The process of this test did remind me how much data Google collects — not just my searches but anything I say to a smart speaker, any search result or ad that I click — which will be a point in favour of DuckDuckGo for some, even if it didn’t perform as well and is harder to read. The data collection does also mean that Google (not just search but also mobile widgets and other services) personalises to me more effectively.
Bing has overhauled its layout over the past year to be very graphic and augmented by generative AI explanations from its chatbot. I like the idea of being able to ask follow-up questions in searches this way, but in practice it’s not really there yet. In a search for “kids BMX bike green” (Google did pretty well and DuckDuckGo ignored the green part), Bing Chat offered a comparison of a few different options but they were too expensive. I asked, “any under $200?” and it replied with more options that were all over $300.
Another observation from this test is that I tend to use the Google Search bar as much for returning to familiar sites as I do for searching for information, which means switching to another provider could be temporarily annoying simply because it wouldn’t remember all the places I’ve been.
Is Google as the default unfair?
In a response to the trial on Google’s blog, president of global affairs Kent Walker said the company’s search product was the favourite of users, browsers and device makers simply because it was the best in the business. And being default was not determinative, he said; changing from Google Search to any other search provider was easy, and the company competes with many others on phones and PCs.
“People don’t use Google because they have to, they use it because they want to,” he said.
“This lawsuit simply ignores how intensely competitive and dynamic the technology industry is today.”
Walker also points out that, when faced with a situation where Google is not the default, users tend to spend some effort putting Google back in charge. When Mozilla made Yahoo! the default on Firefox, most users changed it to Google. And when people set up a Windows device for the first time they frequently sidestep the Microsoft default — “Google” was the number one search query on Bing in 2021 — despite Microsoft making that very annoying to do.
This is a compelling point, and many of us will be well accustomed to dodging all of Microsoft’s pleading while trying to get Chrome and Google set as defaults whenever we set up a new PC. But all of these arguments do ignore the fact that Google and its competitors are not on an even playing field. The question remains, why do users prefer Google? And if the answer, as Walker says, is that Google is simply better, the question becomes whether that’s down to its scale, its longevity, its ubiquity and all the user data it sucks up, which no other company could possibly match.
For the US, which has to prove that Google somehow broke the law to achieve the status quo, this legal case could be an uphill battle. But for the rest of us, it offers a rare opportunity for some insight into what the tech giant does with its enormous market share. Is it still focused on being the best? Or, as has been alleged, does it use its ubiquity to squeeze us for money even at the expense of product quality?
How ‘organic’ are Google’s results?
This month, Wired posted an opinion article by lawyer and privacy advocate Megan Gray, which alleged that Google had accidentally revealed during the trial that it manipulates people’s search queries for maximum ad revenue. The example given was replacing a search for “children’s clothing” with “NIKOLAI-brand kidswear”.
Google rejected this in very strong terms, saying the piece was misleading and inaccurate while denying ever altering search terms. Wired removed the article for not meeting its standards, but the degree to which it was shared on social media and boosted in write-ups at other outlets shows just how willing people are to accept foul play on Google’s part. A suspicion of privacy invasion and inappropriate data handling follows the company at all times.
Grey herself is a former vice-president at DuckDuckGo, a privacy-focused company founded explicitly to counter giants such as Google. She admits that she may have misinterpreted the evidence, but maintains that Google manipulates Search to maximise ad revenue. So what does Google say?
In a post on X, Google’s official search liaison, Danny Sullivan, said ad systems do not affect the organic results, i.e. the list of blue links in search results that are not sponsored.
“Ad keyword matching is a long-standing and well-known process that is designed to connect people to relevant ads. A separate process, which has nothing to do with ads, is used to match organic results to a query,” he said.
“It’s no secret that Google Search looks beyond the specific words in a query to better understand their meaning … If you make a spelling mistake, or search for a term that’s not on a page but where the page has a close synonym, or if you aren’t even sure exactly how to search for something, our meaning matching systems help.”
It is true that some search queries result in more ads than others, and what ads you get can vary wildly depending on your request, which is the result of the keyword matching Google uses to sell ads. But Google denies pushing users towards the most monetisable subjects.
Similarly, it’s also pretty common to see people complaining on the internet with some variation of “Google Search was a lot better 10 years ago”, which could absolutely be true. But it would be a stretch to assume that’s because of the company’s monetisation practices, when there’s a much simpler explanation.
In 2013, people were creating around 9 ZB of data every year (that’s nine zettabytes, or nine trillion gigabytes), which Google had to index constantly, whereas in 2023 it’s more like 120 ZB. And not a whole lot of that is static websites like it was a decade ago, it’s live information that may be nested within various services and apps.
What really happens when we search?
Google has published extensive information about its Search procedures, so it’s not difficult to get its own take on what happens in the background when you send off a query.
Google’s software is constantly crawling the web and updating its central index, which itself is more than 100 million gigabytes in size. When you enter a query, Search uses a few quick processes to decide the meaning of your question, including what kind of information you might be after. Then, beyond just looking for sites with the words you typed in, the system uses machine learning to rank results according to what it thinks you’re looking for, prioritising high quality and easy to use sites.
If you allow it, Google does bring in some information it knows about you to further refine results, including your location and web history. If you’re looking to turn this off, it’s in your Google account settings called “Web & App Activity”. Google says it does not infer personal information like race or religion to shape results.
Google works with independent “search quality raters” to gauge the effectiveness of its processes, and says it ran almost 900,000 quality tests in 2022.
Some critics, including Megan Grey, contend that Google uses semantic keyword matching to make searches less precise on purpose, which widens the pool of content served and increases auctions for its ads. The quality tests, then, would ensure that the systems were still returning results that were justifiably related to the input.
However, Google’s help articles on its ‘broad match modifier’ in Google Ads put that kind of matching purely in the realm of ad-selling, which would affect the kinds of ads you see but not the organic results. For example, if an advertiser was paying for ads on a certain keyword, their ads would also appear in searches that carry that intent, even if the specific word was not included.
The goal is clearly more ads that are relevant to a person’s search, which is good for Google and advertisers but is very different from Google manipulating the search query to affect the non-sponsored results. And though none of that is strictly relevant to the antitrust case playing out right now in the US, it may be very relevant next year when the department of justice takes a second crack at Google, having filed a suit alleging illegal monopolisation of the digital ads market.
Get news and reviews on technology, gadgets and gaming in our Technology newsletter every Friday. Sign up here.