Billions of Views
You don't get to 500 million friends without making a few enemies
And another consulting story — one can never really get enough of ‘em. This one’s from the Summer of 2017, where I worked for the largest syndicated YouTube edutainment network in Asia. Here, I go into explicit detail on my roles and responsibilities at the media empire and how I built tools to manage and expand their already incredible reach, in various… unorthodox ways. I’ll write about my consulting misadventures in New Delhi (Restaurant), Florida (BioPharm), San Francisco (Tech), Boston (Real Estate), and LA (Marijuana) at a later time.
If you’re looking for something more verbose read on.
This was my second time flying for work to Mumbai, the first instance in 2015 when I was seventeen — right after the culmination of my high school board examinations. I’d already received my acceptance letter from NYU and was itching to take advantage of the long summer break. I wanted to make some decent dough to blow in New York.
My Q215 clients had insisted on handholding and onsite work, so once they arranged the itinerary — and after I had handed in the final CBSE paper — I jetted off to the city I’d spent eight years of my preadolescence in. I was apprehensive because I had never left Delhi for work (just yet); nerves hit me while preboarding. All my previous engagements were either in person in the National Capital Region (a fancier way of saying New Delhi and some surrounding cities) or on Skype (for foreign clients). This was all about to change — this also paved the way for my love of flying and hoarding frequent flyer miles. I settled into the two-hour flight with Objective-C for Absolute Beginners: iPhone, iPad, and Mac Programming Made Easy by Gary Bennett and counted down the minutes till I arrived in my childhood city. Cathedral School, Marine Drive, Pedder Road. Unfortunately, I was only in town for a weekend which meant I couldn't revisit my favorite spots for some old school reminiscing or quick bites.
Overall, I had a great experience. I felt comfortable and in control because they had been referred to me by family friends. I planned out the requirements in the first few hours of touchdown and drew out a potential development roadmap they could follow, while also architecting a nascent data model. My primary objective was to build their website, a platform where potential customers could receive personalized shopping recommendations, and receive suggestions. Everything was clear-cut, well-defined, and I didn’t mind going above and beyond to please my clients.
This didn’t exactly hold the second time around.
Two years later, I was finally back in the City of Dreams. But why was I still anxious? Maybe because this time my employers had reached out by unscrupulous means. At the time I was an infrequent visitor of BlackHatWorld,
an internet forum focused on black-hat search engine optimization techniques and services, often known as spamdexing.
Imagine a toxic community revolving around script-kiddy technical nonsense, all hat no cattle. They had come across my profile after seeing some figures I’d shared on their message boards — for the two dozen or so sites I managed at the time, I was holding top organic positions in SERPS for several invaluable, high-value keywords. For a few good weeks, if you searched for ‘best web developers in Delhi’, my company website cropped up. I knew what I was doing, and was good at it. So they reached out, we negotiated a reasonable price for my services and I got to work. But they wanted me to fly to their headquarters in Mumbai. I was reluctant because I prefer to fly under the radar when I’m operating in murky waters. I didn’t care that they might be inviting me so they could understand the processes and potentially cut me out for their next ‘advertising’ cycle, I just preferred to stay at home. I’d also just got into a long-term relationship (London/ New York) and I found it tragically comedic that the one summer we’d get to spend time in the same state I was leaving.
They insisted, and so I left my haven and departed. Unfortunately, the unnerving feeling didn’t go away this time around — even after performing the usual background check on the company. So naive.
Conglomerate and Half
Wait, what exactly did the company do?
Have you ever come across any nursery rhyme video on YouTube which has an outrageous number of views, but on listening in, just sounds odd? And you can’t quite seem to put a finger on why? According to The Dangers of YouTube For Young Children, an Atlantic feature written by Alexis C. Madrigal, the company I worked for
ingests Anglo-American nursery rhymes and holidays, and produces new versions with subcontinental flair.
Think Johny Johny Yes Papa, but narrated by an Indian. Now one might scoff, but that single video (at the time of this publication) has garnered more than 4.3 billion views. A 100-second spot, pushed out in two weeks by two-odd animators, managed to secure their position as the 20th most-watched YouTube channel with only 393 videos to their name. 2nd in the education category by subscribers. They were the third largest YouTube channel in India, behind T-Series (yes, that one) and Sony Entertainment Television. Two media powerhouses and one nursery rhyme channel held the podium.
And this was only one of their channels. They also claim to compete with other big names and traditional competitors in the toddler-education-space such as Sesame Street, Disney, and Nickelodeon. And I can attest to this gravitas: I was present when the founder rejected an acquisition offer upwards of tens of millions of dollars. There were numerous articles about them, featuring their exponential growth and awe-inspiring numbers.
They had 200+ employees spread across two floors in Mumbai’s most sought-after commercial real estate building, Indiabulls Finance Center (IFC), and an upcoming property in the suburbs, AND their previous headquarters in Chennai.
The first floor of IFC housed the artists, in a dimly lit space with all blinds closed (I was told this was so they could focus on animating), and the top floor accommodated management, data crunchers, and miscellaneous employees. The walls were lined with the numerous accolades they received for passing 100K, 1MM, and 10MM subscribers.
They were definitely doing well. During my background checks, I’d done my research, but I clearly underestimated the scale of their operations. So why was I put off?
If YouTube thought this company was a shining pillar of the community, why was I hesitant to put my best foot forward? Even Madrigal, the author from the previously linked Atlantic writeup, figured something was up. He changed the title from Raised by YouTube to the current iteration to accurately reflect his stance. Interestingly enough, the article isn’t proudly displayed on their website compared to other (less reputable) news sources, even though they specifically invited Madrigal to Chennai! First, a brief on some of the projects I tackled during my three-month tenure, along with some roles and responsibilities, with my comments interspersed:
A script that leverages the YouTube v3 API, categorizing, tagging, and updating video titles + descriptions, while maintaining parity across all channels.
This was a trivial task, involving a python script that accepted user inputs and then went and updated the already uploaded video’s metadata to accurately reflect new changes. A solid use-case would be if one wanted to change all social media links across all their video descriptions.
Fairly normal ask.
Technical Product Manager/ middleman for the app development company to ship their AVOD/ SVOD streaming app in time.
This was a hassle — the contracted developers working on the project were incompetent and constantly missed deadlines. The app was still in beta by the time I left. I was invited to the Google Analytics dashboard post-launch, and I’ve some statistics shared below:
Still in the scope of my ‘contract’, and relatively harmless.
Injecting a customizable smart app banner for iOS and Android onto the website
Keyword stuffing, Black Hat Search Engine Optimization, Cloaking, Spam Comments, Backlink Automation, and ‘Competitor Analysis’
Disclaimer: I do not support penetration testing into national systems or any systems for that matter without authorized permission. This was purely an academic and educational exercise.
This was the main reason I was invited to Mumbai — everything else led up to this project. They tasked me to create a dashboard that displayed a competitor overview, relevant information, and whatnot, and the keywords they were using to bump their views. Once we knew the path, we generated backlinks along with their keywords across hundreds of pages back to our uploaded videos, essentially spamming numerous blogs concurrently. At its peak, this program was creating ~3 comments per second across many websites. I doubt this played a part in bumping up video relevance, but they told me to run so I ran. They had previously tried this on YouTube itself (when attempting to automate the removal of certain distasteful criticism) but ended up receiving a stern email and a warning from Support. Now they just disable the comments section entirely, but the like/ dislike ratio is a good indication of audience sentiment.
I would prefer not to get into the specifics of the blackhat SEO shenanigans I partook in, but picture a network of bots that wrote and rewrote articles about the aforementioned company to bump up their position on the coveted search results page. Across the many domains they owned, content flowed between them, and paragraphs were posted on the many blogs to increase ‘site value’. I also created a rudimentary crawler which checked to see the content Google was indexing, and take appropriate action. Among other asks, they also, on multiple occasions, requested me to use their seemingly unlimited bandwidth to create a botnet to DDoS their competitor’s websites, hiding behind the ‘stress testing’ veil. And this would have worked — they easily had a LAN of over 300 PCs, not including the render machines which were hosted offsite. I vehemently and immediately put my foot down.
And I’m really glad I did. A few months prior to this stint, I had the unfortunate chance of encountering a Rutgers University who boasted of using the feared Mirai botnet to launch cyberattacks against a large number of business websites (Sony), some odd-Minecraft servers (an extremely profitable strategy), and his university (to get his pick of classes). He showed off a lot on HackForums. In September 2018, Krebs on Security reported the U.S. District Court of New Jersey sentenced 22-year-old Paras Jha to six months of house arrest and ordered him to pay USD 8.6 million in damages.
All the crypto he gained from his malicious endeavors ended up paying for his lawyer bills.
Simple, Sunny, Consistent?
Again, heavily quoting the meticulously planned and well-written Atlantic article:
Many observers respond to their unexpected success by implying that the company has somehow gamed the system. ‘Whenever we [my client] go to the U.S.’, people say, ‘You guys cracked the algorithm.’ But we didn’t do anything. ‘The algorithm thing is a complete myth.’
‘We do not employ the weird keyword-stuffed titles used by lower-rent YouTube channels.’ The company’s titles are simple, sunny, consistent. Its theory of media is that good stuff wins, which is why its videos have won.
Laughable. But I was getting paid, why the backlash? When did I suddenly grow a conscience? A 2019 analysis by Pew Research Center found that videos including or featuring children younger than 13 get three times as many views as others.
My client’s target demographic was consuming ‘edutainment’ content at immense speeds, regardless of what they might watch, and the only barrier to entry was parents opting into their children-friendly service. But as Pew reported:
YouTube explicitly states that the platform is not intended for children under the age of 13. YouTube provides a YouTube Kids platform with enhanced parental controls and curated video playlists, but the analysis in this report focuses on YouTube as a whole.
In another study, a team of pediatricians at Einstein Medical Center, in Philadelphia, found that YouTube was popular among children under the age of 2. 96.6% of the kids in the study had used a mobile. Young children in an urban, low-income, minority community had almost universal exposure and most had their own device by 4. And that was in 2015. But why was this a problem? Madrigal had some strong opinions:
The new children’s media look nothing like what adults would have expected. They are exuberant, cheap, weird, and multicultural. YouTube’s content for young kids—what I think of as Toddler YouTube—is a mishmash, a bricolage, a trash fire, an explosion of creativity. It’s a largely unregulated, data-driven grab for toddlers’ attention, and, as we’ve seen with the rest of social media, its ramifications may be deeper and wider than you’d initially think.
Madrigal also sought out Colleen Russo Johnson, a co-director of UCLA’s Center for Scholars & Storytellers. Johnson did her doctoral work on kids’ media and serves as a consultant to studios that produce children’s programming. Her answer was simple:
‘Bright lights, extraneous elements, and faster pacing.’ In one of the videos I had her watch, a little boy dances flanked by two cows on a stage. A crowd waves its hands in the foreground. Lights flash and stars spin in the background. The boy and the cows perform ‘Head, Shoulders, Knees, and Toes,’ and as they do, the dance floor lights up à la Saturday Night Fever. Johnson told me all that movement risks distracting kids from any educational work the videos might do.
For kids to have the best chance of learning from a video, Johnson told me, it must unfold slowly, the way a book does when it is read to a child. ‘Calmer, slower-paced videos with less distracting features are more effective for younger children,’ she said. ‘This also allows the video to focus attention on the relevant visuals for the song, thus aiding in comprehension.’
But even in relatively limited doses, these videos can affect young toddlers’ development. If kids watch a lot of fast-paced videos, they come to expect that that is how videos should work, which could make other educational videos less compelling and effective. ‘If kids get used to all the crazy, distracting, superfluous visual movement, then they may start requiring that to hold their attention,’ Johnson says.
Generation Z was being subjected to the same low-quality repetitive bullshit, repeatedly. Let’s revisit their most popular video, Johny Johny Yes Papa:
(as of November 27, 2020),
and the video running
1 minute, 40 seconds long
assuming an average watch duration of
The future leaders of our free world had watched approximately 4098.6 years’ worth of garbage. While Alphabet and in turn Google has not released an official threshold, several third parties reported that ‘views’ do not appear to be counted unless a user (or child) watches at least 30 seconds of the respective video.
The tools I was creating were directly impacting and in turn adversely affecting millions of children. I was disillusioned and distraught. My 19-year-old naive self wanted to make a difference in the world, a grass-roots change, and in the worst sense of the definition, I definitely did.
PBS to Oranges
It’s not all bad. My client’s supposed old-school competitors (PBS, Sesame) were subjected to many regulations and government scrutiny before they created TV content suitable for children. Reformers like the Action for Children’s Television advocated for years to tweak their programming to make sure of this.
Among the specific findings, researchers demonstrated that Sesame Street improved children’s vocabulary, regardless of their parents’ education or attitudes. Another study found that regular adult TV stunted vocabulary development, while high-quality educational programs accelerated language acquisition.
You didn’t need flashing lights to educate, inspire, and connect with children. But the ubiquitousness of the Internet paved way for a ‘new kid on the block’: one that could mass-manufacture clickbait, churn drivel, and use ‘correct tags’ by successfully reverse-engineering whatever someone might enter into the YouTube search box to hit the oh-so-important KPIs like increased retention rate, watch time, and subscriber count. Profit was the end all be all. There wasn’t any novel content no more.
My client’s executives and animators weren’t working with academics who could help them shape their content to promote the healthy development of young brains — instead, they were using analytics and data points to fine-tune content on what ‘works’ and quickly pivot if the goalpost shifted.
It quickly takes a dark turn when you compare content on both platforms. Television has a quality and decency barrier. YouTube videos today are made by randoms with a minimal amount of moderation. The difference is a kid’s TV show in the 90s wouldn’t use an image of a woman dressed as Elsa cutting off her tongue with real scissors to try and clickbait children into watching it. You might ask what I’m talking about, but how can one forget ElsaGate? From the subreddit’s description:
#ElsaGate refers to a wave of videos being produced by different channels, containing pop culture characters (typically Western) that are shown doing bizarre and usually violent/ sexual acts. The reasoning for the existence of these varies. For example — some believe, controversially, that it is a form of communication used by child pornographers; others believe that it is to appeal to children for maximum ad revenue. Others believe that it started as a project for maximum revenue, and has now derailed into a bizarre competition. The majority of the live-action videos are made in Russia and the animations in India.
You slowly start realizing YouTube might have a bigger part to play in this mess.
Rage Against The Machine
If the above tweet doesn’t load (likely a shadowbanned user), it was a screenshot surveying children among the age of 8 to 12 on what they wanted to be when they grew up. From the Harris Poll, funded by LEGO, kids from China are more likely to hope to be an astronaut (56%) than kids from the US or UK (11% in both). In contrast, Western counterparts prefer to be YouTuber’s or Vloggers. Now, this could be attributed to Google’s ban in China, but I do think the survey was a classic misdirection at best, disinformation at worst. Another way to read into this: an astronaut is the 5th most popular profession for kids in the USA and the UK today.*
* out of the 5 professions provided in the multiple-choice question poll
But it still does provide some food for thought.
Money Money Money
As Theodore Porter — the great historian of science and technology — put it in his book Trust in Numbers,
Quantification is a way of making decisions without seeming to decide.
I base most, if not all of my decisions on data. I can attest to this statement by linking to the various articles I have on probability, quant, trading, and chess. And YouTube does so too. But from the above example it is clear that if you torture the data long enough, it will say whatever you want.
Let’s try an experiment. Go to the front page of YouTube on your incognito browser and tell me what you see. Corporations, Millionaires, Product Reviews, Music Recommendations, and Movie Trailers. Ah, really emphasizing the ‘you’ part of YouTube. The most creative people have fallen into irrelevancy, whether it is because of algorithm changes to prioritize watch time, the demonization debacle making some of the most opinionated creators upload less or censor themselves entirely, or vague policies banning people for not conforming to the sites new family-friendly direction.
Last year, YouTube got into hot water for rolling out P-Scores, an internal rating given by Google to channels to rate them based on five key metrics, which are:
Popularity: Watch time. Also, longer videos rank better.
Passion: Leans more on audience engagement. Devoted fanbases who like, comment, and subscribe will propel their respective channel.
Protection: Content’s advertiser friendliness. No swearing, sexual content, or violence, and you’re in the green.
Platform: Reviews the platforms on which the content is being watched, for instance, on phones, TV, or other devices or websites. Beneficial to content that is frequently watched on larger screens, primarily TV screens. Seems to target growing audiences.
Production: Focuses on high production value. Industry-grade editing, professionally shot cinematic masterpieces will succeed here.
Combine these metrics and the algorithm generates a grade to determine what kind of advertisements could appear on your video. The higher the score, the better the ads, the more money you make. Ratings seem to be extremely important not only to an individual video’s popularity but also to the overall channel success. The issue, however, is that creators have no way of determining or identifying the individual values for each of these benchmarks. I also wonder how a proprietary algorithm can determine the value of a video uploaded based on these particular features.
When you look at how they’ve structured this, you’ll realize that YouTube has been internally judging creators’ videos on this scale for a while now without telling them, once again not giving homegrown creators a chance to play fairly.
Around the same time, three YouTubers reverse-engineered the P-score that judged a channel’s advertising worth based on the content they created. They documented over two hundred YouTubers P Scores, and the only channels with over a thousand are news stations or talk shows. The same channels you saw on the incognito homepage.
Creators like PewDePie (the 2nd most subscribed channel) ranked sub-1000. Casey Neistat, a highschool dropout but prolific YouTuber scored 789. And another egregious example. In 2017, Neistat posted mentioning all ad revenue from his video would go to victims of the Vegas shootings. One can only guess what happened.
His score suddenly made a lot of sense. YouTube replied:
If these were the rules, I can respect the stance. But the very same day, the Late Night brigaded and uploaded a video titled Jimmy Kimmel on Mass Shooting in Las Vegas and ads were present. This dichotomy and preference for television networks mean that P-Scores are also manually edited and coded by people at YouTube HQ. No matter what you do, it seems like if you have enough green, you can swing the ratings your way.
I digress: By using P-Score, YouTube is forcing creators to create more cookie-cutter, standardized content, which follows the rules and the book to the tee, favoring brand-appropriate content over videos that might have a more personal touch. And the data-driven edutainment animated shorts that my clients were producing fit this bucket.
What does this say about the content Google is promoting?
Let’s ignore the red flags and assume the content they’re creating benefits children. I read through some of the comments defending my client:
My youngest son has been watching [my client’s videos] and a year ago he could count to 10 in English, and knew the name of every color in English as well. Most 2-year-olds can do that, but not in a foreign language, so for us, it has been educational.
But then I came across an entirely different beast:
My kids watch this stuff constantly and it's just too much and too random to vet it all. I’m putting a lot of trust in YouTube Kids to filter it to quality content.
That trust was shaken the other day when they said ‘daddy did you know you shouldn't sleep under trees because you can't breathe properly?’ I’m like who told you that. ‘A video.’
Turns out it was a ‘learning’ video produced by one of these outfits, not [my client] but a competitor AumSum. It's some weird urban myth that has traction in South Asia. Compare the locations of the authors with pro-myth vs anti-myth answers. It’s a bastardization of tree respiration biology. They think it puts out enough CO₂ at night to impact humans. So don’t sleep near it because then you’ll be oxygen-deprived.
Apparently, someone green-lit that stinker and turned it into a cute animation. This is funny until you realize AumSum as a whole has 139MM views. Their singular mission is to make educational science videos. And my kids are 100% addicted to YouTube Kids.
From AumSum’s About Us section:
We try our level best to create highly creative and refreshing videos of Physics, Chemistry, and Biology. This channel can prove to be useful for students studying in schools, colleges as well as for people of all ages who have a curious scientific mind.
It’s now been edited to a single-line sentence:
We try our level best to create highly creative and refreshing videos of Science.
Think of the Children
And for the explanation of billions of views for these mindless videos? I’ll pass the torch to Calvin & Hobbes (from their 28/12/1998 strip) to assist:
Young children actually prefer watching the same episode(s) over and over again, because it’s rewarding when their expectations (of what happens next) get fulfilled. Dr. Jessica Horst of the University of Sussex, in Get the Story Straight: Contextual Repetition Promotes Word Learning from Storybooks, conducted a study where children were divided into two groups. One read the same book with new vocabulary multiple times. The other read a variety of books with new vocabulary. The children were then tested, and those who had been exposed to the same story over and over demonstrated understanding of many new words, while the control group retained very little new information.
Young people assimilate these stories as a pattern. They are chains of meaning that can decipher better and better. And of course, if their parent is present while they’re experiencing this Brave New World, they get used to the comfort.
Though having a big library at home is a great thing, Dr. Horst says, ‘It’s not the number of books, but the repetition of each book that leads to greater learning. We know that children who watch the same TV program over and over again do better in comprehension tests afterward.’
It’s actually exciting to watch the hero fight the villain; it’s really sad when tragedies happen; if anything, knowing the ending makes it better because they can invest more safely or securely in the emotion of the moment. I may be older now but I still remember those feelings. My client’s library is low effort content. I mean: it’s really quite terrible and a waste of your kid’s time. The only reason they have views is that they know how to game the system. Parents search for a popular nursery rhyme and leave kids with auto-play or suggestions, and channels like my client know how to game the recommendation algorithms to keep the kids within their content.
You might say I’m racist. There are other channels in the toddler entertainment business, from other countries, which do the same thing. I blacklist them as well. There are good Indian educational channels on youtube. My client is not one of them.
Another feather in India’s proverbial edutainment cap is the story of WhiteHatJr, an ed-tech startup that was acquired for a USD $300MM all-cash deal in August 2020.
The story might seem rosy, but numerous articles stating that people who have criticized the kids coding startup’s methods or marketing campaigns have found their posts swiftly wiped off the internet. Experts say this could be corporate censorship, which points to a larger debate around transparency in content moderation. This post is long enough as it is, so I’ve linked to some of the more engaging discussions below:
Even Netflix needs some restrictions and regulations. With autoplay on these services, it’s not like they have the interests of us or our kids at heart. YouTube is designed to tempt you down the rabbit hole. They want to maximize the amount of time you spend on their website. Do we go back to the reform-regulated PBS and Sesame Street then? A lot of the DVD collections of Sesame Street now have warnings that mention their content might be dated and not suitable for small children anymore. Content that once seemed ‘timeless’ to people who grew up with it will seem wildly out of place now. One can find higher quality video entertainment for children. Maybe the newer more-modern PBS Kids Video and Games apps. Another option would be to find them something useful or interesting to do that doesn’t involve being a stationary recipient of an audio-video feed. So what choices do children have? How can we distract them?
Older teen kids do seem to think they’re entitled to continuous digital mental stimulation via passive consumption — I also wonder this instant gratification robs them of the experience of ever being bored. And if that might influence them. Here:
We regularly make sure they get experience doing nothing too. My wife and I are big believers in the importance of doing nothing, for imagination/ creativity, and also just because it’s an important skill to have in life, to be able to sit and wait.
It’s a self-fulfilling prophecy, the more constantly stimulated the kid is, the more difficult it is for them to pay attention when it’s important.
Does the onus then fall on parents? How do one juggle a job, hobbies, engagements, as well as the responsibility of curating the content your child is watching? Is this indicative of an Orwellian controlling future for our progeny?
In yesteryears, television was used as a tool by our parents to keep us occupied, and in this current day and age, YouTube serves the same purpose. It is downright impossible to control what you’re kids are doing online, I myself was browsing weird forums at 15. What you can do is be a mindful and engaged parent and make the content they watch more actionable. If they end up watching a mindless Vox video on Why no aquarium has a great white shark ask them the right questions:
if there could potentially be suitable places to house sharks in the near future,
how would you calculate the resources required to hold one in an oceanariums,
and the environmental and governmental bodies one would have to go through to get the green light.
They should attempt to see the larger picture and connect the dots. Make them draw a plan and a roadmap. Figure out all the edge cases. Or go the Pacifist route, and task them to create a protest worthy of banning shark captivity forever. Or even focus on the animals themselves, and construct and design an elaborate tank. Don’t ever let them slide — a rolling stone gathers no moss. I personally feel like the best approach is moderation for everything, but hey, people who don’t watch media probably have more interesting hobbies than I do. And would probably harass their kids less.
The heavily quoted Atlantic writeup Raised by YouTube by Alexis C. Madrigal
In 2018, writer and artist James Bridle gave a TED talk titled The nightmare videos of children's YouTube — what's wrong with the internet today
Miscellaneous DDoS Scripts, Retrieved from GitHub
YouTube’s P-Scores, Video Throttling, and More: A Comprehensive Guide by Bowblax, Nicholas DeOrio, Optimus, Pescatore
All the WhiteHatJr links provided above, separated by article and Y Combinator Comments.