“If Dr. Roger F. Tomlinson is the father of GIS, who is the mother of GIS?” Prof. Michael F. Goodchild quizzes. “Logically, it should be the earth!” Speaking at Esri India User Conference 2017 in New Delhi, Goodchild is building on the legacy of the father of GIS at the Dr. Roger Tomlinson Memorial Lecture.
Tomlinson initiated, planned and directed the development of the Canada Geographic Information System — the first computerized GIS in the world — in 1965 to produce accurate estimates of quantities of Canadian land. At the time, there were two available methods to achieve this — overlay sheets of dots and count them by the hand, or use a planimeter. “Both the methods were notoriously labor-intensive, slow, inaccurate and even demeaning,” points out Goodchild.
Enter, Tomlinson’s insight: Dedicate a $10 million computer, even though none of the necessary software and peripherals existed in 1965. “Another very, very visionary move was to hire IBM to provide technical support,” tells Goodchild.
The techie diversion was needed at various levels. It was necessary to scribe each map to create a clean, scannable product; build a drum scanner from scratch; write algorithms to vectorize the scanned maps; add attributes to each polygon; write algorithms for polygon overlay and area estimation; and invent the Morton Order (quadtree) to optimize the sequence of maps on tape. “All these things required massive effort,” Goodchild points out.
By 1972, CGIS was in big trouble, but Tomlinson’s vision of a multipurpose technology of GIS had a profound and lasting impact, enabling transformation. “That legacy is still there, but GIS can do so much more — things that Roger could not have thought of at that time,” Goodchild says.
Space vs Place
One major problem with GIS, Goodchild indicates, is that it talks in terms of latitude, longitude, polygons, precise distances and directions. People, however, think in terms of a place, not space. “How many people here can identify the lat-long of where we are using just their general knowledge? Almost no one. And yet, we all know we are in Delhi.”
Places are often vaguely defined, without precise boundaries. They are often context-dependent, with imprecise distances and directions. “In our heads, we all have information about at least 10,000 places on earth. However, the ‘names’ layer is missing from the seven base layers of the US National Spatial Data Infrastructure,” Goodchild tells.
Another matter requiring our attention today is the issue with Big GeoData. How do you integrate multiple sources? And how do you assure quality? “If I use Google Maps for navigation, it produces a personalized map created only for me. This map uses crowdsourcing data to let me know the ETA. Now, this map is quite reliable, but, there is a very intimate relationship between crowdsourcing in real-time and the result,” says Goodchild. “One of the consequences of Big Data is that we know nothing about the quality of the data.”
Today we have multiple windows, multiple scales and interactive tools at our disposal. “All these are way beyond what Tomlinson could have imagined,” Goodchild stresses. “He possibly could not have thought of the software package called GeoDa.”
He couldn’t have thought of Esri’s remarkable ArcGIS suite also. “It never ceases to amaze me,” says Goodchild. “I always tell my students: Don’t worry about where you will find the tool. If you can imagine it, there would be a tool in ArcGIS waiting for you to make it happen.”
Tomlinson’s system was designed only to produce numbers. It wasn’t designed to tell stories. However, stories are one of the best ways to communicate the results of GIS analysis to users who know nothing about GIS, who have no background in spatial thinking — even though they may themselves be contributing spatial data, courtesy the smartphone.
“Data analysts may not be motivated to tell compelling stories. They prefer structured, unambiguous fields like math, statistics and computer science,” Goodchild asserts. “But, we need to understand that stories go where data and quantitative analysis cannot — to our hearts.”
Education is a field that Goodchild is most passionate about. And he makes a compelling case for the need for GIS education versus GIS training. “Education teaches you the fundamental principles behind GIS — that will always be true,” he maintains. “Training, on the other hand, might become redundant.”
Goodchild recalls how the introduction of Google Earth, in one stroke, made redundant what he had taught his students for an entire year. “A 10-year-old could do in 10 minutes what I had taught them in one year. If the average career of a person is 40 years, and at least 50% of what he learns remains true after 20 years, then only we can call his education effective.”
The Fourth Paradigm
Which brings us to the fourth paradigm of data-driven science that claims: Let the data speak for themselves. Goodchild points that this is a very dangerous idea because geospatial data are never geography. It is impossible to measure location perfectly. All geospatial data are subject to a myriad of uncertainties. They are always generalized, abstracted and synthesized. And the detail is always truncated.
“Given all that, when data speaks for itself, it is not geography that you are hearing. A geographer must not delude himself with maps; he must look out of the window. Because however carefully you may map, you are never mimicking the world. There is always something left that the map is not capturing,” Goodchild sums up.