When we google a topic, many of us trust that something called "the algorithm" will magically serve up the right information.
But what is this algorithm? How does it work? And are we all getting different results and therefore different information about the world around us?
To answer these questions, researchers are rolling out a project that enlists the help of the public.
The year-long Australian Search Experience project invites Australians to install a browser plugin that automatically queries Google Search, Google News, Google Video and YouTube several times a day on key topics. It runs in the background and does not access any personal data.
The Australian project will coincide with the next federal election, and address longstanding questions about the influence of search algorithms over the way people vote.
Do people who vote one way get the same search results as those who vote another?
By comparing the search results generated with the demographic details of potentially thousands of participants, the researchers aim to understand how the various search algorithms weigh different factors.
Basically, they're trying to study Google's algorithms by reverse engineering them from search results.
Here's what they've found so far.
COVID search results 'carefully curated'
The project launched last month and so far has more than 500 participants who have "donated" data from more than 40 million search results, which adds up to 20 gigabytes of data a day.
So far the clearest trends are for Google Search, according to the project's chief investigator, Queensland University of Technology's Professor Axel Bruns.
"Very, very early results" suggest most Australians are getting roughly the same information when they google a topic, regardless of age, gender, and other demographic factors.
"From what we're seeing so far ... on Google Search there is probably less evidence of any significant personalisation or problematic personalisation," Professor Bruns said.
This isn't that surprising, he added.
Unlike YouTube, which is designed to keep users engrossed by providing a highly personalised feed of content they want to watch, Google Search is just a handy place to get information quickly.
"Google Search is obviously not where people come to hang out, but to get the right information at the right time, so personalisation is somewhat less useful."
What was more surprising, Professor Bruns said, was the level of curation of search results for certain topics.
Curation is where Google suspends the usual operation of the algorithm to actively promote or bury certain information.
Google has long stated that it does not use human curation to collect or arrange the results on a page, though the line between this and a human tweaking an algorithm to curate search results is a fuzzy one.
Professor Bruns pointed out that a search of the term "COVID" delivers a whole page of embedded content with information on case numbers and symptoms as well as links to official health information.
"The first page of Google results for terms like COVID and vaccines and lockdown and quarantine is fairly carefully curated," Professor Bruns said.
"It shows a whole range of embedded content before you finally get to the actual search results themselves."
Before the pandemic, such an obvious curation of search results would have been very unusual.
Curation on this scale would require "some level of human intervention", Professor Bruns said.
In a statement, Google said "dedicated features on important topics like health and elections" is something the company has done for many years.
"Since the beginning of the pandemic, we have worked closely with national and regional health authorities to make trusted information available to all Australians," a Google spokesperson said.
'There's very little transparency'
In this case, the curation is for a worthy purpose — to reduce the spread of misinformation — but the lack of transparency or oversight is worrying, Professor Bruns said.
As when Google conspicuously experimented with burying links to Australian news sites earlier this year, it shows the corporation's power over the public's access to information.
"There's very little transparency about what topics they've done this for and how such topics are chosen, and by whom, or how they choose which content to highlight via manual curation," Professor Bruns said.
"They're not revealing how they make the choices for curation and [they're not indicating] when you're encountering a curated search page.
"A project like ours aims to generate some more transparency about this, and the workings of search engine recommendations more generally, by observing them from the outside, at scale, using a citizen science approach."
Project to coincide with federal election campaign
The research follows a similar project run by Algorithm Watch in Germany in 2017, which designed a browser plugin that would collect the search results of queries related to the German election.
From more than 5 million search results, it found that the degree of personalisation was relatively limited, said Matthias Spielkamp, founder and executive director of Algorithm Watch.
"We are not much of a believer in the filter bubble theory," he said.
"What we found is that the personalisation is done on a pretty low level."
But he said the "significant" Australian project could uncover more than the German one.
Australia's small population makes it an ideal subject for the study, as relatively few data donors can give statistically significant results.
Google has also become more "hands on" with curating search results since 2017.
The Australian project will also look at YouTube - a platform where the content fed to users is much more personalised.
The project is funded by the federal government through the Australian Research Council's (ARC) Centre of Excellence for Automated Decision-Making and Society.
"We will particularly look at searches we've got queued up for political topics, political leanings, and so on to make sure that we get a good sense of what people see for political topics," Professor Bruns said.
"We'll continue to vary the search terms as other things pop up, whether that's bushfires in the summer or whatever it might be."
Google Australia is an industry partner of the ARC Centre of Excellence for Automated Decision-Making and Society.
Professor Bruns said Google was being kept "very much at arm's length from this particular project."
He stressed that researchers do not have access to participants search history or other details of what they search.
"We're simply running our own search terms," he said.