This appeared last week:
Google
‘Google says I’m a dead physicist’: is the world’s
biggest search engine broken?
For decades
now, anyone who’s wanted to know everything about anything has asked Google.
But is the platform losing its edge – and can we still trust it to tell us the
truth?
Tom Faber
Sat 20 Jul
2024 21.00 AEST Last modified on Sun 21 Jul 2024 02.40 AEST
I didn’t know
I was dead until I saw it on Google. When I
searched my name, there it was: a picture of my smiling face next to the text
“Tom Faber was a physicist and publisher, and he was a university lecturer at
Cambridge for 35 years”. Apparently I died on 27 July 2004, aged 77. This was
news to me.
The problem
was the picture. When you search the name of a notable person, Google may create what
it calls a “knowledge panel”, a little box with basic information taken from
Wikipedia. Somewhere along the way, the algorithm had confused pictures of my
face with the biography of another man who shared my name. According to his
obituary, he was “a distinguished physicist with a literary hinterland”. Google
provides a feedback form to resolve this type of bug. I filled it in several
times, but it made no difference.
I’m not the
only one who has been struggling with Google recently. Many users are saying
its principal product, its search engine, isn’t working as well as it should.
They claim the ingenious vehicle that has enabled us to navigate the internet’s
infinite scroll of information is beginning to rust and decay. That’s not to
mention the company’s endless court battles with rival companies and world
governments, or the rise of ChatGPT, which many tout as a search engine killer;
even Bill Gates
said last year that once a company perfects the AI assistant or “personal
agent”, “you
will never go to a search site again”.
Yet it’s hard
to imagine anything taking Google’s place. Last year it turned 25, and Alphabet, its parent
company, currently ranks as the fourth
most valuable in the world, worth more than $2tn (£1.5tn). Google has a
whopping 90% share of the global search market. More than a tool, it’s
practically infrastructure; the connective tissue that is fundamental to how we
find information online. This gives the company enormous power over politics,
social attitudes and the fortunes of countless businesses – anyone and
anything, in fact, that relies on the eyeballs of the internet to operate. Some
say Google is too big to fail.
It doesn’t
take a distinguished physicist with a literary hinterland to see that right now
Google search looks both deeply vulnerable and totally unstoppable. How can we
be sure the company really has our interests at heart? And can we still trust
it to tell us the truth?
The story of
Google reads like the stereotypical tech company origin myth. A couple of
computer geeks, Sergey
Brin and Larry Page, started a business in a garage in the late 90s and
built it into one of the world’s richest companies.
At the time
the web was growing fast and a few early search engines were trying to organise
the chaos. Page and Brin’s bright idea was to sort webpages not just by their
relevance to a search query, but also by the quality of their information. This
system, PageRank, prioritised webpages based on how many other pages linked to
them. The underlying concept, borrowed from academia, was that if many people
linked to a specific source, then it must have high-quality information.
It worked.
Coupled with Google’s clean, simple interface of a text box and a handful of
blue links, the site felt like magic. “Everybody could see then that Google’s
results were far better than the others’. That was the basis of everything,”
says Dirk Lewandowski, interim professor of data science at the University of
Duisburg-Essen, who has studied search engines for 20 years.
Google
quickly garnered a great deal of trust and goodwill. Its mission to “organise
the world’s information” was inspiring. If you wanted to know something, you’d
ask Google. Most of the time, it would deliver the answer you sought.
Gradually, the other search engines died off. Search became synonymous with
Google, and “Google” became a verb, and began to expand beyond text to images
and video, even mapping the physical world with Google Maps and Street View.
Success
generated more success, and Google captured vast amounts of data on its users
that it employed to improve search algorithms. The company realised that this
data could be valuable. With its search engine, it was capturing users’
thoughts, desires, their innermost questions. Google used this information to
reinvent the advertising industry.
Cory Doctorow, an
author, activist and mordant critic of big tech, explains the company’s ad
system like this: “Say I have an 18- to 34-year-old manchild in central London
who’s got an Xbox and has been searching for information about gonorrhoea. Who
will pay to advertise to this person? Advertisers or bots bid for placement –
and the winners serve an ad to you.”
Many of
Google’s products besides search, from YouTube to Maps, collect data on users,
which enables personalisation of your ads – this model is the foundational
example of what technology commentator Shoshana
Zuboff has called “surveillance capitalism”. Ads became big business for
Google. Last year its parent company, Alphabet, earned 77% of its revenue from
them – that’s $237.85bn.
Along the
way, the company accrued not just economic power, but also social and political
power. Rosie Graham, a lecturer in contemporary literature and the digital at
the University of Birmingham, says we don’t just ask Google for information,
but also for “ways to live our lives”. When we look for answers to social,
religious or political questions, Google judges who are the trusted voices, and
who we should not hear from. “Google has the power to change the way we think
about things,” Graham says.
“It acts like
it’s just another company, but it’s not. It influences countries’ elections. It
has a huge stake in what’s profitable, what jobs can exist … in many ways it’s
more powerful than governments. Gone are the times when it can be this small
company that’s all cutesy and shoestring.”
Somewhere
along its path to success, Google lost the public goodwill it earned in its
early days. Once, its playful motto, “Don’t be evil”, featured prominently in
its code of conduct. In 2018, it was quietly downgraded.
Companies
become too big to fail, then too big to jail, then too big to care. Google is
too big to care
Has Google
search got worse? And if so, what does that look like? Imagine you used to go
to your local library and, when you asked for a book, it was produced
immediately. Now, when you ask for that same book, the librarian tries to sell
you a magazine subscription, waves about some different books they say other
people like, then finally produces a big stack of tomes with your desired book
wedged awkwardly in the middle. You might have an opinion about this change in
service.
This is the
portrait detractors paint of the current state of Google search. Doctorow, for
one, calls the most recent results “garbage”. Former Google employees have
posted scathing articles arguing that the company is floundering. Yet every
day, billions of people use the search engine, and it’s a safe bet that many of
them think it works just fine. How is it that some people believe this
essential internet infrastructure is circling the drain while others haven’t
noticed any change?
It can be
hard to say anything definitive about search result quality, because each
person’s experience is so different. If two people Google the same sentence,
they will get different results based on all manner of variables. Meanwhile,
Google constantly tweaks the algorithm.
Still,
critics do have specific reasons for saying the service is going downhill.
Google search is good only as long as it can serve up high-quality information,
and many claim it no longer does so reliably. They often blame this on Google’s
inability to combat spammers and the much-maligned search engine optimisation
(SEO) industry. SEO companies aim to make websites appear more highly in Google
search rankings to help their businesses. But this can lead to degradation in
site quality, as if content is tailored only to please Google’s algorithms.
Take recipe pages. When searching for cooking instructions, you’d probably want
to see them displayed concisely at the top of the page, yet most food blogs
bury recipes beneath a long anecdote. Food writers do this because they believe
Google ranks this format highly. But readers resent it.
On the
spammier fringes is what’s known as “black hat” SEO, bad actors who use
techniques with fabulously evil names such as “domain squatting”, “reputation
abuse”, “obituary spam”, “keyword swarming” or “parasite hosting” to bring
their content to the top of Google’s search results and turn a quick buck. Spam
pages usually have little meaningful content and are aggressively monetised,
hosting intrusive ads to profit from each visitor’s click. A recent study
claims that Google
does indeed have a big spam problem, but adds that other search engines
face the same issues.
Google and
the spammers are locked in a never-ending battle. The spammers come up with a
new technique, Google tweaks the algorithm to stop it working, then the
spammers come up with something else. Google’s vice-president of search, Pandu
Nayak, describes the dynamic as a “spy v spy situation”. Today the internet is
facing the looming threat of a new wave of AI-generated
spam, which threatens to overwhelm search engines.
Even when the
links returned by a Google search are of high quality, the other
criticism is that it’s hard to find them among the clutter. Where the company
once sought to send users onwards to relevant links as quickly as possible, in
recent years it has started answering more questions within the Google search
interface itself – so if you’re trying to find out something about sports
scores, the weather or film showtimes, solve a mathematical equation, or
perhaps find out the key publications of a certain distinguished physicist,
Google will provide that information in a little box for you, without you
needing to click any links.
Sometimes all
these little boxes get in the way of the answer you’re trying to find. I just
typed in “best smartphone 2024” and was shown, at the top, a carousel of
shopping opportunities, followed by four links, then a panel of questions that
“people also ask” with vaguely related queries (“Which phones last the longest
years?”), then some YouTube videos, five more links, then more related queries
and a further shopping carousel. The links I was actually seeking were buried
by clutter. It’s a far cry from the sleek, minimal interface of early Google.
Sometimes,
Google will populate boxes with information gleaned from the internet that
turns out to be incorrect. Besides calling me a dead physicist, these info
boxes have claimed that Barack Obama was
the king of America, and asserted that Kannada, the official language of the
Indian state of Karnataka, was the ugliest language in the country – Google had
to issue an official apology for that one. It didn’t take me long to find an
inaccurate response. On the third random question I typed in, “How long is
Waterloo Bridge?”, a box came up with the confident answer “2,456ft” (748
metres). It was only when I clicked through to Wikipedia that I saw this figure
was in fact the length of the first Waterloo Bridge, which was
demolished in the 1930s. The current bridge is just 1,230ft (370 metres) long.
Google says
users find the additional panels useful. Critics argue that it is trying to
extract as much revenue as possible from users by keeping them within the
Google ecosystem to the detriment of the user experience. Early on, Google’s
founders realised that commercial incentives might compromise the integrity of
search results. In a 1998 student paper, Brin and Page wrote
that ad-funded search engines would be “inherently biased towards the
advertisers and away from the needs of consumers”. Yet Google started
displaying ads anyway because, as one academic told me, this is the only good
way to monetise search.
I spoke with
representatives of the competing search and answer engines Kagi, DuckDuckGo and
Perplexity, all of whom frame this issue as a misalignment in Google’s company
incentives. They say Google makes design and business choices to earn revenue
for advertisers and shareholders at the cost of search user experience. When I
put this critique to Google’s Nayak, he dismisses it as “an easy narrative” and
continues, “Since the beginning of Google, there has been a clear separation
between organic search [eg regular results] and ads … we make sure that the ad
side of the house does not affect search.”
When I ask
whether Google search results are getting worse, he repeatedly makes the same
point: the fact that Google has a 90% market share in search shows its product
works well. “If the search experience was not good, I have every confidence
people would not use it.”
Is there a reason
people might use Google search even if it weren’t a great experience? The
US Department of Justice (DoJ) recently gave its closing statements in a
historic legal case against Google, which the Financial Times called “the most
significant antitrust trial in 25 years”. The DoJ’s argument is that Google
uses its wealth to operate anti-competitively as an illegal monopoly,
principally by paying other tech companies to be the default search engine on
their devices. Naturally, Google argues that people use its search engine
because it’s the best. But if that’s the case, why did it need to pay Apple
$20bn in 2022 alone to be the default search engine across its phones and
computers? The case is likely to conclude before the end of the year.
Google’s
biggest competitor, Microsoft’s Bing, has only about 3%
of the global search market. A number of other startups scrape fractions of 1%,
many with their own spin on the search engine formula: Perplexity offers written-through answers
to questions, Kagi operates a paid subscription
model with no ads, DuckDuckGo focuses on
protecting data privacy. All three say it’s impossible to truly compete with
Google right now.
Aravind
Srinivas, chief executive of Perplexity, says, “Competing with Google is a
no-fly zone. They’re just too big, they dominate, have all the best technology
and a lot of money. They can just scorch you to death by offering whatever
you’re offering for free.”
Does it
really matter whether there is competition to Google’s search engine? Doctorow
believes it does. He coined the memorable term “enshittification” to explain
the state of big tech companies in the modern age: “Here is how platforms die:
first, they are good to their users; then they abuse their users to make things
better for their business customers; finally, they abuse those business
customers to claw back all the value for themselves. Then, they die.” He calls
Google “the poster-child for enshittification”.
Doctorow says
algorithmic systems are particularly vulnerable to this, because their workings
are opaque to users and easy to quietly tweak. Who knows why you see what you
see at the top of your Google search results, Instagram feed or TikTok For You
page? Is it because it’s judged to be the best content for you, or because it’s
what the platform thinks will make it the most money?
“That’s why
it’s so tempting for companies to enshittify them,” Doctorow says. “They’re
just yoloing it and saying, ‘Well, fuck it, we’re just going to make all
the things at the top of your feed garbage, because we’re too big to care.’”
Referencing a comment made by Lina Khan, chair of the US Federal Trade
Commission, he adds, “Companies become too big to fail, then too big to jail,
then too big to care. Google is too big to care.”
The last
stage of the enshittification doctrine is that a platform dies. Is this going
to happen to Google? Rosie Graham thinks it’s inevitable, at some point. She
points out how X, formerly Twitter, faded in relevance practically overnight after
Elon Musk took over. “No company lasts for ever,” she says. “There are all
sorts of organisations that had huge global influence and power that we don’t
have any more – think about the East
India Company. It’s not a matter of if Google will be influential
for ever, it’s a question of when Google will be replaced.”
Not everyone
agrees the search engine is getting worse. Of the dozen academics and industry
professionals I interviewed for this piece, half said they didn’t think search
quality was declining. Several pointed out people have been eagerly predicting
Google’s death for years – there’s even a Wikipedia page called “Predictions
of the end of Google” with examples dating back to 2007. Lewandowski says,
“There have always been complaints about low-quality results and the interface
getting more cluttered. But in the end, it’s basically the same.”
Perhaps what
is really bothering people is that the internet as a whole feels, in 2024, like
a worse place to be. Those who grew up on the web of the late 90s and early 00s
might remember openness, community and free thinking. Today, we’re probably
more likely to associate the internet with anxiety, loneliness and stress.
Maybe we miss the time when the internet felt more human. This may explain why
many people searching for information look to Reddit rather than Google. That
huge, chaotic forum feels like one of the last truly human places on the
internet, where you can get somebody else’s honest opinion in all its
weirdness, untainted by murky brand associations or affiliate links.
Yet even if
it’s true that the internet has declined, that doesn’t let Google totally off
the hook. Its search engine doesn’t just organise
information on the web, it actively shapes it. If the web is a worse place
today, if it’s over-commercialised and full of low-quality content, if
journalism platforms struggle to make money from good writing and are reduced
to clickbait and affiliate links, that’s partly Google’s fault.
“Google has
never really understood the responsibility it’s got to ensure publishers can
continue to publish content without needing to over-commercialise in horrifying
ways,” says Simon Schnieders, chief executive of the SEO company Blue Array.
“They really need to point the finger at themselves and why they created this
beast in the first place.”
As if the
search question wasn’t tricky enough, today many people are predicting that the
arrival
of new AI technologies is going to change everything. Since the launch of
ChatGPT, technologists have wondered whether AI assistants will one day take
the place of search engines. ChatGPT
is a product of OpenAI, which is in partnership with Microsoft, operator of
Bing. Last year Microsoft announced it would integrate AI answers into its
search engine results, with its chief executive, Satya Nadella, calling it “a
new day in search”.
At its latest
conference, Google seemed to be rushing to respond to this threat when it
announced a suite of new AI tools. The company has been using AI behind the
scenes to improve its search and algorithms for years, but with the launch of
the new “AI Overviews” feature, which it has already started rolling out and
plans to make available to more than 1 billion users by the end of the year, it
will put the technology front and centre. With this feature, Google search will
respond to certain queries with a text box above the usual links, providing a
written-through summary of information from various websites. The slogan the
company keeps repeating around this is “Let Google do the Googling for you”.
Asked ‘How
many rocks should I eat?’ Google gave this AI Overview: ‘One per day because
rocks contain minerals and vitamins’
Responses
have been mixed. Google says in internal tests people found the feature useful.
They claim it’s good for queries that require a number of specific variables –
say you’re trying to find a place your family can eat in Paris that does vegan
food, is open at 7am and is within walking distance of a Métro station. Such
questions, which might previously have required 10 minutes of clicking around
on numerous searches, can be done in seconds by AI (ChatGPT is also pretty good
on this stuff).
But, of
course, the internet immediately seized on the feature and found that, in
response to certain queries, it produced hilarious, inaccurate and sometimes
dangerous answers. In response to the question “How
many rocks should I eat?”, Google presented this AI Overview: “According to
UC Berkeley geologists, eating at least one small rock per day is recommended
because rocks contain minerals and vitamins that are important for digestive
health.” Meanwhile, someone who asked about “cheese not sticking to pizza” was
recommended to “add about 1/8 cup of non-toxic glue to the sauce to give it
more tackiness”. Naturally, it didn’t take long for somebody to make and eat
the glue pizza.
The erroneous
information came from obviously dubious sources. Eating rocks was suggested by
an article on satirical site the Onion,
while the
glue pizza idea was a post by “Fucksmith” on Reddit 11 years ago. Google
was roundly mocked online and responded with a blog post essentially saying
that these were growing pains and that the product would improve.
The new
direction heralded by ChatGPT and AI Overviews is to a world in which we no
longer search for answers ourselves, but rather receive a single, supposedly
balanced answer that has been pre-chewed by an algorithm. “Summarisation, or
the dumbing down of search, is a bad thing for society in general,” Schnieders
argues. “It’s important to get a range of diverse perspectives from search,
from your own trusted or new sources, practise critical thinking and form your
own opinions. AI Overviews claims to do this but it’s too much of a black box
to be trusted.”
Yet the
biggest concern around AI Overviews came not from users who had cheese sliding
off their pizzas, but from writers and publishers online. They’re worried that
if Google summarises the information from their websites and delivers it to
users, then those users will have no reasons to visit the sites, depriving them
of the traffic necessary to fund more content creation.
This question
has sent the media into a tailspin. Nilay Patel, editor-in-chief of American
technology news website the Verge, said AI Overviews would “change the web as
we know it”, while the influential tech reporter Casey
Newton said Google had “essentially put the web into a state of managed
decline”. Reports have predicted that publishers could see their search traffic
fall anywhere between 25%
and 60%
as a result of AI. If even a fraction of this is true, it could have enormous
consequences for the already squeezed media landscape. Various small publishers
have accused Google of killing their
businesses.
When I ask
Hema Budaraju, Google’s senior director of product management for search,
whether Google has a responsibility to ensure a healthy web ecosystem, she
answers with “a very direct, strong yes”. She emphasises Google’s “commitment
to the web” and “ensuring we have a thriving ecosystem”. “As we introduce our
generative capabilities, connecting people to the web is central to our
approach,” she says.
I’m inclined
to believe this is Google’s intention. It’s not in its long-term interest for
the internet’s information economy to collapse. If media platforms can no
longer afford to produce high-quality content, then Google’s AI Overviews will
have nothing left to summarise. Eventually, people would stop using Google
search.
During a
visit to one of Google’s London offices, I told a member of the communications
team that I had been inspired to look into the state of Google search by the
broken knowledge panel saying I was a dead physicist. When I later spoke to
Nayak, I decided to ask him how I could detach my face from the other Tom
Faber. Before I’d finished my sentence, the communications person pinged into
the chat, saying the issue had been fixed. I adjusted my question: maybe it’s
been fixed for me because of having a direct line to Google, but how would
someone resolve this if they didn’t have strings to pull?
Nayak
apologised, saying the panels were created automatically using algorithms and
sometimes they messed up: “These are the kinds of things we’re constantly
improving.” He went on to insist that, on issues like this, Google’s “honest
results policy puts everyone on the same playing field”, so it had nothing to
do with me having contacts at Google. Immediately after, the call was ended
abruptly for going over time.
When I
followed up, I was told someone from the team had submitted a feedback form
using the public channels, just like anyone else might, and this resulted in it
getting fixed. This was perplexing. It seemed more than a coincidence that,
after years of trying to fix the problem myself by submitting feedback forms,
it would finally change weeks after mentioning it to a Google employee, and
that this change would be unrelated to that conversation. But at least the
issue was fixed.
For now,
search engines aren’t going anywhere. “I think search is inevitable,” Doctorow
says, but adds that we’ll always want human voices to cut through the noise and
deliver curated sources of information. “We’re still going to have experts,
reviewers and tastemakers, adventurous spelunkers in information space, and
just the terminally curious – that one friend you have who can’t stop holding
forth about something, whether it’s a new gamer mouse or a band or a new,
extremely hoppy IPA. We’re going to have all of those things, but they’re all
going to need search engines.”
It seems
search is going to remain broadly the same – ChatGPT isn’t about to displace
Google search, and AI Overviews are not about to fundamentally change the
search experience. But there remains the larger question of what the rest
of the internet will look like. As AI steers the online economy into
uncharted waters, the fate of the media is hanging in the balance. And whatever
the future of the web looks like, it’s sure Google is going to play an enormous
role in shaping it.
As I was
editing this story, I double-checked to see if my knowledge panel was still
fixed and discovered, confoundingly, that the problem had returned. The next
time I checked, it was fixed again. I began to regularly Google myself in
incognito mode, on different browsers and devices. Sometimes my face popped up
as a dead physicist; sometimes it didn’t. The ever-changing, algorithmic nature
of it made it feel like Schrödinger’s knowledge panel, both correct and
incorrect at all times.
The
frustration of trying to resolve the knowledge panel issue echoed the process
of trying to get a solid read on the state of Google’s search engine. The
company is too opaque, and its system has too many shifting parts, to make a
clear pronouncement. It is many things to many people, constantly succeeding
and failing its billions of users. Sometimes, as much as you search and search,
there is no single answer to be found.
Here is the link:
https://www.theguardian.com/technology/article/2024/jul/20/google-is-the-worlds-biggest-search-engine-broken
What all this says to me is that
trying to corral the scope of human knowledge and make it findable for the
average punter may just be a leap too far!
If you consider just how hard fact-finding
20 years ago was compared to now one has to concede we have come a very long way
and more worrying is the fact that a quantum improvement from here is pretty
hard to imagine.
I really find it hard to imagine life
without Google search (or some equivalent) and also struggle to envisage what
the “next big thing” will look like!
I wonder where the garage is that
the next step is being dreamt up?
David.