PGT-PPP (Pretty Good Translation, Pretty Poor Privacy)
What’s in a URL? To someone who gains access to your computer, your browsing history will provide interesting information. More interesting, in some cases, than you might think.
Try Google Translate, for example. Say you want to translate “To be or not to be, that is the question” into the language of your choice. Go to http://translate.google.com, type your text, select the source language (actually you can skip this step, the tool will detect the language automatically) and the target language (that you will have to do, Google does not read into your mind yet). You get the translation; rather, a Pretty Good Translation, almost never quite right in my experience, but sufficient to give you a Pretty Good Idea. This is the result of modern work in computational linguistics, based on statistics and large-scale data mining rather than a traditional syntax-directed attempt at perfection.
Now look into the URL:
http://translate.google.com/#auto|sv|To%20be%20or%20not%20to%20be%2C%20that%20is%20the%20question
Your text is encoded in it! This is true even for very long texts. Building up such complex URLs is one of the time-honored techniques for simulating state in the stateless HTTP protocol. But now anyone who sees your browsing history will know the precise texts that interested you enough to make you want to translate them. A Pretty Good Window on your personal interests! And not necessarily something that you want automatically archived.
Google Translate and other translation sites are great tools to facilitate our life, especially when dealing with languages we know superficially or not at all. But maybe there is a way to provide the service without opening such a large window on the detailed questions that occupy our minds?
Eiffel Software
Home page at ETH
Informatics Europe
LASER summer school
Propulsion Academy