Boolean Search Is Dead

Google search strings are often called Boolean Strings. But do you think Google search is Boolean? It is not. Neither is LinkedIn’s, despite what their help says. Both platforms apply semantic algorithms, trying to guess the searcher’s intent; one – successfully, the other – poorly, introducing bugs while at it. Both platforms break the formal Boolean rules.

You can see that the Boolean logic is broken with these simple examples, which “should” return no results, yet they do because of the interpretation:

• restaurant manager -manager (Google)
• restaurant manager NOT manager (LinkedIn)

You may want to rely on Google’s interpretation in cases like researching target companies or associations (etc.). That is a “question to Google,” like top retail companies in china or common hispanic last names, looking for one or a few “right” answers – like most people do. That is what the verb “google” implies. Where you need to apply control to eliminate some false positives, you can put single keywords in quotes, and Google will not vary them.

However, relying on one Google search is a bad idea if you are after as many matching results as possible.

1, You want to modify your search to let Google know your search is “restrictive,” which may produce many more results.

2. Modifying the search without changing its Boolean logic leads to different results! These changes will get different results and their numbers:

• Putting some words in quotes (see 1.)
• Changing the word order (ask Nicolas Darcis)
• Repeating words.

People – Researchers, Sourcers – like us – who want to get lots of results, for example, from an X-Ray,  are not on Google Developers’ radar since we are in the minority. From experiments and a nod from Google’s team, we should be playing “non-Boolean” games and combining results from seemingly equivalent searches. Some automation may come in handy.

Why are we still seeing mile-long OR strings on Google in posts? For a while now, using ORs on Google has been reducing search results. Compare searching for each term in turn and combining results with a search with ORs for any list of words, and you will see. Say NO or ORs.

LinkedIn is another matter.

LinkedIn is a different – and sad – story. Its premium search is buggy – it won’t find some members through keyword searches for words in their Headlines, About, and Job Descriptions. We have not been able to decipher what it does – it just seems random.

LinkedIn Job Search is broken in strange ways.

What are LinkedIn Developers doing? @LinkedIn, we can help you find good ones. Perhaps QA Engineers as well.

LinkedIn Recruiter search is unintuitive, but you can use it well.

When sourcing, you also need to be aware of how restrictive LinkedIn/Recruiter filters can be. When you use standard filters, you are down to searching among 18% of LinkedIn members.

X-Ray, it turns out, is an excellent alternative to Recruiter.

In practice, “broken Boolean” means that we should be smart and combine controlled Boolean and open-ended semantic searches to reach our goals.

I am preparing a webinar on the topic outlining what exactly happens with the Boolean rules and how to take advantage of our algorithm understanding; watch our announcements. In the meantime, join the popular, updated, Find Anyone’s Contact Info on November 9th.

Explore our Sourcing Training Library.

My friend, a philosopher and community leader Hung Lee and me will have an informal chat on the “Boolean Is Dead” topic on November 11th. Join us!