A Brave and Manipulative New World with Google FLoC Tracker

One of the biggest news Google published this year is their new tracking system FLoC which aims to replace all third-party trackers in Google products. However, this new tracker developed by Google themself using their massive user data sets has raised many concerns regarding online safety and privacy.

August 13, 2021 - 11 minute read -
Article Technology Privacy

For many years, people are using their privacy to trade convenience. When they open our browsers, they would like to see customized results suit their own tastes. No one wants to see what they don’t believe in willingly. When they are dining out, they want to get the best matched one on Yelp. When they shop, they want the most up to date merchandizes fit to their needs exactly so it is absolute not necessary to search for hours to find a pair of good fitting shoes. Indeed, human beings desire to be understood by others, even by machines. And being understood feels good for most of us.

Back to the Old Days

Right now, there are so many “smart” devices around us without much notice. We have smart bulbs, smart microwave, and even smart door locks. In order for a device to be “smart”, it has to store our preferences to “understand” our needs. All these devices are tracking information about us while in use, and we are no more than a profile to these devices and companies who make them. In old days, these companies might sell these data to advertisement companies to make profits, or they would most likely analyze these data by themselves to make more sales. For example, if someone uses light bulb mostly during late night time, it is very reasonable to conclude that one is active during the night. Thus, if these users are active on a schedule, then they might be working on a night shift. This information is enough for many advertisements to target this specific group of people. And even further, a person who works on nigh shifts are most likely in a disadvantaged group, and all people in this group can be easily targeted using tailored advertisements. This might sounds a little bit (or very) discriminate, but this is how advertisements companies are doing right row. This is wrong with no duobbt, but we cannot denial the truth that credit card approval, car loan, house mortgage, and many other products are based on extreme discriminations.

Human beings desire to be understood by others, even by machines.

However, even the situation is bad enough, there is a limitation here. Each company has one’s own data to analyze, and buying data from other companies is not cheap. So, the profiling is not quite complete. There is something missing here or there. Thus, third party tracking in the old days is still not “smart” enough. Sometimes, people earn good amount of money who are definitely not in need of money still get emergency loan advertisements. Back to those old days, we were not scared with a targeted advertisement.

The Machine Knows What You Think

Just like the old saying, if the profit is high enough, people would willing to do whatever it takes. With no doubts, the advertisements business has such a high profit and I won’t be surprised if anyone is willing to go to prison for that high profit. After years and years evolution, the advertisements companies have come up a full procedure to target their consumers.

For each person online, there is a profile of that person. And, this profile is not something we fill out when we register a web service. It is not our names, our genders, or our address. It is our browsing history, our clicking rate, and the time we stay on a specific page. We might not be honest when we fill out the preference form, but we are very honest of what we are browsing online. For instance, everyone around a person knows that this person only eat healthy organic food and never ever be anywhere near a burger place because this person only eat organic food in front of others. However, this person secretly eat fried chicken and In-n-Out every week privately without anyone knowing about this. But the cookies and trackers on In-n-Out website know that this person ordered burger and fried chicken. Then, the advertisement companies know this, and maybe next week this person will get a coupon for In-n-Out which lures one to make further payment on buying more In-n-Out burgers.

We might not be honest when we fill out the preference form, but we are very honest of what we are browsing online.

Not just food, the machines know everything about us even better than ourselves. We lie when we talk about ourselves to blend in others, but we never lie when we click on something we are secretly interested, and we never lie when we stay on a page which interests us and close a page quickly when it draws no interests from us at all. All these data are feed into machines which are learning us. This produces a very scary reality. I have heard from more than one friend that if he search Google about something such as toothbrushes, he can see an ad about an attractive toothbrush on his Instagram. And very sad, all ads showed on his instagram are actually something he really wants to purchase. We know how this ends, he bought everything Instagram showed him.

This surreal situation didn’t just happened in something one searched, but even one thought about and talked about. I guess when a person shows interests in buying something, there is a trace on how one brows online. And, the machine can figure out the hidden trace to understand what this person really wants to buy.

The Manipulation

After knowing what people want to purchase and feed them with what they want, the advertisements companies are not going to stop at this point. They want to use these machines to tell people what to buy. In other words, these machines are manipulating people on making them buy what the advertisements companies want them to buy instead of what they really need. As we all know, brainwashing is a standard advertisement strategy, and the core of brainwashing is repeat. As long as you repeat enough times, lies will become truths, and no matter how strong one is, any human being will surrender.

The dangerous part is when someone started trusting one’s browser, and click whatever Youtube present to one, the computer can heavily influence one’s mind by presenting different information.

These manipulation has a long history with advertising activities, but it is never this easy nowadays. Right now, advertisements companies have your personal online profile, and they know exactly which websites you are visiting, and how long you stared on a specific page. By knowing these information, if you are the target, the “thing” ad companies what you buy would show up everywhere you visit. One would have a feeling that this product is everywhere, and at some point, one would at least start considering purchasing this product. However, the interesting thing is, this product is not actually “everywhere”, it is just all over the places you are visiting.

There was a quite interesting experiment I did. I first let my chrome browser learn me what I watch on Youtube on purpose for a while, then burned all my trace to see the “original” youtube. After learning me, all videos showed up on my index page would be something like “cute cats”, “Biden”, “books reading”, “new movies” which seem like fit my made up personalities well. I started having a feeling that the whole world is using the “same” Youtube as I am. However, when I burned my data and viewed the “original” Youtube, I was in a shock for a moment with all the information I feel that would not even exist in this world. The dangerous part is when someone started trusting one’s browser, and click whatever Youtube present to one, the computer can heavily influence one’s mind by presenting different information.

A New Era with Google FLoC

If you thought the ad companies are scary enough, you have underestimated how far they can go driven by profits. During April 2021, Google started testing their very own tracking software where they are trying to replay third-party trackers with, FLoC. Generally speaking, FLoC aims to be the tracker on all Google’s services, especially on Chrome browser, to replace all other trackers we have now. According to Google, this move intend to replace all third-party trackers to have a well-designed tracker under a proper standard. The main reason for doing so is third party trackers are violating too much personal privacy without proper control. However, if this is the goal, why not getting rid of trackers completely? Well, I guess this statement of reason for why they develop FLoC which first seems reasonable is in fact another Google bullshit to cover their ultimate goal of making more profits.

Here come at least two potential vulnerabilities within the algorithm itself, first, the potential to identify each user, and second, the targeting based on vulnerable groups.

Anyways, what is FLoC exactly? Long story short, FLoC is a tracker developed by Google which profiles everyone using Google’s services to serve ad companies better when targeting their potential consumers. FLoC runs on a cohort system where it classifies each person into a cohort, and within each cohort, users share some similarities. For example, a cohort of users who all like hiking and potentially buy more hiking shoes. Here come at least two potential vulnerabilities within the algorithm itself, first, the potential to identify each user, and second, the targeting based on vulnerable groups.

Within FLoC, each online user would have a tag to identify oneself without revealing one’s true identity in the mean time while tracking. In order to do so, Google states that their cohorts will be large enough to make everyone’s true identity not reversible within each cohort. Specifically, each cohort will have at least thousands of users so sites are not able to cross contrast to figure out who exactly is in a specific cohort. However, the risk still exists if sites have access to all cohorts and have a bit knowledge of a person, a person can be easily identified, or at least identified into a very small group of people. Sadly, every partner of FLoC would have access to all cohorts’ user data. The way Google approaches to solve the issue of how to keep users’ identities in secret while tracking raises another important issue. After putting at least thousands of people in a cohort, on the contrary, there are as many people as possible within a group so ad companies are able to target a huge group of people easily. Think about it, if a cohort of users all share similarities in hiking and potential of buying hiking shoes, won’t it be easier for ad companies to target a single cohort of users in stead of finding a bunch of cohort to target together?

Things are not that simple. And if things are ever that simple, it would be splendid.

Besides the danger of putting users’ real identities on an edge while tracking using FLoC, there is another important issue regarding Google’s FLoC tracking system, the targeting towards vulnerable groups.

As Google states, machine learning is heavily involved in the FLoC cohort grouping algorithms. In which case, machines are going to group people into cohort ruthlessly and robustly without any humanity. Thus, vulnerable groups are going to be identified regardless, and ad companies who have access to these vulnerable groups are able to do very bad things such as manipulating people with depression and many others. Of course, Google claims there is a solution within FLoC algorithm for this issue. FLoC’s solution on this is whenever there is a vulnerable group grouped out, the algorithm is going to dismiss this group and re-group everyone inside. However, how would you find out this is a group of people with vulnerability? For example, for people with depression, not all of them are under medicine, even worse, many of them are not aware of having a depression. Things are not that simple. And if things are ever that simple, it would be splendid.

Fake Socializing

Beyond what we have discussed above, I would like to talk about a very interesting scenario I have been seeing these days regarding online profiling by human beings. The fake socializing. Indeed, being social is “normal” to many people and being self-satisfiable is “abnormal” to them. This leads to that in many cases, people are pretending to be social even if they are not. In order to do so, people start to publish their connections and information, aka personal information, online. For example, many successful business such as Facebook, Instagram, and LinkedIn are based on this need. People connect to people they don’t know to look better and impress other people. Now, there are more and more platforms where you can follow others and get followed to show how social you have, and people are even buying followers to make themselves look better.

The online fake socializing has also made people believe that what they have learned about others from their online profiles are indeed what other people are, and what they present to others online are indeed what they are. Regardless the knowledge learned by the machine which is always a rather true reflection of personalities of a certain person, what people directly see online are not always the truth. These people are living in bubbles created by themselves and believe nothing but the fancy reflections these bubbles made. The horror story is, maybe after my generation, people would believe this is the way to make friends or even connect to people. In a fact, I have came cross people being worried about networking after I told them I got rid of LinkedIn. Since when LinkedIn equals to networking? Indeed it is a networking tool to do some networking, but it is definitely not all of it. I can always meet people for dinner, grab lunch together, sit on the grass and chat with each other. Would’t that be more enjoyable than posting links on each others’ web feed and pretend others have actually read it?

Come on, just ask me out for a lunch or hiking or whatever, and we can chat about who I am and who you are from there.

Besides above, I also feel that there is an extreme gap between the online social world and the reality these days. Specifically, the differentiation of online friends and friends in reality. I have come across people who indicated to others that I am a friend of them with only knowing me through reading the about page I put online. They even have not had a cup of tea with me or knowing what I usually do when I am free. If these people are able to claim I am their friends, I could easily say that I am a friend of Barack Obama since I know his whole life even where he was born. For me, friends are people who spend time together either studying, reading, watching movies, or just hanging out together doing random stuff. However, for many people nowadays, friends are just knowing what each other posted on Instagram. A friendly catching up is no longer finding a time to have dinner together, instead, it is reading Facebook posts. They even forget how to be friends with others. I have met people who wish to make friends with me, and of course I would also like to make new friends, but they are freaked out once they knew that I do not use any social media. They do not know what to do to make friends with others other than following them on social medias. Come on, just ask me out for a lunch or hiking or whatever, and we can chat about who I am and who you are from there.


References: