Google Map Maker’s Edit and Authority System – Part 1
This and the blog I will publish tomorrow are part of a series examining Google Map Maker and how its edit and authority systems function. I apologize for the delay in posting this work, but I decided to wait for the reviews of my edits, as the reviewing process is the most critical component of the Map Maker system. After I finished writing my analysis, I realized that the document was too long and have chosen to publish it in a two installments, one today and the next tomorrow.
1. I am not an investigative reporter and do not look for trouble on purpose. However, I find that when you look, poke and prod, you often find things that you did not expect. This length of this blog is a prime example. I had not expected the topic to require more than one essay. I was wrong.
2. My purpose in contributing edits to Map Maker was focused on finding out how they would evaluate these edits and accept or reject them. As a consequence, tomorrow’s installment provides a look at the authority system behind Map Maker, while today’s installment is a brief review of what I encountered when I tried to use Map Maker to edit the Google Map Base.
3. I apologize in advance for generalizing about some of my conclusions. However, when you read the information described in this series, I think you will agree that the conclusions are reasonable and, perhaps, something that someone should look at in more depth.
4. Google, you put this stuff out there, so blame yourself.
5. In the way of background, the Google Map Maker software was invented at Google India, in Bangalore by Alit Karaganda and Dimple Bart and launched in India in 2006. An article at livemint.com provides details of the history
6. I looked at the Google Map Maker videos on You Tube, read the pages for beginners and the available help files. None of these were very detailed, but it looked like that was as much as I was going to find. There may be other materials that help the newbie figure out how to work with Map Maker and I apologize if I missed these documents. But what the heck, let’s see how Map Maker works.
I went to Google Maps, zoomed into Mission Viejo, California, clicked on the “Report a problem” tab at the bottom of the page, and, then, clicked on the “Now you can edit the map yourself, on Google Map Maker”. After having tried to use Map Maker to correct Google Maps, I think the ‘Map Maker link’ should be reworded and phrased as a disclaimer. Perhaps something like this would work.
“You (as in you yourself) might or might not be able to edit this map using Google Map Maker. However, in either case, you should proceed only if you have a high tolerance for goofy editing systems and you are willing to ignore the numerous mistakes in our map database that we will not, under any circumstances, let you correct. Further, our trusted reviewer of your work might or might not be personally familiar with your local geographic area and might not find your edit acceptable, even though it reflects reality.
Further, in order to use this system, you must sign-in to your Google account to establish your identity, even though we might regard your edits as ‘spam edits’ and disregard them. This cute little trick helps us to increase the number of our Gmail accounts and allows us to read your Gmail -email to better target local ads for goods and services that you may or may not be able to find on our maps, regardless of your intent to help us improve the representation of local geography in Google Maps.”
However, since there was no disclaimer to be found, I decided to attempt to use Google Map Maker to enter several map corrections I had researched. Unfortunately, I found the system so lacking and difficult to work with that I gave up after trying to deal only with three simple edits. I decided that attempting the complex edits I had planned was asking too much of a volunteer.
I had expected some difficulties, but Map Maker just isn’t that hunk of burning funk for which I was hoping. Let me be blunt, based on my experience, Google Map Maker is a system put together by software engineers who apparently do not understand best practices in map compilation. Nor do they appear to have spent any measurable time studying the Human Factors aspects of computer-user interfaces. On the other hand, I guess if you had never used anything else to edit maps, Google Map Maker might seem like a godsend. In a similar vein, you could tell a person who had never seen bacon before that it was Zebra meat, and they might accept your description. But perhaps, I have put the cart before the horse. Let’s look at the simple edits I tried using the Map Maker edit tools.
First, I had noticed that some streets in my development had small cul-de-sacs that were unnamed on Google Maps and thought that I would help them out by supplying the names. To make sure that they were not named, I zoomed in as far as possible and, sure enough, names did not appear on these stub-streets. However, when I put the icon of the “Street View man” on these stubs, I soon found out that Google Maps did know the street names. In addition, sometimes when I right-clicked the stubs and queried ‘what’s here’, the service would provide me the name of the stub.
I find this lack of name display curious. At several of Google Maps’ higher zoom levels there would be no difficulty rendering these street names on the display, nor any visual design reason not to do so. I guess Google must regard them as undesirable map clutter.
So, Google put this in your Map Maker play book – if you want people to edit your maps, show them what is on your maps. Don’t make them hunt for data to correct, as they have better things to do with their time. (While I am on this brief diversion, the mapping functionality on Navteq’s website did not show the stub names for the same streets, leading to this question, “Why are the display properties set for these two map databases identical?” Could it be…? Oh well, that’s another topic.)
Scanning further around the neighborhoods with which I am familiar, I found a blunder that I thought would be easy to fix. While representing the parking lot associated with a local medical office complex, Google had mistakenly shown one of the lanes in the parking lot as connecting to a local thoroughfare. Below are two photos taken the day of the edit, showing the end of the lane in the parking area from the west (the parking lot side) and the east (from the street-side). It is relatively easy to see from the photographic evidence that the parking lot lane does not connect with the street (Marguerite Parkway).
The Google map of the parking lot is shown below and the lane in the parking lot is shown as intersecting with Marguerite, rather than ending approximately twenty feet to the west, as it does in reality.
Depending on whether or not the designers of the Map Maker system contemplated allowing users to interact with connectivity and topology, I realized that I might need to edit the intersection with Marguerite, as well as the line segment itself. So, I started with the Intersection, but that did not work out well.
After grabbing the intersection, I noticed the error message below and could find no way to recover the situation or to save my work. Oh well.
The error message reads “Incorrect address. Internal Error: Bad Data (Bad Feature). Please report this error with a link to this page.” I had no idea to whom I was supposed to report the error while including a link to the page. I admit, I was shocked. Has Google not heard of error trapping? Does the system not store the what, where and the specific page where I was editing? Curious.
I thought that, perhaps, the next best path to take would be to edit the street segment, rather than the inappropriate intersection. However, when I attempted to do so, the edit page revealed that I was editing Multiple Road Sections and warned that “The geometry of this feature cannot be modified.” Hmm. Well, then, what exactly was I supposed to edit to remedy the situation? I looked for alternatives, but there appeared to be none that were appropriate. The only option that made much sense was to leave both a description of the error and a separate comment indicating that the segment was incorrectly represented and that it did not intersect with Marguerite, noting that the lane terminated at the edge of the parking area. I did so and saved the information for Google’s review.
I have to admit that the intersection error continued to nag at me, so I went back for a second attempt at editing the intersection. However, in the Google Map Maker system, it appears that if someone is editing any line associated with the intersection, you cannot edit the intersection that is attached to it until a moderator decides whether or not the edit suggested for the line segment will be accepted. Locking features can be a good idea in interactive systems, but not in one in which the moderation necessary to release the segment takes over a week, as happened in this case. Well, I have no one to blame but myself, but it is hard to imagine that leaving a comment and not changing any element would lock an edit system.
I guess this is an example of Map Maker’s highly inefficient, volunteer, moderator-gated approach to map editing. After all, I and other contributors have nothing better to do than to come back at some later point in time to make another attempt at correcting the issue that I could have corrected when I was at the website and which they could have cued-up for future resolution. However, that would take more “programming smarts” than Google apparently decided to put in Map Maker. Or maybe this just reflects that the edit system is so heavily moderator driven that corrections depend, at least theoretically, more on the local knowledge of the moderator than on the topological data available to the system. Unfortunately, the speed of the process, also, seems gated by the availability of moderators who can evaluate the validity of the contributed edits.
A mere seven days later, the edit for this feature was accepted by “trusted reviewer” Hemant. Unfortunately, the road geometry remained unchanged on Google Maps and the parking lot lane still intersected with Marguerite. In response, I wanted to tell them that the critical issue had not been resolved by editing, but apparently could do so only by editing my previous action, which you might remember was a comment. However, once again, it appeared that all I could do in this situation was to add another comment, which, of course, locked me out, once again, from editing the intersection.
However two days later, a new “trusted reviewer” identified as Nigar must have read my second comment and responded “Hi, as per the input provided by you and the street view I get to see that this segment of road does not intersect with the ‘Marguerite Pkwy’. I am approving the edit and making the necessary change. Thanks.” Unfortunately, the intersection has still not been corrected on Google four days after the receipt of the approval of the edit.
I had better luck suggesting to Google where the actual access to the parking lot in the medical office complex was located.
When I reviewed my edit a few days later, I was shocked to learn that I had picked up fluency in a new language that also includes unique symbolic phrases. How could a reviewer make sense of this gibberish?
However, the edit was accepted nine days later, by “trusted reviewer” Shalini, who provided this note “Hi, thanks for the edit. The ‘Segment usage’ can be left as ‘None’ and there is an intersection error. The road need to be connected to near by road. I will do the edits for you and approve.”
Yes, all true. I tried to grab the end of road to extend it to intersect Marguerite, but simply submitted it as it was after several tries where it appeared that my mouse’s on-screen pointer must have been coated with grease. Curiously, I have heard from others that they seem to be having the same problem with Map Maker when attempting to extend lines by grabbing a node and moving it.
Next, I tried to add an underpass/tunnel that crosses under a local, divided street. The tunnel under Oso Parkway is actually a horse crossing tunnel, but one that is used by pedestrians who like to walk along the horse paths in this area. In order to accomplish this task I added a line to show the beginning and ending of the tunnel, but soon found out that there was no attribute for a “horse underpass” (at least not that I could find). I decided that I would categorize it as a Pedestrian/Bicycle underpass and correct the notion with a comment.
However when I later reviewed my comment, I found that I was no longer writing Mandarin, but I had now somehow decided to write in a form of HTML. How was a reviewer supposed to understand this gobbledygook?
This html-speak scrolled so far beyond my endurance that, so I soon lost interest in whatever it may have been attempting to tell me.
You may have noted from the image that the underpass is located in Galivan, Ladera Ranch, notations which were provided by Google, followed by the names of locations that I added during the edit process.
Galivan is designated by the U.S. Census as a Populated (Community) Place or U6, which is defined by the Census as “A populated place that is not a census designated or incorporated place having an official federally recognized name.” Google, following Navteq’s lead (hmm – or maybe that’s everyone following GDT’s or, perhaps, ETAK’s lead) shows Galivan on their maps.
I’ve got news for the Census, and all the mapmakers. Galivan is not a populated place. Even the commercial Storage Center that is within fifty feet of the center of Galivan is not in Galivan. As a matter of fact, Google shows Galivan right between a pair of railroad tracks. Of course, since this is railroad right-of-way, there are no buildings here that could be construed as Galivan.
There is no Galivan. It does not have a federally recognized name, because it appears there is nothing to recognize – no people, no home, no boundaries, nada. Of course, everyone can be forgiven for this since Galivan is listed in the Geographic Names Information System (GNIS) maintained by the USGS. However, if you look closely, you will find that this record was first picked up in 1981 from a 1:24000 Topo Map and the Decision Card on the feature reads “No Data Found”. Oh those cartographers. What a sense of humor. In addition, what a cruel trick to play on those conflation routines! Of course, this is exactly the type of situation that could benefit from local knowledge, if only Map Maker effectively solicited local knowledge.
Ladera Ranch, the other entity mentioned by Google as a related location in the tunnel image, is an unincorporated, planned-community (no official boundaries), which, unfortunately for Google, is located in the eastern portion of Mission Viejo, which is east of Laguna Niguel, which is located east of the mysterious Galivan, which is located east of Laguna Hills, where the underpass occurs within the borders of a Laguna Hills community known as Nellie Gail.
However, my edit was accepted after 9 days by the “trusted reviewer” Abhilash. Well, the tunnel was accepted, but not my information about the horse crossing or Galivan Yep, the accepted feature is a pedestrian crossing located with this string of modifiers “Galivan, Ladera Ranch, Laguna Hills, Orange, California, United States 92653.” It seems that my input on localities made the situation even worse.
As noted at the start of this blog, I gave up my editing career after these three attempts. I admit that I do not have a high tolerance for the kind of goofiness that I ran into with Map Maker and others may have more success than I. Another reason that I gave up was that the feedback was taking so long to get back me that I lost my enthusiasm for the process.
However, when I got the feedback and saw the reviewer comments, I started thinking about the feedback and the data I had provided in the edits and began wondering, “How did they decide to approve my edits?”
The tunnel edit was particularly interesting since the feature cannot be seen in Google’s satellite imagery, nor can it be seen in the Street View Imagery, raising the question, as to what evidence “trusted reviewer” Abhilash relied on, other than my say so. Since this was my second edit, I had presumed that I had no “cred” in the system. How, then did they decide? I’ll publish a critical analysis of the Map Maker Authority System tomorrow. Trust me, you won’t want to miss it. It’s a corker.