Is there even a need to teach search?
A goal of Google search is to be so simple, so obvious, so effective, that the searcher never needs look for help. In principle, search just ought to work the way you imagine it will. The assumption is that Google searchers will search for things we can understand… and that the Google system will just “do what’s right” automatically.
If this were true, then there wouldn’t be any need for courses to teach people how to search. Yet, when you look around, it turns out that there are a fairly large number of courses that are being taught. Classes like “Become a super Google searcher” promise to teach you the tricks you need to search effectively.
How many courses are there? In truth, I don’t know. But I started looking, and once I started looking, I couldn’t see the end of the list. Essentially every library I checked in California (something like >95% hit rate) runs a class to teach Google search (or as you’ll quickly learn, it’s usually called “Internet search,” so as to be fair and equitable for all search engines). There are entire companies that do nothing but teach people how to use Google search in the enterprise.
As we've found in our fields studies, when you actually go talk with people doing search in their home, roughly 90% of the people we interview say some version of “I wish I could search better… I know there are things out there that I just can’t find…”
So the need is perceived, there seems to be a market, and from direct observation I can tell you that while most people CAN search, by the same token, most people don’t do a good job because nobody ever really showed them how.
What needs to be taught? Here’s my short list of things that need teaching; a list of the three most important skills and pieces of knowledge that searchers need if they’re going to be great at search.
1. Search Strategies
I teach a lot of classes on “how to search.” In the past year, I taught 14 classes of students (that’s around 700 students), and looked over the shoulder, helped out and watched many of those students. Some were advanced, some were the rankest of beginners—but the overriding thing they had in common is they knew how to add terms to a search box and how to get the results... but if the answer wasn’t there, they had little strategic knowledge to fall back on when things went wrong.
Luckily, many of the questions searchers bring to Google are pretty simple. Many are simple navigation queries (which Google does spectacularly well), many are searches for simple facts (“Who won the battle at Little Big Horn?”). But once the question gets to be slightly more complex, you need some fairly deep knowledge about what to do next. Once the searcher gets beyond the basics—“What was the Army’s goal in having General Custer at Little Big Horn?”—then the searcher needs a search strategy.
In its simplest form, strategic knowledge can be knowing how to try out multiple versions of a search, testing out various combinations of terms to explore variations on a query. But you’d be surprised how many people create a query, do a one or two modifications of the original query, and then give up. This is a difficult behavior to see in logs, but one that I see all the time when helping people at the library.
What’s worse, though, is that most searchers don’t know what to do when the results they’re seeing are all off-topic with respect to the question they’re pursuing. When the results coming back don’t look right—you need to have something to do, a way of making progress. That’s what a strategy does for you.
One basic skill that self-organized learners have is the ability to triangulate; that is, to look for multiple sources of information, spot potential biases among the sources and ferret out independent means of validation. We know that Google searchers frequently get mistaken answers to even simple questions. (In one recent study, 20% of the answers to a simple, factual question were blatantly incorrect. Why does this happen?)
How can we help make our users competent searchers in an age of many competing and conflicting information resources? Once our users begin to step beyond the simple question and navigational query, they need to know about other ways to approach the search problem.
We need to teach strategies for doing searches that are more sophisticated than simple fact-finding and navigation searches.
2. Credibility and authoritativeness assessment
In a recent email a friend wrote: “… During my first trip to Kenya I got to observe a journalist training program (organized by Internews) and was shocked by how journalists used the web. The journalists reported back with a brief report on the pros/cons of second generation HIV drugs. Most of the information in their reports, which they got from the web, was incorrect (and frankly, a bit scary). They just looked at the first couple of links and took the information at face value. These journalists worked for small radio stations across the country and their programs had quite a reach in rural areas where misinformation can be very damaging…”
I’ll make a major wager here: I’ll bet a dollar that this is a very common behavior, and not just among African journalists. I’ll go further and say something that’s not widely believed: Most people actually are not very good at searching. The situation of the African web just makes the problem more apparent.
It’s tempting and easy to think that you’re a good searcher: Can’t you just Google something and get the something you seek back as an answer?
Yes, you can… but the key thing is… it’s just something back. How often do you, when you’re searching, check that what you got is really correct? How do you assess the credibility of a web result?
Most people say they just look at the results and—to get right down to it—make an intuitive judgment about it. Does the page look professional? Does it seem consistent within the visual style? And, most of all, does it line up with things I already know?
This last factor is important because confirmation bias is a well-known source of problems. If you’re expecting a particular outcome, or even a particular KIND of outcomes, then we know you’re more likely to see that in the results. Your eye even has the ability to spot confirming data on the page in a pre-cognitive scan.
Oops. That can be very bad. Take a look at the famous “tree octopus” website. (Don’t know about it? Search for [ tree octopus ] and look at the first result.) It looks very polished and professional. More than one teacher has been taken in by its gloss and surface-level coherence. If you’re an adult who’s grown up knowing some basics about marine fauna, it usually only takes a minute before you realize that it’s totally and utterly bogus.
But suppose you didn’t grow up with marine biology textbooks in your public library. Then, the tree octopus site looks really good—very credible.
And, what’s even worse, in cultures without a long tradition of visual styles that convey authoritativeness (e.g., newspapers in east Africa), implicit web authoritativeness is rampant.
So what DO you do to assess the credibility of a web page? Turns out there’s no simple answer. www.martinlutherking.org seems credible, until you look at what other sites link to it, or use a WHOIS lookup and discover that the site is funded by a white supremacist group. But if you don’t know how to do these kinds of fairly sophisticated background checks, how would you ever know?
To an extent, we’re all third world people for some domain that lies outside of our immediate knowledge. If you’re a Silicon Valley person, how would you gauge the credibility of web information that tells you about low-cost fertilizer treatments for your crop of Kenyan millet?
We need to teach people how to look at information they find on the web, via Google, and make assessments of credibility.
3. Teach what kinds of content are available
A crucial factor for being an expert searcher is knowing what kinds of things you can find. When I teach classes at public libraries in Silicon Valley, I’m always surprised at how few different resources people recognize or how many different kinds of resources exist.
This problem extends beyond knowing what Google has indexed to include knowing what's possible to find with web search. The equally terrible implication is that some people honestly believe that if they can’t find information on Google, then it doesn’t exist, and they give up in their search.
In short, people who don’t have a good idea about what’s possible to find don’t ever consider looking for it—they don’t ask for things they don’t know about.
As a simple example, how many people know that Google Earth has archival aerial imagery of many parts of the United States. That knowledge is critical to being able to do many kinds of comparisons of land use patterns, watershed analysis and deforestation over time. Without that kind of knowledge—without knowing that archival maps are available—the searcher is stuck, and unable to progress in their search.
Just as important as knowing what kind of information is available, skilled searchers need to know out to information can be mapped from one form to another. The simplest example is just a dictionary: A dictionary indexes from word to definitions. There are many different kinds of information maps. A reverse dictionary maps from words used in the definition to individual terms. (It’s incredibly handy for looking up terms you don’t know about. Try finding a metal that’s reddish in hue, could be copper, but I was thinking of bismuth. This is a trivial reverse dictionary lookup. Handy when you need a synonym...)
Knowing at least what’s possible, knowing the full range of human expression and information availability opens your mind to asking the question. What you know you can search for is a powerful conditioner of what you do.
An important job that we need to take on, I believe, is to teach people about what kinds of knowledge are possible. Knowing about something isn’t the same as knowing that thing in depth; but it does open the possibility of searching on that thing in the future.
We need to teach the range and scope of what information is available in order to let people know what they can find (and understand).
Summary
Search engines are remarkable tools. Google is simple, powerful and massively transformational. It has changed the way everyone thinks about information.
Yet even simple and powerful tools need knowledge to be used well. Even Apple, famously known for their emphasis on ease-of-use, has to teach their users how to use. Witness the Genius Bars at the local Apple stores, the Apple-dedicated magazines, or the plethora of online forums that people use to exchange tips and ideas. That’s a lot of support for something that is claimed to be easy. The reality is that computers, even Apple computers, are incredibly sophisticated devices. Once a user gets beyond the mere basics, they need help.
I want to argue that Google has made many right decisions about ease-of-use, but that there is still a long way to go. Google CAN do more than provide fast answers to navigational questions, fact lookups and pointers to Wikipedia articles. Google can, and should, teach people how to use Google to see more deeply into a topic, to come to a greater understanding of anything they want to understand.
I want to argue that Google should teach people how to think about research, rather than just search. There’s an important shift in thinking here. The difference is one of scope and scale. A research task is what people do to solve sophisticated problems, while a search is what they do to solve sub-pieces of the larger problem. Google supports search very well. But we don’t help our users with their larger research task.
In the end, from a business perspective, better searchers will do fewer searches-per-task than unskilled searchers, their time-to-result will go down. That’s good—we want searchers to have short, fast, effective experiences (we know that brevity of search task is highly correlated with happy users). BUT we also know they will also take on more search tasks overall. In every field study and class I’ve run, follow-up surveys and history analyses have shown that more skilled searchers search more. They grow in their capability and knowledge.
So what should Google do? Improvements in search quality (such as spelling corrections, suggestions, related searches, and so on) will continue to make everyone better as the technology advances. But we can do more to make Google users significantly better in addition.
1. Build learning into Google search. This means making tips and help content be targeted and precise; making better use of the Google Help Center; let users know when they have opportunities for improving their search at both tactical and strategic levels.
2. Create a search curriculum that can be used widely. We’re doing this already (see: 9 Search Lesson Plans). Google clearly can’t teach everyone how to search, but there are literally thousands of librarians and teachers willing and ready to teach… if only they had some authoritative and reputable content to work with. We also are busy creating some engaging short videos that can be used to teach individual concepts.
3. Continue Google teaching and outreach to the community through classes. Allow every Google office to offer search (and other product) classes. One of the great values of teaching person-to-person is that we come to learn what the key issues and drivers are for our users. Clearly, Google outreach won’t teach more than a tiny fraction of our user population—but it can be an important factor in reaching key players, the teachers, parents and librarians that will carry our message to the rest of the world.
This is a great aspiration for Google to have: We should be in the business of showing people how to search and research in order to deepen their understanding of the world. We want to make the world’s information available. We’ve done a decent job at getting a good fraction of it online. Now, let’s teach people what they can do with all that.
They can do more than look up the latest celebrity news, they can transform their lives.
