Google Plays Whack-A-Mole?
I had thought I would take the weekend off, but the following blog just forced itself out and you can blame Mike Blumenthal.
I presented an online presentation in the Joint Worldwide Universities Network Virtual Seminars in GI Science and Technology in November of 2008 on Neogeography, which was the focus of the fall 2008 program. My topic, “Data Quality and Neogeography” contained, I thought at the time, lots of interesting material for future research. You can find the presentation here.
One of the comments that I made was that even with the benefits of moolah ($$$) major navigation databases were out-of-date, inaccurate, non-comprehensive, of variable quality and too expensive to maintain. The summary of the presentation was twofold. First, I believed that UGC and neogeographers would become an integral part of the process used to build spatial databases. Next, I concluded that hybrid data collection systems using UGC and controlled data, including formal field gathering and measurement, were where geospatial database building efforts were headed.
Near the end of the paper, I asked “Will the winner of the map building competition be established commercial companies that capitalize on UGC to augment their data or new competitors that commercialize UGC and augment these data to compete with established systems?” Since we invent the future as we move towards it, I decided it was going to be fun to see what happened in the industry over the succeeding two to three years.
Since that time, Google has emerged as the leading new competitor commercializing UGC and augmenting it with more traditional data gathering methods (e.g. field activities (Street View),data mining etc.). On several occasions, I have blogged about Google’s map creation efforts and forecast that they would eventually need to hire an army of map compilers if they were serious about being in the navigation and spatial database building businesses. As you can see below, I should consider opening a fortune telling business, or, as least, posting my “Psychic for Hire” sign. (On the other hand, isn’t that what consultants do anyway? Wait, that includes me! Ugh.)
Late last night (see, there is an advantage to being a night owl), I received a note from Mike of Blumenthal’s blog who was passing on a note from one of his contacts (Chris Silver Smith) with a link to an article in TechFlash titled “Google hiring 300 workers to pinpoint bugs in Google Maps.” This appears to be another of those comedy pitches that are hard to find anywhere other than in the world of mapping.
What, Google needs to hire people to pinpoint the bugs in their maps? I thought that was the reason behind Google’s use of UGC and the map-feedback-loop? However, I have some advice for Google, “Use your maps and you too will be able to find the errors in them.” Oh, wait a minute. You do use your maps to support your local advertising system. And what did you find out? Oh, that bad?
The article from TechFlash starts “Google is hiring an army of 300 temporary workers in Kirkland as part of a yearlong campaign to improve the accuracy of Google Maps. The search giant is working with temp staffing agencies to find “computer geeks” familiar with geographic information systems and mapping software.”
I suppose there must be a couple of million “computer geeks” familiar with GIS and mapping software living in Kirkland, Washington (presumably these are the same people who used to work with the GIS giant Costco in Kirkland). Later in the article, it gets even better when the author of the piece (who seems to have done his homework), indicated that the two search firms helping to fill these position are looking “…to find 300 ‘visual data specialists’ to improve the product.” Jordan Newman, attributed as a “Google spokesman” in the article, apparently said “…much of the work will focus on data management.” The article goes on to indicate that data management includes “…keeping current on new business listings and the latest information on things like bike trails.” Whew, that’s a relief.
So, let’s be sure that we have this straight. Google apparently needs “computer geeks” familiar with GIS and mapping, who are “visual data specialists” and knowledgeable about data management topics such as keeping current on new business listings and the latest information on things like bike trails. You know, in all of my years in maps and mapping, I don’t remember ever meeting a person who had all of those skills, much less finding them in someone applying for a “temporary” job.
Get a realistic strategy, Google! Improving the quality of your maps is not a temporary need. The whole concept of “keeping current” is not an effort you can do in a year and then abandon because you decided a year earlier that this effort could be accomplished in twelve-months.
I realize that the emphasis here is that Google apparently believes it needs to hire a temporary work force to eliminate the map errors they did not catch in the original compilation process used to create the Google Base. However, my belief is that they will find out that research conducted by a professional team of map compilers could be of significant benefit to them. However, Google has a history of not liking solutions that are non-algorithmic. Imagine that, having to hire humans. Time will tell.
The fly in ointment in respect to Google’ approach to navigable map database building is that you cannot build a database of the quality needed for navigation, based on conflating data from public domain sources. In addition, you cannot build a navigation quality database sourced only from User Generated Content. Google, go back and read the slides from the presentation noted above – you need an integrated hybrid approach and a part of that activity is to have talented, professional map compilers who know GEOGRAPHY and how to find “source” on all thing geographic.
To be honest, talented map compilers are hard to find. In fact, you do not find them. Nope, you find people who have the talents and then you train them, because there is no public educational program (university-based or trade school) that trains competent map compilation experts focused on building navigable map databases. The talented map compilers who do exist in the world are classic data geeks, but ones who are interested in geography.
While I do not think it likely that the group being hired by Google will have strong map compilation skills, I do agree that they will be able to help Google build a GIS database with reasonable, but not exceptional quality levels. Building a navigation quality database that provides legal routes and route guidance capabilities, along with accurate address records, well, that’s not something Google is going to get to without different people and different tools.
However, one has to wonder who at Google has the experience that NAVTEQ and Tele Atlas have gained over the last twenty-five years of compiling navigation quality databases? In other words, who at Google is going to train this new, but temporary group of map compilers? Who at Google really understands the map compilation process enough to teach others how to do it at a professional level? Who at Google has experience with field data compilation for building navigation databases? In short, this is not a technology transfer program that can be accomplished in a year. Google needs to build a map compilation branch and the duration of that effort will last as long as Google wants to be in the mapping business.
I suppose many might think it is a “cheap shot” to ask the questions posed above. However, if Google had the answers to these questions before they built the Google Base, they would not be hiring 300 “crack” map compilers after releasing it for public use. After all, as noted in one of the comments to the TechFlash article, Google is the company who brought you Randolph, Vermont the popular city, apparently located at the bottom of Lake Champlain.
Now, once again bringing out my ESP capabilities, I predict that Google has yet to feel the real pain of being a quality cartographic provider (maybe, because the company does not yet merit that title). Just the other day, I saw a note indicating that the government of Vietnam had complained about Google’s depiction of the Vietnam/China border. I wonder who is going to research that question and others like it for Google?
In a similar vein, I wonder how UGC, the use of imagery and Street View are going to help Google answer questions about geographic features that are not visible, like boundaries. Guess that means greater reliance on data mining. Guess that, also, means greater reliance on authoritative sources. Guess that means more reliance on metadata? Google? Metadata? Do the two go together? Hmmmm.