Hooo-ah. Just got back from bushwhacking through the far-right fringes of online conspiracy-theorizing.
It all started with StepRep. A while back, as part of our Great Dog Food Experiment, I “adopted” a used bookstore called the Book Bin, in Salem, Oregon. I created an account for the Book Bin and started monitoring their reputation using StepRep. Why? To get a feel for how everyday users interact with the product. To catch bugs. To come up with ideas for how it can be improved.
This week in my top keywords list I noticed something strange.
What the…?
That’s weird.
Clicking through to the page in question, part of an online pamphlet called The Globalists and the Islamists: Fomenting the “Clash of Civilizations” for a New World Order by end-of-the-world expecter Peter Goodgame, I tried to figure out what this weirdness was doing among my search results.
Goodgame’s thesis is that “militant Islam has been a card played by the global elites of the dominant Anglo-American establishment to achieve the long-term goal of a world government.” If you’re wondering whether the Freemasons, the British royal family, and the Bilderberg Group have a part in this plot, rest assured – they do. But how is Salem’s favourite used bookstore tangled up in the conspiracy?
With the help of Ctrl+F, I found the following passage:
However, bin Laden’s time in London has since been confirmed by Saudi-based journalist Adam Robinson in his book Bin Laden – Behind the Mask of the Terrorist.
Then this:
[I]n 1973 the Islamic Council of Europe was created with headquarters in London. The Council’s long-time Secretary General was a prominent Muslim Brother by the name of Salem Azzam …
So there you go. The name of the business, plus the name of the city where it’s located, appearing together on a single page. For a small business without a lot of online mentions, that’s enough to make it into this week’s top keywords list.
You see, at its core, StepRep is pretty simple. It collects data through queries to Google, Bing, and all the other major search engines. If you didn’t mind submitting multiple queries to multiple search engines all day long every single day, and sorting through all the hits to eliminate duplicates, you could get the same results StepRep does.
After the results are gathered, StepRep does some data analysis to determine sentiment and relevancy. But as the Bin Laden example illustrates, it’s not foolproof. The problem is that StepRep isn’t smart enough (yet) to figure out from context that the phrase “his book Bin Laden” has nothing to do with a store called the Book Bin.
Is that a damaging confession for me to make? Well, how would you do it? …How would you design a search algorithm clever enough to screen out bad results like “his book Bin Laden” without also screening out good results like “His Book Bin excursion was a success…” or “The Book Bin, laden with rare finds…” Bear in mind, this algorithm also has to work for every other business name in the English-speaking world.
If you’ve got a solution, you should definitely get in touch with us, because we’re looking for smart people to help us improve the relevancy of our searches. I thought this story was worth sharing because it illustrates just how difficult it is to achieve perfect accuracy. StepRep is getting stronger and stronger, but (unlike the Bilderberg Group) it’s not all-powerful…yet…




