Disruption comes to Google

THE tech industry is extremely dynamic and the only constant is change. Companies that once appeared to be invulnerable can be regularly upset by upstarts unless they adapt. We can see this in the difference in fortunes between Microsoft and Intel which both saw the PC displaced by mobile devices. Microsoft adapted by pivoting to the cloud while Intel's fortunes have declined as ARM, TSMC and Nvidia are ascendant.

Google is facing similar potential disruption in its core search business due to the rise of answer engines like Perplexity, ChatGPT and Microsoft's Copilot. The media narrative has turned against Google, especially towards its CEO Sundar Pichai, with articles with titles like Alphabet needs its own Mark Zuckerberg and Alphabet faces ‘clear and present danger’ of falling short in AI being published almost daily.

However, this negativity seems somewhat extreme in comparison to the facts. ChatGPT's growth on the web has flatlined since peaking in May of 2023 while Alphabet reported record profits of US$20,7 billion last quarter which was a 52% increase from the prior year. So which is it? Is Google dying or is this just the media's love for an underdog story making them run away with the facts? The truth as always is somewhere in the middle.

Google stumbled upon one of the most symbiotic and profitable business models known to humanity. People come to Google looking for goods and services they want to buy and it shows them ads, people click on these ads and Google gets paid. It is amazingly simple and lucrative.

Websites create content that makes Google useful, free of cost to Google and in exchange get valuable traffic from people who click on links in search results. This created a feedback loop whereby the more people there were using the internet, the more content there was for Google to index and the more traffic there was for Google to send to websites.

The more people there were using Google, the harder it was for other search engines to compete since Google had more data to train the relevance of its engine than anyone else. Google then put its thumb on the scale by paying out over US$26 billion a year in payments to ensure it captured the most valuable search entry points, thus preventing any upstart from doing to them what it did to search engines like Alta Vista and Excite decades ago.

For decades it has been structurally impossible for an upstart search engine to compete with Google due to these dynamics as Microsoft testified in court. However, technology disruption can render structural advantages irrelevant. The same way digital cameras made Kodak's advantages in film cameras irrelevant and touch-based smartphones rendered Nokia and Blackberry's advantages irrelevant, LLMs (LLM-based answer engines) have threatened to do the same to Google Search.

Fifteen years ago Microsoft launched Bing as a decision engine, a new search engine with a pitch that it had a different point of view from Google's approach of showing people 10 blue links (plus ads) in response to their queries. The following excerpt from the press release feels prescient when you consider the pitches of answer engines like Perplexity and Microsoft's Copilot.

Panelists at a big technology companies gave the following views.

“Bing is specifically designed to build on the benefits of today's search engines but begins to move beyond this experience with a new approach to user experience and intuitive tools to help customers make better decisions, focusing initially on four key vertical areas: Making a purchase decision, planning a trip, researching a health condition or finding a local business.”

“The result of this new approach is an important beginning for a new and more powerful kind of search service, which Microsoft is calling a decision engine, designed to empower people to gain insight and knowledge from the Web, moving more quickly to important decisions.”

Even back then, it was obvious that in many cases what people want from search engines is an answer to their question and a tool that can help them with decisions they need to make, not a collection of links to websites. This is especially true today when a bunch of these links are to SEO content spam websites even on Google.

However, the problem with Bing was that the marketing pitch was disconnected from the technology. Bing was just another web search engine that returned 10 blue links regardless of how much the ad campaign said otherwise. And thus it failed to gain traction for the structural reasons previously mentioned.

Today, answer engines powered by LLMs can do what Bing's marketers claimed it could, but actually could not 15 years ago. I personally now have entire categories of queries I only perform on ChatGPT instead of Google because I know I will get a direct answer instead of having to trawl through a bunch of poorly designed websites searching for an answer.

Additionally, LLM-based answer engines have broken the symbiotic model of web search, Whereas the 10 blue links model drives traffic to websites thus incentivising them to make their content available to be crawled by search engines like Google, LLMbased answer engines answer user queries inline and do not send traffic to websites. Thus websites who were already moving towards paywalls to address the various attacks on online advertising, such as Google Chrome deprecating third party cookies, are further incentivised to hide their content from search engines thus making Google's declining product quality even worse.

So Google is now at risk of entering a negative feedback loop whereby in competing with tools like Perplexity and ChatGPT, it actually undermines its core product and harms the overall web ecosystem.

Finally, Google’s business model is built on showing people ads as the first result over organic content, in some cases filling the entire page with ads and you have to scroll before seeing organic results. A user experience where you simply provide a user an answer provides no opportunity to show ads.

⬤Read more on www.newsday.co.zw

⬤This article first appeared on medium.com