Friday, 20 January 2017

Why Search Engine Privacy is bad for Search Context

Mention search, and a discussion of privacy usually isn't far behind.

Whether it's Google's (in my opinion unnecessary) masking of search queries for logged-in users, or concerns about using your browsing history to suggest products or even web sites you should visit, it seems everyone has an opinion.

Now search engines are starting to use enhanced privacy in their market differentiation.

Here's how Search Engine Journal's Matt Southern ("Qwant, a French Search Engine, thinks it can take on Google.") puts it:
Like search engine DuckDuckGo, Qwant’s competitive advantage is privacy. It protects users’ privacy by not tracking what they’re doing or searching for online. Qwant doesn’t use cookies, collect browsing data, or do any kind of data profiling.
Sounds great. My privacy is protected.

But that also means that the search engine doesn't know anything about me.

It doesn't know where I am located. Nor does it know the kinds of sites and search results that I've visited in the past.

In fact, all that the search engine has to go on is the page I visited last (the HTTP Referrer) and the query I've just entered.

Search Engine Context, Choice & Privacy

The job of a search engine is to act as a librarian/gatekeeper for all the content in its index.

Whether the engine is responsible for finding a product (i.e. Amazon, eBay, etc.), a person (i.e. LinkedIn, Fiverr, Facebook, etc.) or a resource (Google, Yahoo, Bing, Qwant, DuckDuckGo, et al), the users -- you and me -- expect a result list that contains the most relevant, highest quality, useful and valuable items that they know about.

But what's useful and valuable to me, might not be useful and valuable to you for the same search query.

It's to be hoped that a real world librarian, when asked for a "romance" book, isn't going to give the same recommendations to a teenage girl as they do a middle-aged housewife.

By the same token, knowing the search context is vital to a search engine's success. Building a competitive advantage out of treating all queries (with the same keyword phrase) as identical, to my mind at least, misses the point.

Enhanced privacy has the power to remove context, and thereby render results less relevant to the end user, leading to lower quality results.

It won't just hit organic search, either. It's across the board -- taking paid inclusion and sponsored entries as well as the various AdSense / AdWords style PPC services into account, too.

Squaring the circle requires giving end users the chance to activate more private -- i.e. context-restricted -- browser sessions. It's one area where I agree with Alphabet: making users login to get the most out of a search engine like Google is a good thing.

At the same time, giving users the option to start a private browsing session also solves a lot of the issues that Qwant and DuckDuckGo appear to be building their entire competitive advantage upon.

However, users need to both be educated, and use the private browsing features, as well as encouraged to view access to things like location, language, browsing data and subsequent data profiling as a way to help the search engine better serve their needs.

A Simple Example of Search Context

One example is the "near me" query. As in: "bank near me".

  • My usual logged-in Google session gives me a list of banks that I am a customer at, near my location.

  • Qwant gives me a list of reasonably random US based sites that include the term "bank near me" in their web page. None of them are "near me".

  • A private browsing session with Chrome, using Google, gives me a slightly different list of banks near me -- probably because it doesn't have access to my browsing or search history, so doesn't know which one I favour.

That's the effect of removing context.

As mobile search becomes ever-more popular, the chances of using, say a search engine like Qwant, and getting a reasonable local search result seem, if not remote, then at least less likely.

It's not a reason to drop privacy concerns completely, but it is an argument for education and choice over blanket removal of all search context; something that Qwant ought to take into account if it wants to challenge the established players going forward.