24 03 2014
Will mobile apps redefine the Web?
The growth of the Internet has historically been based on Web sites, search engines that help locate sites of interest, and Web browsers to display the search engine and Web site content. The search engines that drive all this are free, supported (lucratively) by advertising revenues. The result is synergistic: Search makes content findable, encouraging the generation of more content.
This model is being challenged by the explosion in the use of mobile devices to access the Web. Of course, an individual can use the Web browser on a mobile device just as they would on a PC. But, increasingly, access to information and services on the Web is through apps, software that works around browsers and Web search engines to give direct answers or directly launch specific Web sites without going through a classical Web search. For example, if I ask Siri, “What’s the best Chinese restaurant nearby,” it will provide a list and then automatically display the restaurant with the best Yelp rating and ask if I “want to try that one.”
Eric Schmidt, Executive Chairman of Google, testified before a Senate Antitrust committee in September 2011: “Apple has launched an entirely new approach to search technology with Siri.” Google has responded by launching its own personal assistant (natural-language voice or text search initiated in some cases by saying, “OK Google”) without calling it a personal assistant. Samsung has the S-Voice personal assistant on Galaxy phones (based on language technology from Nuance Communications), and Nuance has its own Dragon Mobile Assistant, a free download. Microsoft is rumored to be developing a personal assistant named Cortana (based on a video-game character). Supposed leaked screenshots of an upcoming Blackberry OS release show an icon named “Intelligent Assistant.” Obviously, something fundamental is going on.
General personal assistants like these try to be your preferred entry to the Web, and for good economic reasons. Schmidt also testified in 2011 that Siri was a threat to its advertising revenues. If the personal assistant substitutes for search engines as a starting point for most individuals, those apps can eventually generate most of the ad revenues driven by the Web.
And there is no reason the same personal assistant can’t be available on your PC as well as across your mobile devices. The small size of mobile devices and the lack of a full keyboard encourage the use of speech to interact with these devices, and the difficulty of navigating through many screens on the small devices encourages the use of natural language (even when entered as text) to accomplish a task in fewer steps. There is the long-term possibility of the app model becoming the preferred approach at all entry points to the Web.
And, just as specialized web sites developed, specialized personal assistant apps are beginning to appear. Such a trend upsets the Web paradigm further: We will be able to navigate between apps by returning directly to a general personal assistant app like Siri or Google, much as we return to the Web browser today.
Skeptics will say the speech and natural language understanding technology isn’t ready yet. And, indeed, there is much improvement possible. But the requirement isn’t that a computer fully imitate human skills (as is often the claim), but that it handle things that a human doesn’t do well. A hammer can’t do as many things as a fist, but it certainly is better at driving nails. Computers are also tools, and similarly better at many things than humans, while not-so-good at other things. When we converse with a computer, it is usually to have it do the things it can do better than a human alone.
For example, I may say ask for a good nearby Chinese restaurant as discussed earlier. I expect the virtual assistant to understand that it should use a search engine, GPS location, mapping system, knowledge of my current location, and a restaurant evaluation service to fulfill my request. If it gives me a list of options, I might touch one and say, “Make a reservation for two at 8 tonight.” It would confirm if it is able to do so. This example is possible today.
There will be continuing battles to be the virtual assistant that an individual turns to. With Apple’s Siri, one can inquire by voice, without text entry of requests being an option, a limitation that makes Siri unusable in situations where speaking isn’t polite, privacy is an issue, or there is too much background noise. Google’s “OK Google” allows speaking a request like Siri, but a request can also be typed into a Google search box in natural language, making natural language an option in all cases. Google for some reason doesn’t like to characterize their natural-language feature as a “personal assistant,” but it serves that purpose. Processing a natural language request requires more computer power in the network than a classical web search, as well as continuing development of the core technology, so Google may simply be cautious about encouraging too rapid a switch to a natural-language approach.
Given the history of speech recognition and natural language understanding taking much longer to approach mainstream use than expected, it’s dangerous to make too optimistic a prediction. But there is a problem with current digital devices and applications that natural language can help solve—digital overload, too much of a good thing. The user interface that worked when the Web was smaller and devices had fewer features and apps, is getting over-burdened. That is a hurdle to individuals adopting new devices, apps, and services and will drive the development and use of natural-language apps. Beyond serving a need, advances in the personal assistant model will be driven by competition: Competition between providers of general assistants and competition between companies to provide effective specialized assistants to help with their products and services.
I suspect this will be one of those trends most obvious in hindsight. Companies with foresight can profit.
– William Meisel