On the Necessity of Diagrams

Stephen Ervin of Harvard University delivers a fascinating presentation about object-oriented diagrams in GeoDesign at the 2010 GeoDesign Summit.

Jan 5th, 2010

Start From:
Player Color:

Right-click on these links to download and save this video.


00:01So, I want to be, not contrarian, about terminology, but offer refinement.

00:08And I want to talk about diagrams because I think they're really, really interesting.

00:15And essential to our enterprise.

00:17And they're like maps, and they're like sketches, both words that we've been tossing around quite a lot lately.

00:24But they're unique.

00:27I think most of us can call that a sketch.

00:30It had characteristics of having been made quickly.

00:32It has characteristics of representing some real-world attributes like texture sometimes, or color, or outline, et cetera.

00:43Many of you will recognize that as a map.

00:46It also has some relationship to some real-world stuff.

00:50Either shape or scale, but not both.

00:55And other kinds of abstractions to be sure, but kind of one-to-one mappings.

01:00That's one reason we call it the map to the world.

01:04And then there are these critters called diagrams, which are graphic things.

01:10And they talk about, or represent, or abstract, real-world and spatial things, but in a different way.

01:19And in critically different ways.

01:23Diagrams graphically represent high levels of abstraction, just as maps do...

01:28...and perhaps in a different way that sketches and photographs do.

01:34They emphasize critically elements and relations, a good basic database concept, such as topology, for example.



01:47Duality, singularity, and other kinds of high-level topological constructions...

01:53...rather than, for example, measurement or detail that you would find often in a map or a sketch.

02:01And they're often used in the process of explanation or argument...

02:05...whether with yourself or your client, your instructor, your students, your colleagues.

02:12There are many kinds of familiar diagrams in different disciplines.

02:17They all have the characteristics I talked about, having elements and relations in them.

02:22They use color.

02:24They often use arrows for various reasons.

02:28They are...even though they may...some of them are more maplike than others.

02:33They're not maps in ways that we can distinguish them, and to be sure, there are...

02:38...this is a continuum, and there are sketchlike maps, and there are maplike diagrams.

02:41And there are things that are odd hybrids that we can't quite figure out until we see how they're being used.

02:46And that's an important part of the thing here.

02:50Here's a famous one. Garden Cities of the Future, Ebenezer Howard.

02:56This is a fragment of a larger diagram, and he says, "Note: This is only a diagram."

03:03We can't make the plan; we can't make the map...

03:06...and I'll use those two terms interchangeably for now...

03:08...until we have a site.

03:10So there's something different here.

03:12This is a set of concepts, ideas, elements, and relations that have yet to land on the ground.

03:20And when they do land on the ground...

03:22...then they will generate maps and plans and work orders and construction equipment and all that kind of stuff.

03:29And here's a canonical, familiar bubble diagram from the architectural and land planning enterprise...

03:36...where once again, we see elements and relations and topological constraints like connectedness and maybe family relationships...

03:46...where color is used to designate there's something the same about these, but we don't know what it is, perhaps, even.

03:52Or maybe we can begin, it's so obvious what it is.

03:54And there's a very simple graphical palette that's constrained by all those semiotic rules...

03:59...that Bertrand and others have helped us understand the uses of...

04:03...that are used to map these concepts, not their terrain-based implementation.

04:11But what's so cool, I think, and interesting, and fabulous, and important, about these...

04:16...is that in fact they have this...some kind of a relationship to maps.

04:21They can be derived from maps.

04:23We can look at a map or a plan and diagram it.

04:26Or, we can take a diagram and instantiate it as a map or a plan.

04:32And so they hold this interesting...

04:36...dare I say, epistemological relationship to what we know and how we know it, and where it lands...

04:42...in the birth of maps and plans, and the design process.

04:48And Carl showed us multiple examples of diagrams in his talk this morning.

04:49So, to reiterate, diagrams embody concepts, or ideas, like inside and outside and edge and large and many.

05:01And if you go back and listen to the video of his tape...

05:04...you'll see he talked about diagrams and he said that designs are different 'cause they have different diagrams.

05:10I collected these diagrams and we combined them.

05:13And I kind of knew what he was going to talk about, but I didn't actually set him up to say all that.

05:19Ideas can be implemented and manifested in diagrams, and diagrams can generate maps.

05:26And you can run this back.

05:28You can say, from this map let's extract the diagram.

05:30Let's see what the ideas embedded therein are.

05:34Now one important role of this is if you want to make a change...

05:39...you can take your map up to the diagram and plat the diagram in a different terrain; a different site.

05:44Maybe you'll have a different map, and a plan, but the same diagram.

05:48And it's a one-to-many relationship.

05:50There can be many ideas embedded in a diagram.

05:52A diagram can generate many maps.

05:55A map can embody many diagrams together, as we've seen.

06:01In object-oriented terminology then, that sort of idea that's frankly borrowed from Lenaeus, and it manifests in modern programming...

06:11...you could say that diagrams are instantiations of ideas, so ideas are classes in a sense...

06:19...and diagrams become instances of them, and maps and plans, in turn, are instances of the diagrams.

06:26And it's a, I believe, lattice rather than a simple hierarchy, so that indeed, a diagram can have multiple ideas.

06:35And a map can have multiple diagrams.

06:38But these are three very different levels of inferentional...inferential argument and discourse.

06:47We talk about the map, so we care whether there's a cliff, and an obstruction, and a south facing.

06:54The idea was very simple. Two nodes and a line between them.

07:02So diagrams support logical analysis and inference as well as spatial.

07:09And that's important, because we talk about these in slightly different terms interchangeably.

07:14And logical analysis is not necessarily visual, whereas spatial analysis of maps often is.

07:21And diagrams sit in between here and support both kinds of inference.

07:26Well, if the cliff is between these things, then there's a problem.

07:30That's a spatial inference.

07:32That they are connected is an ideological concept, or even logical.

07:38I bow to the diagrams made by ModelBuilder...

07:40...but they're a small subset of all the kinds of diagrams that we need to be able to support in this process.

07:46Here's an example from the planning literature.

07:48This city river district diagram describes the recurring relationship.

07:53It further suggests potential linkages.

07:55The same elements that the generic diagram (city center, main street, peripheral 00:08:15

08:03...one role of a diagram is to be able to say, what's similar about this situation and that situation...

08:09...which are very different terrain and topography, but have similar urban functioning or connected kinds of roles.

08:15And you could take this diagram and create a map from it.

08:19Or you could look at a map and derive a diagram from it.

08:22But we don't have very good tools for doing that other than the human eye-brain combination.

08:29Here's a diagram that I caught in the wild this fall at the GSD.

08:35And I didn't have my camera with me, so I had to reconstruct it.

08:39You may not know what it is yet.

08:41What if I tell you that's a waterway?

08:46That's a highway.

08:48That's some green stuff.

08:51And that's some built stuff.

08:54Well, then that diagram can be instantiated.

08:58And the differences between the instantiation of the diagram can be noted.

09:05And here's a diagram that's a different plan of the same diagram.

09:11And the diagram possibly needs yet another diagram that says they don't have to be parallel.

09:17And the highway doesn't always have to be between the green and the blue.

09:22And they can go under bridges and get intermixed.

09:25Those are design concepts that we want to be able to record and move forward with.

09:30And so, the act of putting down this one is a lot like building a legend.

09:35And whether the legend or the diagram comes first, I don't know.

09:38But making a legend and making a diagram are exchangeable, interactive parts of the design process in important ways.

09:47And there's an instantiation of that, with some green stuff and some gray stuff and a highway.

09:54And on it goes.

09:57Diagrams are subject of active research.

10:01My Ph.D. thesis at MIT had diagrams in the title.

10:04That was 20 years ago, and in the last decade...

10:07...here, for example, is the fifth international conference on diagrams and diagrammatic representation and inference.

10:13The artificial intelligence community is very interested in these things for, I think, exactly the same reasons I've talked about.

10:20The interesting potent inferential role they play...

10:24...in graphical and visual knowledge and intelligence and other domains of argument and representation.

10:34So diagrams are distinguished not so much by their graphic style and conventions...

10:38...although there are many of them, and we teach and promulgate them and sometimes invent and break them...

10:43...but by their roles in design inference.

10:47And they're not sketches. And they're not maps.

10:49But they function halfway in between concepts on the one hand and maps on the other hand.

10:59And that's what makes them, in my view, so potent.

11:03Whether something is a map or a sketch or a diagram depends a little bit on how it's used, not so much...

11:08...you can't always tell by looking at it.

11:12But I believe they're essential in our GeoDesign enterprise, and that they require their own research agenda and software support.

11:21Thank you.

Copyright 2016 Esri
Auto Scroll (on)Enable or disable the automatic scrolling of the transcript text when the video is playing. You can save this option if you login


No comments. Be the first to write one below.

Comment on this Video