A Glimpse Into The Future Of Search With Futurist Ramez Naam
Ramez is a futurist, a best-selling author (check out his science fiction novels Nexus and Crux and his newest nonfiction book, The Infinite Resource: The Power of Ideas on a Finite Planet), a startup founder, and a former search guy at Microsoft/Bing.
Search, and computing in general, is going to evolve beyond our wildest dreams. Movies like Minority Report and Her are entertaining representations of the future of computing, but they only scratch the surface.
It really will be a brave new world beyond our imaginings. That’s where futurists like Ramez come in.
They help us paint a more complete and accurate picture. Because we as humans tend to visualize the future linearly, we don’t “see” the curve of the planet as we gaze into the horizon. We don’t see where exponential technologies and the Law of Accelerating Returns is taking us.
I sat down with Ramez over Skype to discuss some recent and upcoming technological and engineering breakthroughs and their impact on our lives as SEOs. This is only an excerpt of a longer interview, specifically just some of the more SEO-related aspects of our interview.
Stephan: Where do you think we are heading in terms of search — and search technology — overall?
Ramez: Three big things:
1. Search is going to be dominated by mobile. Desktop is still there, but people are spending more and more time on mobile. Think about if your content works on a mobile device — because even if it is optimized really well and gets to the top of the search rankings as a result, you’re kind of screwed if the user can’t interact with it on a mobile device.
2. I think that we’re going to keep working on the types of inputs that mimic natural language. But that is going to be slow to develop. Whereas the mobile development is just rapid-rapid-rapid, the progress on speech will be bit-by-bit and the progress and on language, bit-by-bit.
Take the example of Siri. Siri is an awesome demo of the future. How often do you actually see somebody use it? (Or, if you have an iPhone, how often do you use it?) Not that often, I would guess. The reality is that the error rate is too high. We’re not okay with 5% error. Maybe 1%. But absolutely not 5%. We cannot live with that.
Additionally, we’re in these environments where other people are talking. We don’t want to impose what we are doing on them. Like, “Siri, find out what tomorrow’s weather is.” I don’t want to say that in front of the person in the cubicle next to me or the person on the train sitting next to me. Text is a nice semi-private, less obtrusive input method.
3. Something that I see people doing a ton of (and Microsoft is also starting to do more and more of) is the use of structured data. Do a query on any movie name, for instance. You know the big strip at the top of the query that comes up? This is Google making more and more use of structured data.
Google has talked about being able to scrape data off the web and then supporting schematized data in their pages. I don’t know anything specific for companies to do right now, but watch that and be aware of it. Because that’s in the future, to not just have keywords or even natural language, but to actually offer chunks of data so the search engines can understand and give you better exposure.
Stephan: Don’t you see it as a risk? If I provide enough structured data, won’t Google start answering questions directly and not referring the searcher to me? Someone might go to their search results and decide, “Oh, this is the birth date of the tenth president. No need to go further.” Because Google just confirmed it from 10 other websites using their structured data. That doesn’t seem very nice.
Ramez: I think there is a risk there. But I think that there are reasons not to worry. First, when people perform SEO, they are generally really focused on commercial content and some action to be had.
Second, people often want more than that specific answer, they want a deeper dive. Google is usually pretty good at giving a citation noting where the information came from. So that could be a way to actually get your content promoted instead of somebody else. What if they did just want that little snippet of information? Often your search listing gives that away anyway. But still, having the right content gets the views.
Stephan: How important do you consider freshness as a ranking factor?
Ramez: It used to be that freshness didn’t even matter at all. Nowadays, on 20% or 30% of queries, there is almost always stuff that has been updated in the last 24 hours or 48 hours near the top. If people want their site to maintain relevance, they are going to have to have fresh content. That’s something to have people think about. My numbers are soft, but I know industry experts talk about it.
Stephan: And how does “social” factor into that idea of freshness?
Ramez: The signals from Facebook and Twitter are obviously key signals to both Google and Bing already. They’ve probably become even more key signals to what is relevant.
Everyone keeps thinking, “Make your content, have the right terms and have the link with the right terms.” But how do you get people to tweet your content? How do you get people to share your content on Facebook or Google+? Because those are important signals to the general algorithms.
They are also are super important for that freshness signal. And that freshness signal is a ticket to the top of the search results for the first two days that it is out there. You want your site to be constantly popping to the top of the search results by the steady pulse of content that is getting there.
Stephan: Where do you see progress happening with speech recognition and natural language?
Ramez: I think speech and natural language are two separate things. Speech runs into all these issues with shared spaces and privacy independent of the natural language. You could have a speech interface for queries without natural language. The query could be “Hotel Las Vegas,” and that’s what I say aloud.
The typed interface is a little more natural language or a little closer to natural language-rich. “Hotel Las Vegas on the strip under $200 a night.” I think what you’ll see is that the latter will just kind of creep in, bit-by-bit. If you went back 5 years of time and took some long queries, you’d get much worse results than you get today. They are just getting smarter and smarter at saying, “Oh, these two words together mean something, and these two words together mean something, etc., etc.”
Stephan: Do you see AIs (artificial intelligence) becoming curators of content? The search engine will respond, “Okay, I’ve run through this research to answer your question on the topic to give an answer to you, and here is my summary based on that.” Do you think that’s coming with AI?
Ramez: I think that summarization is a possibility. We’ve had it for a while, but not on a large scale. It’s tricky because it’s easy to lose a sentence in a summary and blow the whole thing. It’s a summary crafted by a computer, which could be wrong. There are articles on Wikipedia created by people, that if you omitted a sentence, you would lose the meaning or would flip the meaning in a different direction, and that’s an easy mistake for an AI to make.
I think the way that summarizations will make more of an impact is in collapsing documents. If I was worried about summarizing documents in one particular industry, it would be the news business.
Reuters or AP puts out a piece of content which is the basic news story. Then 5,000 news sites print it verbatim with maybe a bit of embellishment. The Washington Post will take an AP report and add 3 paragraphs to it. Summarization could then list the AP report and the 10 places that added the 10 paragraphs. And that’s very much a threat to the way news works. That’s why news sites hate Google News.
The industry very similar to news, in terms of mass duplication of content, is shopping. You have the same base product information and the same base description. When you use Google shopping, they have the item that’s been defined and you see all the offers. It’s really not the content on the page, it is the merchants. It’s their ability to offer the structured data to say that they have the item in stock, and then they compete on price and shipping speed.
Stephan: Any other major industries have to worry?
Ramez: Any place where the content has the same function as somebody else, you are in deep trouble in the search results. You have to have some way to differentiate. What is it that makes you special and helps you stand out to users and to the search engine? For a news site, it’s original content produced by you or your users, it’s original images, it’s being the first to report on something, it’s an excellent social media presence that has people tweeting and sharing your stuff the most and the fastest to get your name to the top of the results.
For the retailers, it’s similar. It’s original content, it’s reviews that users are posting, it’s original images, it’s users posting images of the products, and it’s social media presence. People say, “Hey, I want to have really good SEO so people can find my site,” but if you can get people engaged with your site — build an engaging experience that gets people talking — that will naturally pump you up in search results and hopefully get them to share your stuff with others.
Stephan: That’s pretty hard for a startup that doesn’t have the funding to buy all that traffic from Facebook and Google AdWords. You’re building a baseball diamond in the middle of a cornfield (gratuitous “Build it and they will come” Field of Dreams reference), going broke waiting for traffic to show up.
Ramez: Google might not always be the way to get people initially. People usually want to create this broad, generic site right off the bat. But it might be better to build something that is more in a niche, and then lead with something unique that you do in that niche.
Lead with personality, lead with attitude, and lead with depth that allows you to be unique. That often gets traffic via word-of-mouth (or nowadays, social media). You have something that no one else offers. If you have people that like your stuff and are therefore sharing it on Facebook and Twitter, you will start to pop up in the rankings of Google and Bing.
Say you’re an online retailer and you want to go into apparel. That’s obviously a crowded market. Maybe you’d be best to get into a very specific market. Maybe you want to be the best place for scarves, maybe you want to be the best place for little black dresses, I don’t really know. You might find that you are more able to focus on creating really good, high-quality content there.
I have a friend in Seattle that started a wine site, and wine is also a relatively crowded category online, and she’s doing pretty well. What she did was focus on creating unique content, and that content was so cool and so unique and so useful.
A Seattle reporter found her and wrote an article that drove tens of thousands of hits to her site, which drove more sharing of her stuff on social sites, and led to a mountain of new traffic. That’s what you really want to have: unique content and value that nobody else has. Then you will naturally draw an audience and naturally build a following.
Everything you need to know about SEO, delivered every Thursday.