Ed Parson's wrote a blog post about Google Research Awards today. What follows is my comment to his post. His blog choked on the length of my comment (with no error message) or he moderates comments. Either way, after clicking "comment", I got no message. So I thought I'd stick my thoughts over here. The comment started off picking a fight over words but shifted to one of my favorite stumps: Google's opacity relative to how disruptive their technology is to GI Science. Comment follows:
Also, reading the University Awards page, I noticed that the funding is really geared to support a PhD student and not faculty salaries. Further, the funding is setup as an unstructured gift to try to bypass University indirects. I wonder how common this style of funding is in Computer Science versus Geography? I wonder how this style of funding is viewed by tenure boards in Computer Science versus Geography? My suspicion (although it's not at all based on knowledge) is that Computer Science departments see a greater "variety" of funding types and that tenure boards are less picky. Geography departments tend to have a very wide variety of faculty. I suspect that winning an "unstructured gift" from a big corporation, like Google, wouldn't go over well with a Marxist Geographer on a tenure board.
And clicking through to the Google University Relations FAQ, I am immediately drawn to the statement:
Google’s Faculty Research Awards program funds work in computer science, engineering, and related fields. Our awards tend to fund projects that are highly technical in nature.There are many GIScience researchers who are comfortable grouping themselves in "related fields" but there is a constant tension in GIScience between "highly technical," which can easily translate into "building tools," and theoretical. GIScience researchers in Geography struggle constantly to maintain that what they study and teach has a unique theoretical basis. To do that, the trend is to avoid the "highly technical" in favor of softer, less technical topics.
For instance, my thesis proposes a new way to structure metadata for spatial data when the data comes from multiple, heterogeneous sources (like Google MapMaker or OpenStreetMap where lots of different people contribute data). My committee member from Computer Science suggested I needed to develop a tool to aggregate the metadata whereas my committee members from Geography who specialize in human geography suggested I needed to survey the people who currently work with this metadata. The advice from Computer Science was to go more technical whereas the advice from Geography was to go less technical.
Another problem with Google's statements about funding is highlighted a little further in the same paragraph:
The work funded through Research Awards tends to be of the type and caliber that the results may be published at top conferences and in top publications in computer science.This relates directly to my post last year where I examined the publication history of Googlers. If you really want to bring more GI Scientists on board with Google's research agenda, you're going to have to make sure Google is fostering the GI Science research community rather than explicitly focusing on Computer Science.
I think there is a world of value both to Google and to academic GI Science if the barriers to the two working together are broken down. And I don't think it would take much to start breaking down those barriers. Just rewording Google's FAQ could elicit more responses from Geographers.
But I think the real watershed event has to come from inside the Googleplex. Google needs to start communicating with academic GI Science about some of the more disruptive technologies it developed. For instance:
- Google Maps has largely demonstrated that the academic GI Science community around automated generalization and multiple representation databases (MRDBs) were too busy navel-gazing. To a large degree, selection is the only generalization method that matters. And MRDBs need to focus on tile caches versus dynamic web mapping services because usability matters more than flexibility.
- Google Maps has also (just about) put the final nail in the coffin of "map scale" (which Mike Goodchild argued was a "legacy concept" over a decade ago). Academic cartography has always stumbled over itself trying to get students to properly use the terms "large scale" and "small scale" (which have the inverse meaning in Cartography as they do in pretty much all other usages). It turns out that map scale is only important when you plan to use a ruler with the map. When was the last time you used a ruler with Google Maps?
- Google MapMaker and the "report an error" link on Google Maps hooks into a verification process inside Google that is so opaque, folks in the OpenStreetMap community assume evil lurks about (especially when Google uses its financial foundation to advocate the use of MapMaker). I shudder to think how much effort I have personally put into the USGS National Map Corps which probably reproduced results that could have been gleaned from Google's internal experiences.
- Google Maps also has made astounding leaps in spatial data integration. Once Google dropped TeleAtlas in the US and started building its own basemap, crazy levels of detailed data cropped up. Property boundaries in a US nationwide dataset? That's bordering on sacrilege. Sure, it's common-place in countries like Sweden where the national mapping agency is the "National Cadastre" but in the US, except for the odd private data reseller, no one really has tried this. What did Google learn in the process? What has the user backlash been like? How about municipalities that try to fund their GIS departments by licensing this data?
- When I was at the 2011 Annual Meeting of the Association of American Geographers in Seattle, WA, I attended a series of panel presentations organized by the AAG Cyberinfrastructure Specialty Group. Up first was an Industry Trends panel in which Kailen Wright talked about how Google Earth Engine can execute raster spatial operations on a global scale. The following panel was Agency Perspectives on CyberGIS in which my boss, Lynn Usery, talked about how one possible outcome of CyberGIS would be the ability to execute raster operations on a global scale. In other worse, Kailen said "Google Earth Engine does this" and then Lynn said "Maybe, with some more research, we might be able to do this". Of course, Lynn missed Kailen's presentation. There are GI Scientists currently working to replicate aspects Google Earth Engine when they could be focused on how to apply and improve Google Earth Engine.