projections in web browsers are terrible and you should be ashamed of yourself
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
| Title |
| |
| Title of Series | ||
| Number of Parts | 188 | |
| Author | ||
| License | CC Attribution 3.0 Germany: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. | |
| Identifiers | 10.5446/31763 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language | ||
| Producer | ||
| Production Year | 2014 | |
| Production Place | Portland, Oregon, United States of America |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
| |
| Keywords |
00:00
Projective planeWeb browserComputer animation
00:17
Presentation of a groupProduct (business)Revision controlComputer fileXML
00:38
Web 2.0File formatPlug-in (computing)State of matter
00:54
State of matterQuicksortNumberProjective planeComputer animation
01:21
Physical systemEndliche ModelltheorieShape (magazine)Multiplication signLipschitz-StetigkeitXML
01:55
Point (geometry)QuicksortPhysical systemSphereExecution unitCoordinate system
02:20
Transformation (genetics)Euklidischer RingPlastikkarteXML
02:36
Different (Kate Ryan album)Projective planeVideo gameWorld Wide Web ConsortiumQuicksortPhysical systemRadiusComputer animation
03:04
NeuroinformatikMathematicsCalculationComputer animationXML
03:22
NeuroinformatikLoginXML
03:40
Social classMetreMassProduct (business)LoginSoftwareLecture/Conference
04:20
Projective planeTrigonometrySupercomputerState of matterMultiplication signComputer animationJSON
04:45
Matrix (mathematics)NeuroinformatikInternetworkingQuicksortMathematicsShared memoryDynamical system
05:02
PlanningState of matter1 (number)MereologyRecurrence relationMetrePhysical system
05:46
Revision controlMetreWord
06:08
SphereMathematicsFigurate numberSpacetimeRevision controlLecture/Conference
06:28
VariancePhysical systemState of matterPlanningProjective planeDifferent (Kate Ryan album)Computer animation
06:49
Projective planePredictability
07:09
Fiber (mathematics)State of matterPlanningMassSelf-organizationWebsiteServer (computing)HypermediaWeb 2.0Computer animation
07:51
Object (grammar)HypermediaComputer animation
08:32
MassState of matterProjective planePlanningKeyboard shortcutComputer animation
08:51
SummierbarkeitWeb 2.0
09:11
Projective planePole (complex analysis)Metric systemBijectionRevision controlExecution unitProduct (business)
09:51
String (computer science)Web browserDatabasePositional notationDifferent (Kate Ryan album)
10:12
GodQuicksortFile formatDifferent (Kate Ryan album)MeasurementString (computer science)
10:41
QuicksortData storage deviceDifferent (Kate Ryan album)Data miningMereologyFile formatSubject indexingLecture/Conference
11:10
Revision controlVideo gameWordFile formatQuicksortState of matter
11:53
PredictabilitySlide rulePresentation of a groupComputer animation
12:27
Video projectorArc (geometry)State of matterData conversionMappingMetreDirection (geometry)Error messageMultiplication signProduct (business)Projective planePresentation of a group
13:22
World Wide Web ConsortiumLocal ringPlanningState of matterTesselationLecture/ConferenceComputer animation
13:44
WordControl flowPole (complex analysis)World Wide Web ConsortiumTesselationWeb 2.0Level (video gaming)Lipschitz-Stetigkeit
14:14
NP-hardRight angleVideo gameProjective planePosition operatorShape (magazine)Pole (complex analysis)Lattice (order)Product (business)TheoremWordFilm editingDatabaseMathematical singularityAngleRectangleAreaConsistencyPartition (number theory)QuicksortMappingSpherical cap
15:48
Extension (kinesiology)Multiplication signLevel (video gaming)NeuroinformatikRoboticsWorld Wide Web ConsortiumQuicksortProjective planeAdaptive behaviorMoment <Mathematik>
16:57
Projective planeQuicksortCASE <Informatik>Scaling (geometry)Radical (chemistry)Selectivity (electronic)ArmLevel (video gaming)Data conversionAreaInteractive televisionVolume (thermodynamics)TesselationClient (computing)Exception handlingTwitterVector spacePresentation of a groupParameter (computer programming)Standard deviationWeb serviceEndliche ModelltheorieMassCuboidFile formatWebsiteDreiecksmatrixEntire functionAddress spaceWeb 2.0Line (geometry)Different (Kate Ryan album)Pole (complex analysis)Web pageDistortion (mathematics)GeometryGraph coloringGoodness of fitAdaptive behaviorTriangleComputer animation
21:14
Shift operatorComputer animation
Transcript: English(auto-generated)
00:03
Welcome to projections on the web browsers are terrible. And you should be ashamed of yourself. So my name is Calvin. You might know me, and I have opinions.
00:21
You might know me from my previous presentation. Things you haven't heard of. Not really explained to me. Also the new version of product 4.js is not terrible. That terrible, I should say. And leaflet plug-ins for file formats that really should not
00:42
ever be used on the web. First, I'd like to apologize in advance to non-coastal states. It's not you. It's that state next to you that you don't really like.
01:05
So at a geoconference, you have a surprising number of people that actually don't really know what projections are. They are these things that they've heard of that they might have battled with. But they're just sort of a boogeyman there.
01:24
What they usually mean is a spatial reference system, which has a bunch of stuff in it, like a spheroid, which is a model of how the Earth is shaped. Because as a GIS person, whose true GIS person will
01:41
extend you in very explicit detail, the Earth is not a sphere, and it's on the lips. It's a geoid, which is a shape that's shaped like the Earth. So you also got a datum, which is a way to sort of reference points measured on the sphere.
02:06
It's sort of basically a way of saying, we're going to start here. And then you got a coordinate reference system, which are units and starting points and how you actually talk about stuff.
02:20
And also, a transform, which is how you flatten it, because it's how you go from the orange to the example of the orange. And then for the sticklers, when I refer to WGS as a projection, I'm talking about card player, however you pronounce that.
02:42
For the purposes of like most people trying to get stuff done, the difference between WGS and Web Mercator is not actually a difference. They both are sort of projections in people's mind, even though WGS is not a projection, it's a coordinate system.
03:00
Nobody cares in real life about that difference. So the background. Math is hard, especially if you don't have a computer. Trig and logs, those are really hard without a calculator.
03:24
The very first computers were literally invented to do logs. This is not like, these are hard problems to do without a computer. So traditionally, projecting something is hard, because nothing but trig, logs, stuff like that.
03:45
An agency picks a projection and does everything in that one. So I used to work for the Massachusetts Department of Transportation. Everything was Massachusetts mainland meters, 983 classic, that was everything we did.
04:01
This is great. If the printer is the way you transfer data. And up until 20 years ago, that was all you did. And desktop GIS software is still built around this concept.
04:20
And there are assumptions that projections rarely change. And they are never to be mixed because trig. So, fast-forward to now. My cell phone is literally a super computer. There were times where my cell phone was faster than my desktop computer, where I was doing GIS
04:41
for the state. Your desktop computer can run other computers inside it like it's the Matrix. And there's the internet. This sort of changes some of the dynamics of data sharing.
05:04
But everybody still has their own projection. There's 124 different equations for transforming spherical coordinates into flat ones for the state plane system. They basically make it so that the part of your data
05:23
that you're in, you can pretend that it's a piece of paper. And they are in transverse locator for tall states. And they are in Lambert Conformal, something else that starts at the sea, for wide states. And because the Alaskan Panhandle is the sole place
05:40
in America that's diagonal, that has oblique Mercator. But there's also feet and meter versions. These are not the same because everybody uses meters in Massachusetts except the Boston Police Department. They use feet, but they don't say that anywhere.
06:00
If you try to take their XY coordinates, it's somewhere off Newfoundland. Also, there are two different datas, data. There's NAD 27, which is the original one. And then there's the new version, NAD 83. That's North American datum. It's basically saying, let's figure out
06:23
how the sphere works for us. And then, there's a lot of math there. But there are two variants. There's the high accuracy something something, and then the new something reference system, 2007 variants.
06:44
Which means you have 992 different state plane projections. And that's just America. Don't get me started on UTM. Sweden has a lot of projections. There's a special one that's just for Switzerland
07:02
that all the documentation is in German. Yeah, there's a lot. And this is terrible. Nobody cares about your flyover state plane. Nobody.
07:26
This is a lot of the problems that a lot of governmental organizations do not realize that there are people that aren't in their organization. I used to be the person that had to take the mass meters, road data,
07:41
and put it up on the website for people to download, because people want that in that way. So, the thing is that if you need precision that you can't get with doubles and WGS84,
08:00
and you have data that the accuracy matters, then if you don't use WGS84 lat longs, you're just making it hard for people, because usually most of this, a lot of this authoritative data, the precision is like, oh yeah, but we have a plus or minus 10 feet.
08:23
But we have to be accurate enough with our projected stuff, because it might be off. But, you know, it's plus or minus 10 feet. So these are cargo cult projections. It's that people have always been using mass to use a state plane, or whatever your state plane is,
08:43
or whatever the thing that they've been using, they just use it, and they don't realize that nobody else cares. They're confusing, they're complex, and this is just on the desktop. On the web, they're really unworkable.
09:01
Self-documenting data is folly. It's, we've learned lessons from XML to never, well we're gonna do them again in like five years, but you know, there's a long tail of crazy that you really, to get something that just works, there's, you can get something,
09:21
you know, the version of products for JS that gets, you know, 80% of all projections, that's like less than 10K. The version, the full version is like 60. Because you know, you have the projections for, you know, near the poles. You have the projections for Switzerland. You have the projections for weird units
09:42
in the Caribbean that I'd never heard of, but you know, there are, they're like metric furlongs, I don't even know what they are. And the EPSG database, you know, are too big. You know, you can't just put that in a web browser and expect it to work, you know, on a mobile. You know, that's, it's a meg for the one with prod strings.
10:04
Because I tried that. And there, because there's also three different notations that people use, and then they're not consistent. You've got prod strings, you've got well-known text,
10:22
and you've got well-known text that comes from Esri stuff, you know. And they're literally three different formats that you have to sort of deal with differently. They give you different things for the same projection. They, that's, in many ways, one of the biggest problems with prods4js is just having to deal with all of that.
10:43
But the thing is that storage and interchange are sort of different things. When I send you data, I don't give you stuff about, details about how it's indexed on mine. You know, that's a sort of part of storing it. It's not part of, you know, giving it to somebody else.
11:00
Of course, shapefiles do that, so. Sort of sane formats don't do that. So, GeoJSON, the, I think 1.1 or two, or the newest version does not just have CRS support.
11:21
It's WGS 84 for life, there. And this is not terrible. This seems like a step backwards in some ways, as a sort of a GIS format that you don't have that, like, bread and butter of GIS, by the hand. When you're sharing these kind of things on the web, nobody cares that you use flyover state plane,
11:42
you know, furlongs in your town, you know. The rest of us are just gonna use what we use. What, you know, you can do what you want when you actually put it on the map. KML was always WGS 84 only, also. But KML's terrible for other unrelated reasons.
12:03
A lot of this really makes us forget. Predictions actually serve a purpose. You know, I really couldn't say it much better than he had the entire keynote, and hence why about half of my presentation
12:21
has now been replaced by this slide. But, you know, there are some, A, innovative stuff that you can do with these projections that, you know, don't, that actually, you know, help your maps work better.
12:42
You know, everybody talks about the project, the conversations about projections that people have in, like, middle school end up being much more, have a lot more substance to them than a lot of the conversations we have about projections and GIS. You know, projections, you know, give hidden meaning, they do all that stuff, and most of the time it's just like, oh yes, but this agency uses feet,
13:03
and we wanna use meters, because Esri gives us an error message if we try to mix projections, so we don't wanna do that. Which is literally a conversation I had once. Except they called it GIS, not Esri, because in state you often see people use GIS as a synonym for Esri ArcMap.
13:24
So, Web Mercator doesn't actually work for everybody. A lot of people say, I've heard, you know, some people say, you know, let's just use Web Mercator. And no, I'm not saying that it doesn't work for your, you know, local state plane, because, you know, there are, you know, states wanna have, you know, tiles in, you know, Nevada state plane, and you don't need
13:41
Nevada state plane, you know, you don't. Let's try telling that to, you know, a Finnish person. They don't like Web Mercator, you know, telling them to use Web Mercator, that's fun words, because the polls are where that kind of breaks down.
14:00
I'd like to see somebody do a web map of McMurdo base. Seriously, you know, I will give you $5 if you can get me one in the next five minutes, tiled. You know, these are, you know, hard problems that people, that projections have trade-offs,
14:21
and they solve, so, you know, there's a thing in databases, the cap theorem, it's the something or other, something or other partition tolerance. It's the consistency, availability, partition tolerance.
14:43
Pick two. You know, if you make a database, you can have two of those. So with projections, it's about like seven things, and you can have five, but it's the same sort of thing, you know, you have area, you can have angle, you can have fancy words for angles, you can have shape, you can have, you know, the polls,
15:02
you can have various, you can have a rectangle, you can have them all, and that different maps have different trade-offs. I once heard a very great talk about somebody doing sea lane, it was all about like shipping lanes and stuff, so they had a, and the whole big thing was there, was that every single projection has a cut in the ocean,
15:24
except one that stretches out from the North Pole looking down, because if you have the, because they don't care about anything on land, so there's that one projection that has a singularity in Antarctica on the South Pole, but all of the oceans are contiguous,
15:40
and that's, you know, never a projection you're probably gonna ever use, but if you ever have to do sea lane stuff, it's gonna save your life. And there are also, we've heard some other talks of adaptive composite projections of, you know, I have a robot that bakes me bread when I wake up, you know, this is the future,
16:08
the computer can figure out based on what map extent is, how to make it look good, you know, I don't understand why there's like a paper that does it to one image,
16:21
and that's sort of what we have, you know, that, you know, we're gonna be seeing this in like three to five years, they're all gonna, nobody's gonna care about, you know, Web Mercator, except when they have to pre-cash their tiles, you know, because there's gonna be something cool that maybe somebody in this room is gonna actually invent, or maybe somebody that knows somebody
16:41
that's like unemployed, and doesn't have to spend all their time making money, like I end up having to do, and doing less cool stuff these days. But yeah, you know, this, that's what's gonna happen. So go forth and be less terrible.
17:01
That's sort of what I want people to get out of this, and use projections for what they're for, and not because, you know, they're there. That's it.
17:24
Questions, comments, complaints?
18:00
There's that West Wing episode,
18:03
that everybody always sends me. No, it's, the Wikipedia article on projections has remarkably like good documentation of like, this is for this. So if you then have to go back and be like, okay, what does constant bearing actually mean? Then there's some stuff on radicalcartography.org,
18:23
has some stuff on what projections are for, line with maps, what that guy said. Somebody, that guy seems to have an idea of it.
18:42
Bernie Jenny, the guy who brought us adaptive composite map projections in the first place, also put together a tool for projection selection at different scales and geographic areas. It's an interactive tool, web map tool, that actually tells you what appropriate projections for those parameters are.
19:02
I don't have the address in front of me, unfortunately. I'm trying to find it, but if you go to his website, I'm sure there's a link. All right, so that guy did the color brewer of projections, it sounds like. Yeah, yeah, he did, basically. It's very comparable. I remember in February, there's the cardioid projections
19:22
that if you put on your Twitter timeline, I'm gonna unfollow you. Is there a question over there? If it is there, as you walked.
19:41
Curious about our tile map standards and how we've kind of baked distortion at the poles into that entire standard. I was kind of expecting the vector tile conversation to at least mention abandoning that.
20:01
If we are moving towards WebGL and vector tiles, we should be able to also tile our data in a triangular format, something that isn't embracing the distortion that we have currently.
20:21
So one of the things about tile map services is that they get it done and that Mapbox doesn't have any customers in McMurdo, as far as I know. So they are totally fine not caring about the poles. And I don't care about the poles, I've yet to have any clients in,
20:42
or my company, AppGeo, plug. We don't do anything in the poles. And so for America, for Europe, most of Europe, and it works for Scandinavia, it's fine. It gets stuff done so that one of the most important
21:05
thing about a format is not the actual format, it's the people using the format and tile map services are the worst tile format except for all the others, and it gets stuff done. So that's probably why they haven't bothered fixing it.
21:20
Somebody can come up with a better one, but nobody has.