Bird-brained computers stand in the way of the voice search revolution | John Naughton

Smart speakers such as Amazon’s Echo won’t answer our search queries until they acquire an understanding of natural language

Type “What is the future of search?” into Google and in 0.47 seconds the search engine replies with a list of sites asking the same question, together with a note that it had found about 2,110,000,000 other results. Ponder that number for a moment, for it reflects the scale of the information explosion that was triggered by Tim Berners-Lee’s invention of the web in 1989-90. Back then there were no search engines because there was no need for them: there were very few websites in those early days.

Google turned 20 recently and the anniversary prompted a small wave of reflections by those who (like this columnist) remember a world BG (before Google), when information was much harder to find. The nicest one I found was a blog post by Ralph Leighton, who was a friend of Richard Feynman, the late, great theoretical physicist.

The story starts in 1977 when Feynman mischievously asked his friend “whatever happened to Tannu Tuva?” Leighton denied that any such country had ever existed. But he was wrong: Feynman, who had collected stamps as a boy, remembered Tuva as “a purple splotch on the map just outside of outer Mongolia”, which in the 1930s had issued dozens of beautiful triangular and diamond-shaped stamps. Then the country mysteriously disappeared: it was absorbed into the USSR during the second world war.

Leighton records how he and Feynman set out to learn everything they could about Tuva. They used all the tools and resources available to them in the 1970s – card catalogues, microfilm reels, cross-library listings, books – in the hope of finding useful nuggets of information. “Each nugget, rare and unexpected, delivered a small delight of discovery and kept our quest alive,” wrote Leighton. It took months, for example, to find a single grainy black-and-white photograph of Tuva.

And now? A Google search for “Tannu Tuva stamps” yields about 29,000 results.

So has Google cracked the search problem? Only if you think that the solution is finding web pages that sort of match a query. Which means that, good though it is, Google has only scratched the surface of its vaunted mission to “organise the world’s information and make it universally accessible and useful”. The search problem, in that sense, is only 5% solved. Which means that the question obsessing everyone working in the area is: where next?

The standard dinner party answer at the moment is “voice”. “By 2020,” says Comscore, a marketing consultancy, “50% of all searches will be carried out via voice.” This is probably what lies behind the current feeding frenzy for networked smart speakers such as Amazon’s Echo, Google’s Home and Apple’s HomePod. A few moments’ interaction with any of these devices, though, is enough to make Comscore’s extravagant prediction seem absurd. Apart from concerns about privacy and security, there’s already a burgeoning genre of the daft misunderstandings enabled by these gizmos. (Example: “Alexa, can you turn the porch lights on please?” Alexa: “You don’t have a Porsche, so I’ve turned on the Dacia Duster lights instead.”) And, of course, there’s the case of the “shopping parrot” – a bird that ordered a consignment of gift boxes from Amazon via its owner’s Echo.

Many of these absurdities stem from the fact that computers’ understanding of natural language is still inadequate. But even when it gets better, the central problem with voice interfaces for search – that they lead users into cognitive culs-de-sac – will remain. They respond to a query with a single, definitive result. In the words of one analyst: “It might be possible to follow up that query with subsequent searches, or to carry out an action (eg ordering pizza, hearing a recipe, receiving directions), but otherwise, the voice journey stops there. You can’t browse the web using your Amazon Echo.”

Yep. That’s why the future of search has to be about cognition rather than queries; it’s about concepts rather than words. As one Google search engineer says, it will require three breakthroughs: a shift “from answers to [intellectual] journeys”; providing relevant information related to a user’s interests, even when they don’t have a specific query in mind; and moving from text to visual ways of providing information. Of these, the first is really, really difficult and the second is creepy (Google knowing what you want even before you know you wanted it); only the third is relatively easy.

So search is only 5% solved. Just to check that, I asked Apple’s Siri “Why is internet search so hopeless?” She provided three links: a page from Psychology Today headed “Sad, worthless, hopeless?; “21 tiny ways to stop feeling hopeless”; and “A letter to those who feel hopeless about life – get busy living”. Back to the drawing board.

What I’m reading

Power to the people
One small step for the web... Tim Berners-Lee is working on a project to wrest control of our personal data from the silos of the tech giants. Quixotic? People once thought the same about the web.

No friends of Facebook
Facebook executive David Marcus posted an emotional critique of the WhatsApp and Instagram founders who have fled the Zuckerberg nest. He claims FB is “just about people”. Reminds one of HL Mencken’s crack about the impossibility of explaining something to a person whose salary depends on him not understanding it.

Like my new ‘home’?
Instagram “influencers” can rent a cool SoHo pad to show them in their best light. Beyond parody.


John Naughton

The GuardianTramp

Related Content

Article image
Alexa, how did Amazon’s wrong call on voice assistants tee up a $10bn loss? | John Naughton
The tech giant’s flawed business model for its popular smart devices has cost the company a fortune and thousands of jobs

John Naughton

26, Nov, 2022 @4:00 PM

Article image
Amazon’s Echo seems great, but what does it hear? | John Naughton
The voice-controlled home assistant works well, but on the other hand it is a networked listening device, with wider implications

John Naughton

15, Jan, 2017 @7:00 AM

Article image
Five easy ways to make your home smarter
Turn up the heat, dim the lights and cue the music, all with the touch of a button or a voice command. Here’s the latest kit to connect your world

Samuel Gibbs

05, Feb, 2017 @9:00 AM

Article image
Computers have learned to make us jump through hoops | John Naughton
Machines are supposed to be tools that serve human ends, but the relationship is slowly shifting - and not in our favour

John Naughton

18, Nov, 2018 @7:00 AM

Article image
Computers can write their own code. So are programmers now obsolete? | John Naughton
A London-based AI company recently revealed its new engine AlphaCode can already match many human developers

John Naughton

12, Feb, 2022 @4:00 PM

Article image
Computers need to make a quantum leap before they can crack encrypted messages | John Naughton
The latest claims by scientists that they are able to break the most common digital encryption system are far-fetched

John Naughton

14, Jan, 2023 @4:00 PM

Article image
Is Google’s domination of the internet finally over? Search me… | John Naughton
Questions are being asked about the web giant’s waning powers, but it may just be a storm in an online teacup

John Naughton

19, Mar, 2022 @4:00 PM

Article image
Are blockchain-based DAOs really a utopian revolution in the making? | John Naughton
Decentralised Autonomous Organisations are member-owned communities promising transparency, democracy and security – but the financial reality doesn’t always stack up

John Naughton

12, Mar, 2022 @4:00 PM

Article image
Can democracies stand up to Facebook? Ireland may have the answer | John Naughton
Liberal democracies struggle to tackle the might of the tech giants when it comes to protecting users’ data

John Naughton

26, Sep, 2020 @3:00 PM

Article image
A self-driving revolution? We’re barely out of second gear | John Naughton
The Department for Transport’s breezy predictions about driverless cars transforming Britain into a ‘global science superpower’ are both premature and optimistic

John Naughton

23, Apr, 2022 @3:00 PM