Bing Engineers Serendipity, Not Just Irrelevant Results

May 5, 2018

It is Saturday. Innovation in search never rests. I read “Bing: Search Engines Have a Responsibility to Get People Out of Their Bubbles.” The headline is one guaranteed to give me a headache.

My view is that when I use a search system I expect, want, and need the system to:

  1. Process my keyword query, accept Boolean logic (AND, OR, and NOT arguments), and generate a list of results that optimize relevance.
  2. If I need more results, synonyms, Endeca-like “facets”, I want a button or a menu option that allows me to specify what I think I need to get the information I seek.
  3. I want to have ads, sponsored content, and SEO skewed content flagged in a color which is easily visible and put within a ruled “box.”
  4. I want to know [a] the date at which the displayed result was indexed, [b] the date assigned by whoever wrote the item to the specific article, and [c] an explicit link to a cache in the event the page indexed has been removed or is otherwise unavailable.

I have other requirements for a commercial search system; for example, Diffeo’s or Recorded Future’s approach. But these are specialized and inappropriate for a Bing style Web index.

The Bing approach, according to the write up:

Bing has launched a new feature called Intelligent Answers. When you enter a question with several valid answers, the search engine summarizes them all in a carousel to give a balanced overview.

I don’t want answers. I want a list of relevant locations which may contain the information I seek. For example, I needed to identify the term for a penance device and access images of these gizmos. Bing, Google, Yandex, and even lesser known systems like and failed.

The systems returned everything from a church calendar to a correction of penance to pennant. I did not want baseball information.

Now Bing is going to identify from my query my “question” and provide a range of answers. I don’t want this to happen. If I search LOCA, I want information about a loss of coolant accident, not this:

loca the song

I am happy to add a field code for power, nuclear if such a feature were supported by Bing. I would also add key words to get something close to my term.

The complete and utter silliness of Bing results exists right now. The company which has managed minimal progress in search now expects me to believe that its “smart software” can provide answers.

The write up states:

“Take a simple query like ‘Is coffee good for you?’” said Ribas. “There are plenty of reputable sources that tell you that there are good reasons for drinking coffee, but there are also some very reputable ones that say the opposite. Deep learning allows us to project multiple queries in the passages to what we call the semantic space and find the matches.

Based on my limited experience with whizzy 2018 search technology, I am not sure if Bing’s innovation will be helpful to me. When “semantic space” is concerned, the systems with which I am familiar, provide a number of other tools and functions to ensure relevance and accuracy.

Even with those tools, including state of the art systems from developers from Madrid to San Carlos, the user has to think, analyze, and run additional queries. Phone calls, interviews, and even visits to libraries are often required to obtain helpful information.

Bing promises “intelligent answers.”

Sounds like MBA infused marketing with a few notes added from engineers with better things to do than explain exactly what a content processing component can do with 80 percent accuracy.

Time out. The referee wants the coach to get the MBA marketers off the field for intellectual fantasizing. This is the same outfit which owns Fast Search & Transfer, created the racist chatbot, and missed the mobile phone business by a country mile. Why not ask Bing a question like, “How did these missteps occur?” Perhaps Watson would be able to take a crack at “intelligent answers”?

Stephen E Arnold, May 5, 2018


Comments are closed.

  • Archives

  • Recent Posts

  • Meta