Your grandkids are going to laugh at the notion of "searching the web"

In the next 5 - 10 years our good friend and faithful digital companion, the search bar, is going to ride off into the sunset.

Googling has forever changed how we find answers to almost anything we could ever want to know, because the answer is most likely there buried in the first 10 links brought up in response to anything you type into the search bar.

When you think about it though, having to trawl through those 10 links is actually not a great way to answer a question. If you were to ask your mechanic what was making that noise in your engine, you want the answer, not rambling descriptions and discussions on engine design and maintenance. But that is exactly what you get if you were to ask Google the same question: pages of forum posts when you only want the last one that actually has the answer; a blog post where that engine noise is coincidentally mentioned in the comments; or a 10 minute YouTube video where the information you want is 9 minutes in.

That is all about to change. In the not too distant future you will type your query into a question bar, and the result will be a plain English answer. No more "this search returned 10000000000000 results"; no more clicking through web pages to find the information you are looking for; and, if Google isn't careful, potentially no more Googling.

We have seen a glimpse of this move from searching to asking with Wolfram Alpha. You enter some search criteria, Alpha identifies real world entities or concepts in the search query, and then displays a bunch of cross referenced information. It's a great tool if you are looking for something that is easily identifiable, like the manufacturer of your car, and possibly even your car's model. But Alpha is not going to tell you what that sound in your engine could be.

That's not to say that Wolfram is giving up. He has made some vague reference to a new knowledge based programming language:
We call it the Wolfram Language because it is a language. But it’s a new and different kind of language. It’s a general-purpose knowledge-based language. That covers all forms of computing, in a new way. 
Apple has made some bold moves away from searching with Siri. Granted, Siri is often just a speech to text converter for plain old web searches, but it is not hard to imagine Siri connecting the dots between the questions you asked about "that noise in my engine" and post number 13 in a forum conversion that mentions a fan belt with the same ease with which she connects you request to set an alarm with the settings in your iPhone alarm clock app.

Microsoft is also making a move towards asking questions:
"Our vision of search is 'just ask'. Search is the ultimate human interface. It should be able to cope with any input."
And then there is IBM, which bested some of the brightest minds ever to grace the game show Jeopardy with their system Watson. Watson is probably the best public example of a system that can answer general questions with plain English responses, and that power is going to be available to developers:
Developers who want to incorporate Watson’s ability to understand natural language and provide answers need only have their applications make a REST API call to IBM’s new Watson Developers Cloud. “It doesn’t require that you understand anything about machine learning other than the need to provide training data,” Rob High, IBM’s CTO for Watson, said in a recent interview about the new platform.
Google has been fairly quiet about this post web search world, probably because it is of no benefit to them to suggest that searching is going anywhere, but their acquisitions of data sources like Freebase, the sidebar that displays relevant information about any entities in your search query and Google Now do indicate that they are looking to move beyond just plain old web searches too.

So when my grandkids ask me what the internet was like when I was their age, I will tell them of the great information hunts I used to go on, carefully crafting my search queries and then staking the answer through pages of unrelated and often unhelpful information. And they will laugh.


Popular posts from this blog

MinHash for dummies

Authenticating via Kerberos with Keycloak and Windows 2008 Active Directory

Fixing OpenVPN "Authenticate/Decrypt packet error: cipher final failed"