in technology, Esri’s GIS software is in use in over 350,000 organisations – governments (defence, planning departments, etc.) and utilities are a sweet spot.
(Dangermond, who set up Esri with $1,100, told Inc.com a few years back: "In our 43 years, we have never missed a quarter. We've never had any layoffs; we've never had any downsizing. It's just been very conservative, systematic growth. One thing that has made us so successful is that we've never taken outside investment. That means we can concentrate on what our customers want -- not what the stockholders or the VCs want. That's a strategic advantage.)
The company has 49 offices worldwide, with each country set up a discrete entity, and invests up to 30% of its revenues annually in R&D – something that has allowed it to keep an edge in the field and evolve its technology.
(GIS used to require serious hardware and deeply expensive software to run. As Kennelly tells The Stack: “When I first started working in geospatial, the computer cost £50,000, the software was another £60,000”. He adds with a grin: “I used it to work out where I wanted to live. I mapped commuting time to work, I mapped house prices, crime rates, rainfall… Those maps didn’t exist; I had to make them. At that time I was probably one of a handful of people in the country who do that type of query. Today, you can you can ask that question for free on the internet; the data exists, the processing power is there to do it, and the systems are in place. So if you know a location, even one dot,can tell you a lot about a place. Essentially, location is the ultimate primary key.”)
Spinning up a demo, Kennelly shows what the software can do -- taking us over a city, down to a building, through it, under it, to see the gas and electricity lines, the transport links; how planners and architects can plug in CAD models to broader landscapes, assess impact in 3D, and more.
The amount of data that can be layered into Esri's tools varies, needless to say, by country: some are more open than others with public datasets: "For example, the Netherlands have got incredible data available to everybody because most of the data that their government captures is open sourced," notes Kennelly.
"Whereas in the UK, the volume of data available from government sources is getting better, but it's not at the level it is in most of Europe. Primarily, the major data sets, administrative boundaries, demographic data, traffic data, height maps, soil maps, they they're pretty consistent around the world though."
He adds: "Most other mapping providers' business model is coloured by people where people are. We've come from a background of environmental research so our data tends to be much more consistent. We are as much concerned about the Scottish Highlands, as we are about the centre of London; our customers work in both."
(A high-profile use of Esri's software recently was for the John Hopkins Covid-19 map, built using Esri’s ArcGIS and at its peak accessed over 8000 times a second. While that might have showcased the company's ability to handle high volumes under a new SaaS model, it didn't capture the compelling Google Earth-on-Steroids feel of the tool's capabilities; it's not hard to see why defence departments are fond of deploying it strategically.)
For those brought up on the easy click-and-view powers of Google Maps or Google Earth, it's hard to understand how a fairly low-profile company that started out requiring some monster Unix workstation to run has a) retained a niche and b) kept up with emerging technology firms, even as many other legacy technology providers folded.
As Kennelly notes: “In a modern context, GIS is not necessarily more complex than handling imagery or AI.
"However, in the early days of computerisation the complexity of the information often made GIS prohibitively expensive. This resulted in its development remaining a relatively niche affair, and consequently the understanding of how to make use of it has remained a specialist domain. As an illustration, it requires 64 bits to record a simple point location to a meter accuracy in the UK, this translates into roughly 256 transistors in a typical processor design. To represent something more complex than a single point - such as a building footprint or admin boundary - requires a list of points that are processed as a single entity.
"The first Intel chip, released in 1971, had about 2,500 transistors, making the processing of GIS data a highly intensive process. Additionally, the maths needed to process the relationships between shapes are way more complex than that needed for typical database or accounting uses. Think of the process of deciding if two complex shapes are exactly equal compared to the process of seeing if one number is bigger than another. The true value of geographic data is when you’re able to easily view intersecting layers of information and draw analysis from it. Nowadays, rather than simply connecting a few dots of information – we’re able to connect thousands of layers of information and easily see how they interact with each other.
He adds: "If you compare it to maths, it’s the difference between going from basic addition to trigonometry. In summary, the huge jumps forward in computing power have massively unlocked the power of geographic information – and we’re still seeing organisations wake up to what’s possible."
One challenge for Esri -- other than tackling perennial cries of "it's too expensive!" -- has been shifting from a long-standing model of providing on-premises software licenses, to more flexible models.
The vast majority of Esri’s solutions are now available online, however, with the company most recently launching its ArcGIS Platform in February – a ‘pay as you go’ SaaS service specifically aimed at developers, that lets them incorporate Esri’s capabilities into their existing stack. (Users can tap the client API to work directly with its location services using open-source mapping libraries like Leaflet, GL JS, OpenLayers, and ArcGIS REST JS. The platform comes with a bundle of APIs, SDKs, and services for creating location-focused apps, including for web development, mobile and desktop platforms, and low-code app builders.)
Esri UK CTO Charles Kennelly notes to The Stack: "One big benefit I’ve seen from the shift to the cloud is that when you’re delivering things online in a web browser, you have to put so much thought into the user interface. We’ve learned so much from that, and that knowledge has filtered through to the rest of our software – meaning our ease of use has gotten much better. Most of Esri’s customers now operate in a hybrid-way, using a mixture of on-prem and cloud solutions. Many companies are at different stages of their digital transformation journeys and some industries are highly regulated, so customers are able to operate in a way that works best for them."
One recent use case demonstrated at the company's federal conference in the US was analysis of fires in California: "They want to understand how many properties are at risk? What action do we need to take if the fire moves into that area? They didn't have the necessary property data to map the building footprints, so we used an artificial intelligence tool that takes the imagery and automatically creates building boundaries for it: it's been trained to recognise where buildings are, so you send a plane across once, get your latest imagery, you run the AI tool against it, and it returns you a set of polygons that represent where people live. You can then take fire information and run analysis against it in a way that used to be prohibitively time expensive. That intersection of the application of AI tools to geospatial tools, that I see that there's, you know, the potential is just unlimited."
