Timeo Danaos… Or: Why I’m Wary of Google Docs

During the last couple of weeks I had to use Google Docs quite heavily in order to work on a large, collaborative document. Without any doubt, Google Docs is a convenient platform for collaborative work, offering collaborative real-time, browser-based, WYSIWYG editing for various types of documents. However, the experience was not a good one, as the 150-page document clearly tested the limits of Google Docs. Loading the document was slow; navigating even more so. I wasted a lot of time trying to export the document as PDF file; sometimes it worked, but most of the time, I could only download an empty file, regardless of whether I used Safari or Google’s own Chrome browser. For editing, however, Chrome eventually turned out to be required, and I was forced to use it, even though I have serious issues with Chrome.

What’s the problem with Chrome? It’s not the browser itself, but rather the computing and business models of which it is an important component. The computing model is that of the mainframe (or the mini): You log into a large central computer using your dumb terminal in order to do your work; computing power and storage are centralized on the mainframe.

The mainframe is now called “the cloud,” and the terminal is now called “Chrome” (and the Chromebook actually is a physical Google terminal) and it’s all Web-based and graphical and fancy, but the model underlying Google Docs, Gmail, etc. is essentially the same. Now, the centralized computing model certainly isn’t all bad, but it’s clear that it gives users less control than a distributed model. And of course, the mainframe’s admins have full control, including access to all users’ data, whereas users have next to no control—they can’t even delete their own data, as the admins will always have a backup. Again, this is not necessarily a problem if your sysadmin is trustworthy and not in the business of selling your data. In the case of Google Docs, Google is your sysadmin, and—wait: Google’s business is selling your data.

Now, all this isn’t new. Eight years ago, Edward F. Moltzen already talked about the “Google mainframe,” and five years ago, Jackson West already made essentially the same point that I’m trying to make here. Chrome has two important functions for Google: First, it allows them to track your behavior even when you’re not logged into a Google service; second, while Google’s applications can still be used to some extent with other browsers, eventually they will only be usable with Chrome (try the Google Maps preview with Safari, for example), so people will have to use Chrome to access them, enabling Google to track more people, lock them into their applications and services, and thus sell more advertising. From a business perspective this makes perfect sense, Google is a for-profit company, so why should they not go that way?

As I said, all this isn’t new. It seems, however, that surprisingly few people are aware of this. Don’t get me wrong, I’m not an anti-Google zealot; for some use cases, the deal offered by Google is certainly good, and I’m using Google services myself. On the other hand, there are use cases for which you probably shouldn’t use them; for example, is it really a good idea to store confidential research documents on Google Docs? Google is mining documents and messages for advertising-relevant keywords; Google provides U.S. government agencies with metadata and data of users; while it would probably seriously hurt Google’s business if there were any indications that they also mined users’ data for purposes of competitive intelligence, are there any guarantees that nobody in this huge company is doing some analyses “on the side”? Say, on foreign users’ data, in a patriotic effort to support U.S. research and industry? Remember when Google recorded WLAN traffic? Company officials claimed that they had not been aware of this—and it may be true—but someone at Google had put in the data collection code.

What could be an alternative to using Google Docs? Well, how about writing your documents in LaTeX (or Markdown) and checking them into a Git or Subversion repository on a server you control? Not user-friendly? Well, how user-friendly is a word processor that’s slow as molasses, can’t print, doesn’t support cross-references, floating figures, formulas, or bibliographies, and stores all your data on some foreign corporate entity’s servers?


Leave a Reply

Your email address will not be published. Required fields are marked *