In deciding to read an article with such a title, did you have in mind how Google might be falling down on the job? When I started to write this article, I had other things in mind. However applying the fresh blogging approach I have described recently, one news item provides the perfect lead in to the topic.
CNN reports that Google apologizes for results of 'Michelle Obama' image search. Over the past two weeks, some strange things have been happening.
A user alerted Google to an extremely offensive altered photo of Michelle Obama via an online help forum two weeks ago. Google at first banned the site with the photo, but when the picture reappeared elsewhere, Google let it stand in the Image Search results. A Google ad apologizing for offensive content sometimes appeared with the photo.
Google then seemed to be defending its actions when it eventually provided an official comment.
The California-based company explained that search results rely on computer algorithms that take into account thousands of factors. "The beliefs and preferences of those who work at Google, as well as the opinions of the general public, do not determine or impact our search results," it said.
The company said that the integrity of its search results is extremely important. "Accordingly, we do not remove a page from our search results simply because its content is unpopular or because we receive complaints concerning it."
This is not only an embarrassing incident for Google, but it does point to the deficiencies in the Google approach. If you wish to explore why those deficiencies occur then a recent post, Hold Me Tight And Engaging Websites, has the answer. Here we will give some further explanation of why Google may not get the answers right.
What Google Does Well
When it comes to search, Google does a great many things right. Although it works on only a few keywords that the searcher may give, it tries to make as much sense of those keywords as it can. In the early days it was just a question of parsing key words so that for example a singular word would trigger a display of the corresponding plural words as well.
Since then, there are much more elaborate semantic analyses that are done of the meanings behind the key words. Anyone who has explored key word searches on Google Insights can see this demonstrated over and over again.
Very recently Google has undertaken a major upgrading of its search methodology which was labeled the Caffeine update and this has produced a striking increase in the speed at which results are provided.
It also attempts to identify the searcher's needs from their search history and provides more personalized results that hopefully will appear more relevant to the searcher. What has been achieved is all well and good but it has some limitations which we will now explore.
Is That All There Is
Google's mission is to catalog all the information that exists and help human searchers find what they are looking for within that immense body of knowledge. Their aim is to to deliver information which human beings will find relevant to their search needs. However they use logical, computer-based methods to find the information and then to index it. You might imagine that this would work well for information where the knowledge content is entirely digital. In such cases, the right choice can be determined wholly by logic. However human beings use more than logic when they view the world.
Malcolm Gladwell has pointed out in his book Blink that visitors to a web page can sometimes switch away from that web page in the blink of an eye. Somehow in microseconds they realize that the web page is not for them. Perhaps they are using other abilities than the logical part of the brain in making such an assessment. Reaction occurs in a flash. It seems to be instinctive and not triggered by any logical thought process. Perhaps it is some other part of our central nervous system that gets involved.
This would certainly seem to be a possibility if you follow the thinking of Fritz Glaus and Stephen Goldberg, the cofounders of the concept of Three Brain Synergy.
They suggest that we each have in a sense three brains. These are labeled ”head brain”, ”heart brain”, and ”gut brain” by the neuroscientists. The head brain is concerned with logical reasoning, the heart brain deals with emotions and the gut brain deals with those instinctive reflex actions generated by the limbic area of the brain.
Humans will react in any given situation based on how these interacting parts of their brain function are interacting. Without conscious thought, it may be that our gut brain has set us up to react in a certain way to inputs that our head brain or our heart brain are processing. The gut brain is where the fight or flight reaction is triggered when we are suddenly shocked by a possible threat or danger. The body chemistry reacts immediately in flooding the body with adrenaline to allow extremely fast reactions to the threat. This is not something we control. It is instinctive. We may not be aware that it is happening.
An interesting confirmation of the importance of the got brain is provided in the book, Hold Me Tight, written by Sue Johnson. She is a professor at the University of Ottawa and a long-term developer of very successful techniques for couples therapy. Couples are often at loggerheads or even almost in a war very often through attitudes triggered by the gut brain. Thereafter the logical and the emotional brains are all working within barriers and constraints fixed by that gut brain.
This instinctive behavior could well operate in a whole host of situations, and that includes how we assess particular web pages. If our reaction is a resultant of the interaction of our three brains, how can Google mimic that with processes that use only single brain logic. Their algorithms are using only a fraction of the data that the human being is assessing.
What Google Must Add
If the Google search process is only partially handling what is important to human beings, how can Google redress this shortcoming.
Even if more people were willing to provide data on how they assess the choices served up by Google, it seems unlikely that most searchers would wish to get involved and put any credence on such data. In these cases of course, actions speak louder than words.
Google is aware for some websites that use Google Analytics just how visitors to the websites are reacting to the available choices. That does suggest that Google could include a more broad scale evaluation of the website by using this analytical data. The computational challenges would be significant and yet the data would be available only for a fraction of websites. This is hardly a viable solution to ensure that key word search suggestions will be more relevant to the human searchers.
The same reservations can be made about incorporating social media information such as the voting or ranking that could be captured via Facebook, or Twitter. Those expressing opinions are only a fraction of the total population and indeed they may have axes to grind in expressing opinions one way or the other.
Can we ever expect that the Google search processes can be strengthened so as to draw on the total information that human beings use as they apply the power of their three brains. Can Google ensure for example that its Image Search will avoid displaying images that so many searchers may find highly irrelevant. It seems most unlikely that Google can avoid such unfortunate incidents as we have all seen in this case with Michelle Obama.
Search algorithms work best on digital data and some problems will not be well handled by the Google technology. Perhaps Google must accept that what they are doing now is about as good as they can do.
4 thoughts on “Google Search Does A Half-baked Job”
Good article Barry. I find it amazing how much goes into making sure I get relevant sites when I am searching for something on Google. I think there may be a plateau at some point where Google just cannot get any better.
Roy Paeth – Chicago First Time Buyer
What alternative do you want to suggest instead of Google?
That’s an excellent question, Ligurien Guy, and perhaps Roy Paeth has suggested the answer.
Google will handle purely logical stuff well by algorithmic methods. That may be say 40% of what the Internet population is interested in. Even there they will have problems since spam article writers can fool the algorithms.
For other websites, where the other brains (heart, gut) are involved, then you probably need to involve social media in some way and get the people involved. If Google wanted to help, it would certainly know from its analytics that visitors act “irrationally” on certain pages as per what their algorithms are telling them. In those cases, Google might indicate by a quality score for its algorithm that their evaluation is weak. However that’s probably too much truth for Google to reveal.
.-= Barry Welford recently posted: Stop Sending Junk E-Mail =-.
Comments are closed.