The Shape of the Web
On generative search and the future
I’m Tim Gorichanaz, and this is Ports, a newsletter about design and ethics. You’ll find this week’s article below, followed by Ports of Call, links to things I’ve been reading and pondering this week.
The web has been changing shape right beneath our noses.
It’s hard to notice slow changes. And your most-visited websites are probably the same as they were a decade ago. But things have changed, on many levels. Aesthetics, of course: If you pull up a 1990s or 2010s screenshot of a website you know well, it’s hard to miss how different things look.
The way we navigate the web has also changed, and just as dramatically. I think of this as unfolding in three phases:
At first, you had to know the web address for each page you wanted to go to. You felt a little bit arcane typing in :// for the first time.
Later, directories and search engines were built, connecting you with pages on topics you sought out (in the form of a list of search results)—and eventually connecting you with excerpted information from those pages. Now you can search “tom cruise age” and Google will tell you “62” right at the top of the page.
Today, we’re seeing a new way of navigating emerge. Custom content can be generated to respond to your needs, such as by summarizing numerous existing sources or providing a top product recommendation. We have Google’s AI Overview feature, for example, as well as more cutting-edge and experimental offerings such as in the Arc and Dia browsers from the Browser Company.
One concern is what this new way of navigating will do to the business model of the web.
Since the mid-1990s, the web has been advertising-powered—first with banner ads, now with big data. In terms of where they make their money, Google and Meta are advertising companies, not tech companies. Up to now, the way internet companies get paid to drive relevant visitors to other companies’ websites. But now with generative search, you don’t have to visit any other websites. You benefit from others’ content without “rewarding” them with a visit. How might that change things?
A Business Model Under Pressure
The effects of generative search are already visible in web traffic data, and an article in Barron’s this week says: Google Search is Fading, The Whole Internet is at Risk. As that article reports, across many industries, search traffic is down—U.S. tourism sites by 20% year-over-year, ecommerce by 9%, news by 17%.
Internet search traffic has been falling for much of the past year as web surfers experiment with artificial-intelligence-powered search from OpenAI’s ChatGPT and AI start-up Perplexity AI. So far, referrals from AI search engines have replaced about 10% of the traditional search losses. […]
Across the web economy, the trend is clear: Search is drying up, and Google is no longer the clear-cut way to drive audiences to websites. The changes have begun to force a reckoning across various industries. […]
The traffic conversation has the feel of the 1990s and early aughts before Google arrived and companies were still trying to figure out how to attract audiences across the World Wide Web.
Web companies are already reportedly cutting staff due to lower visit counts. The business model that the web has been built on is under pressure.
Some commentators have suggested that Google is in its sunset era, soon to be replaced by a nimbler, AI-first company such as Perplexity. But Google is also one of the foremost AI companies today, and has long been. They’re certainly at work to define the future shape of the web.
As that Barron’s article concludes, “Google will be fine. It’s the rest of the internet that should be worried.”
Finding Information Online
The notion that generative search will replace all other ways of navigating the web is overstated. So far, even as the successive phases of navigating the web have come along, the old ones have remained. You can still type in a web address manually, and in 2035 you’ll still be able to find a list of search results.
And that’s for the better.
Seeing search as a “command line for the web,” as tech journalist Brian McCullough recently put it, or as a Q&A system, somewhat oversimplifies search. Yes, sometimes you just want an answer to your question. But other times, search is about learning—it’s about building an understanding the information space around a topic.
One of the key skills for finding information now and for the foreseeable future is to be able to identify what sort of information need you have. Is it something for which a direct answer will do, such as Tom Cruise’s age, or is it something where you need to learn about the conceptual space around a topic, such as the different sides of an argument and different forms of evidence?
Generative search will not help you figure that out. It’ll give you a single answer regardless. It is built to give you false feelings of understanding.
The business of the web may be about to change. We may see new kinds of affiliate referral systems (paying AI companies to boost your products or perspectives), or search engines paying royalties to their information sources.
But for now, what’s not changing is that the onus is on the user to think well.
Ports of Call
Quotation marks in fiction: James Joyce famously hated quotation marks—he thought they were unsightly—so his fiction makes use of italics and other indicators to mark speech. Apparently there is something of a trend in this direction, particularly in Irish literature today. The Economist reports: “The inverted comma [what Brits call quotation marks] is falling out of fashion. ... In the 1970s, 94% of Booker-nominated novels used them, compared with just 72% in the past decade.”
AI 2027: A bold (and almost certainly wrong) prediction that superhuman AI will emerge in 2027, communicated in a month-by-month timeline. This will be one to look back at in a couple years to see where things fell short.
A bleak job market for new college grads: It’s a tough job market for new college graduates. Right now unemployment is higher for those with a college degree than those without. Many commentators (such as the New York Times) cite this as proof that AI is taking jobs away. But that’s overblown. What we’re experiencing now is a trend that started in 2009, with the financial crisis. As the Economist reports: “Lots of contingent factors are responsible. Many industries that traditionally employed graduates have had a tough time of late. Years of subdued activity in mergers and acquisitions have trimmed demand for lawyers. Investment banks are less go-getting than before the global financial crisis of 2007-09.” The question, “Is college worth it?” assumes that the proper outcome of a college degree is a job. Taking that for granted, the answer is becoming less certain. It’s interesting to note that “recent graduates’ rising unemployment is driven by those who are looking for work for the first time.” This suggests that if students can get work experience during college, they’ll be more likely to find a job afterwards. So, +1 for the cooperative education model at places like Drexel University.