More on the Million Short search engine and on adding search engines to your browser
In my most recent post I introduced the search engine that lets you remove the most popular web sites from its results, Million Short.
Removing those websites should help you counteract some of the bias that search engines may have towards popular websites.
I felt like checking out if the concept actually works, so I wanted to make Million Short my default search engine. The current post is about the roadblocks I encountered.
The first roadblock was adding a new search engine. The only reason this even was a roadblock was because of myself. I had assumed that this would be a tricky process that involved finding a browser addon or writing one of my own, so that is what I set out to do.
Looking in the wrong places meant it took me a while to stumble onto the fact that there is something called the Open Search API, that lets search engine manufacturers offer some basic information about how they work; based on this, browsers already have all the information they need to easily add tons of search engines.
The way this works in Firefox is if you go to the website of a supporting search engine, the magnifying glass icon in Firefox’ search bar will get a green cross overlayed. When you click the icon, you get an extra option “Add NAME”, where NAME is the name of the search engine.
Google Chrome works slightly different: it remembers all the supporting search engines you have come across during your travels on the world wide web and offers them to you on its settings page.
Million Short’s Open Search implementation uses the engine’s default settings and that was the second roadblock. For some reason the default setting of Million Short is to show all the search results and not filter out the top million websites.
One of the cool things about Open Search API is that you can write your own implementation of somebody else’s search engine. All it takes is a copy of the spec and some minor knowledge about how HTML and URLs work. The only caveat is that your implementation has to be served from a web server, the browser won’t let you read the file from a local harddisk.
So I ended up writing my own Open Search implementation for Million Short that adds the query remove=1000000 to the URL.
Million Short also filters your results by country and this too is a default setting. The result was that I sometimes got more Dutch search results than I had bargained for. I am unsure how I feel about this. Google does the same, but doesn’t filter out the million most popular websites. It feels like on Million Short this works as a multiplier and sometimes that is what you want, sometimes not.
Finally, a large stumbling block during the first 3 weeks I used Million Short as my default search engine is that it kept presenting me with a captcha. Since search engines are a utility, you really don’t want them to cause any sort of friction. If the calculator app on my phone kept asking me for a password, for example, I’d install a different app.
After three weeks the captchas mysteriously disappeared, which made Million Short a lot more pleasant to use.
In the end though, I found I kept switching to Google for ‘proper’ search results. I think this is because for many of the things I use a search engine for, the most popular result on the most popular website tends to be the right one.
Million Short presents the user with a number of filters and invites you to suggest your own. I imagine that using these filters (‘facets’ as they are called in search parlance) is the proper (if slightly unwieldy) way to use Million Short.