You don’t have to give up your privacy for personalization, remote search

People are up in arms about the recent release of Google Desktop 3.0 beta.

There are different sides to the issue.  The EFF has taken a stance, and so have bloggers from the industry rags (like eWeek and InformationWeek). The discussion goes back and forth about whether certain features of Google Desktop 3 compromise your privacy.

Either way this is old news.  Google Desktop 2.0 profiled users over time, gathering data about what they click on so that it could present news clips tailored to end-user interests.

There are a lot of problems with profiling.  But it all boils down to the fact that somewhere, some software has a record of your interests over time.  Someone could steal that record, or subpoena it, or even worse, your interests might change, but you’re stuck as a Britney Spears fan for the rest of your life.

The same issue is present with the new “remote search” feature of Google Desktop 3.  Google says they need an intermediary server in order to make this feature possible.  They also say that you need to have Google Desktop installed on the PC from which you’re doing the searching.  Interesting.  So your data has to go through Google, and you have to install their software on another machine.  The Google desktop website says that all of this is necessary.

The reality is you don’t have to give up your privacy or share your data with third parties in order to get either of these functionalities.

When we launched Watson, we demonstrated you could have personalized content, without sending any data to our servers or by building a user profile that could be inspected.  This is a shameless plug, but we’re proud of it.

Gnutella and other, more legitimate file-swapping utilities like FolderShare show that you can have a distributed architecture for remotely accessing files on your PC.  Google could have developed remote searching with all the benefits of redundancy, etc., with no central server to hack, subpoena, or monitor your activity.  In fact, why didn’t they just allow you to enter a username and password to access the web-based search interface to Google Desktop?  This would have been a simple hack.

There must be something else going on.  I’m not a conspiracy theorist, and don’t think Google is out to harm people or violate their privacy intentionally.  But it is a little suspicious that such smart engineers are willing to risk so many users when there are proven designs that don’t require a single bit be stored on a central server.

-j

Tags: Watson, Privacy, Google Desktop, EFF, Profiles, Search

Comments

[Jeroen]

I think the word 'harm' as in "... (I) don’t think Google is out to harm people ..." is what makes this all so dubious.
Is collecting the information alone harmful? Is targetting you with ads matching your profile across services harmful? Is even spyware by itself harmful?

I think practically, no. I wouldn't sleep a minute less. But from a bigger perspective, it is. Why would I allow some company to target me with information that goes beyond group generalisation and in fact becomes very personal?

Google's ad strategy shows it is very succesful financially, so I'm not surprised they are trying to push it a little more. Looking at the line of products with GMail, it all makes sense. It is where the money is, but whether people should accept this remains questionable to me.

[Watson]

One could argue that user acceptance shows that collecting data isn't harmful in and of itself. However there is potential for collecting data to be harmful should that data fall into the wrong hands.

When bank records are stonlen and criminals steal people's identities they are causing harm. The existence of the bank records enable this. Did the bank participate in harming its customers? Maybe.

But in the bank record case the records are necessary to deliver banking services. My point is that for the current services being delivered in search, this level of record-keeping just isn't necessary.

So why even go there?