The streaming giant, which has renewed every single one of its original offerings for at least two seasons, is set to premiere the second season of its talk show foray, Chelsea, beginning on April 14. But whereas the first season of Chelsea, starring comedian and host Chelsea Handler, featured three half-hour episodes hitting Netflix every Wednesday, Thursday, and Friday, the second season will feature one-hour weekly installments, premiering every Friday night.
Chelsea’s second season will comprise 30 episodes — the same amount as the first season, Variety reports. Check out Handler’s quirky announcement video below:
Netflix initially announced that it had renewed Chelsea back in July — despite the fact that, by some accounts, it had gotten off to a shaky start, with no interviews breaking into the cultural zeitgeist and showrunner Bill Wolff departing a mere three weeks after the show’s premiere.
The streaming giant, has, however, placed a stated emphasis on comic programming in recent weeks, and has acquired standup specials from Chris Rock, Dave Chappelle, Amy Schumer, and Sarah Silverman.
Filed under: News by Geoff Weiss Comments Off on Netflix’s ‘Chelsea’ Will Transition To One-Hour Weekly Episodes For Second Season
YouTube creator Hannah Hart has entered a first-look deal with global entertainment goliath Lionsgate, in which Hart, 30, will star in and executive produce feature films. The first project in development, Lionsgate announced today, is a still-untitled LGBTQ romantic comedy.
Lionsgate signed a similar overall deal with Hart’s friend and frequent collaborator Grace Helbig in October for both movie and TV projects. Last year, Lionsgate also acquired Dirty 30, a feature film starring Hart, Helbig, and Mamrie Hart — a comedy trio known as YouTube’s ‘Holy Trinity’. Lionsgate also made a $25 million annual commitment to films and series led by influencers in coming years, with projects to be released via on-demand platforms.
“I’m extremely pleased to be partnering with [Lionsgate] as I continue my journey into film,” Hart said in a statement. “My hope in working together is to bring more inclusivity and diversity into storytelling overall.”
Added Jordan Gilbert, Lionsgate’s VP of digital production: “Having worked with Hannah on several successful projects, we know that her distinctive worldview, comedic timing and dedication to fresh content is what makes her an impressive writer, producer, and star.”
Hart, who counts 5 million followers across all of her social platforms, rose to fame thanks to her YouTube cooking series, My Drunk Kitchen. She has also released two New York Times Best Selling books and has a Food Network series in the works, in which she will travel across the country in search of budget restaurants. Hart is represented by UTA, attorney Ryan Pastorek, and managed by Linnea Toney.
Filed under: Hannah Hart, Lionsgate by Geoff Weiss Comments Off on Hannah Hart Signs First-Look Deal With Lionsgate, Will Star In LGBTQ Feature Film
Welcome to YouTube Millionaires, where we profile channels that have recently crossed the one million subscriber mark. There are channels crossing this threshold every week, and each has a story to tell about YouTube success. Read previous installments of YouTube Millionaires here.
Highlights from top soccer matches (or football matches, as our friends from across the pond say) can be hard to find on YouTube, but for many fans, the only necessary recap comes from 442oons. Dean Stobbart‘s animation channel, partnered with Whistle Sports, provides a nonstop stream of parodies and satirical music videos from the soccer world. Whether Arsenal is blowing another big game or Cristiano Ronaldo is once again showing off his figure, Stobbart will be there with a jokey cartoon at the ready. Here are his answers to our questions about his channel.
Tubefilter: How does it feel to have one million subscribers? What do you have to say to your fans?
Dean Stobbart: It’s still hard to process on a numbers level! Although I’ve worked (and still work) my backside off to make the videos, the fact that over one million people have subscribed to my channel is bonkers. Working hard is never a guarantee that people will like the videos, come back for more, and subscribe, so I’m still so grateful that my subscribers have stuck around. Without them this wouldn’t be my full-time job!
TF: Your channel has grown tremendously in just three-and-a-half years. Why do you think your videos have been so popular?
DS: I think the humour is the main thing — it’s very light-hearted and pokes fun at the ridiculous world of football in an affectionate way. The parody songs are very popular with my followers and definitely my most viewed videos on the whole. A big factor has been the turnaround time. Creating a cartoon overnight based on the previous day’s action helps the videos to be incredibly relevant and more successful as a result.
TF: Considering that animation is known for being time-consuming, how do you manage to put out so much content?
DS: Late nights. Early mornings. No breaks. That helps! I’ve got a similar work ethic to N’golo Kante. Sadly mine requires much more sitting down and far less exercise, but never mind. Thanks to the growth of my channel, I have two animators now which is a huge help and they do they donkey work of animation so I can spend more time scripting, editing and adding the final 2% to their animations.
TF: How do you manage to stay completely up-to-date on football news and results as you create your videos?
DS: I’ve always followed footy closely as I love the game, so it’s just keeping up with what I already used to do. My phone is constantly buzzing, the TV is usually on and showing football, and I’m usually listening to football podcasts. The 442oons fans are great too as they’ll often give me heads-up on breaking stories and the main talking points in a match.
TF: When you’re preparing to add a new player to your animations, how do you decide which of their characteristics to highlight?
DS: I just look at their face or picture them from all the games I’ve watched and think what stands out. Prominent features always help and some of the crazy haircuts instantly make a player more recognizable in a caricature, although keeping up with [Paul] Pogba’s hair is a full-time job in itself.
TF: Have you heard from any of the players you have impersonated? If so, how did they respond?
DS: Thomas Muller is probably the biggest player who has responded to a video. When I created a cartoon called Bayern Mambo No 5 when Bayern beat Arsenal 5-1 (the first time!) he said he can sing far better than my impersonation of him. I challenged him to a cartoon vs human sing-off but sadly, he must have been too scared!
TF: Which character of yours would you say is the hardest to animate and perform?
DS: There are lots of voices that are hard to do, especially as I add more characters. How many Spanish style accents can one person do?!
I typically find the players from London hardest as that’s not my best accent. [John] Terry and [Frank] Lampard are tough – although my Harry Redknapp isn’t too shoddy.
TF: Any plans to do animations for other sports, or are you strictly a football guy?
DS: My love of football far outweighs the other sports, although I follow them loosely. And as football is such a huge sport compared to most others, I’ll stick with that for now.
I’m hoping to start a second channel but with a focus on the equally ridiculous world of celebrities and that surreal culture.
TF: What’s next for your channel? Any fun plans?
DS: Two million subscribers! Mainly I’m trying to keep the match videos going in the usual madcap 442oons style, and until the games are played, I rarely have any plans for what the video will be. The celebrity angle is one that excites me although I don’t want to get too distracted.
Filed under: Articles, Featured, Millionaires, News by Sam Gutelle Comments Off on YouTube Millionaires: 442oons Animates The “Ridiculous World Of Football”
Kjellberg does use small pockets of his response video to apologize to anyone who may have been upset or offended by the anti-Semitic allegations levied against him. About his now-infamous “death to all Jews, subscribe to Keemstar” prank and other indiscretions, he lamented he is a “rookie comedian” who doesn’t always make the right decision.
Most of his video, however, is reserved for criticism of what he called “old-school media,” a group he says “does not like internet personalities because they’re scared of us.” He notes that many outlets focus incessantly on his wealth rather than the content he puts out. “This whole thing is not a post, it was an attack toward me by the media to try to discredit me to try to decrease my influence and my economic worth,” he says. “I made a point that the media takes what I say out of context, they take that and put it out of context to use against me and to portray me as a Nazi.”
Many of Kjellberg’s arguments are the same ones used by his defenders, including Ethan Klein of h3h3productions, and at the end of his video, he makes sure to tearfully thank the people who have supported him to this point. For his detractors, however, he has no kind words. “I’m still here. I’m still making videos. Nice try, Wall Street Journal,” he says. “Try again, motherf**kers.”
Filed under: Articles, Featured, News, PewDiePie by Sam Gutelle Comments Off on PewDiePie Responds To Wall Street Journal With The Middle Finger, Isn’t Going Anywhere
Popular gaming creator — and newly crowned YouTube Red series headliner — Dan ‘DanTDM’ Middleton is bringing his massively successful tour across the pond. Kicking off in Boston on March 24, Middleton will hit 21 U.S. cities, wrapping up in San Francisco in May.
DanTDM On Tour is being produced by Endemol Shine UK, the management agency OP Talent, and live events company Live Nation Entertainment. The family-friendly show features games, puzzles, and appearances from Middleton’s pet pugs as well as other characters from his YouTube channel. When it kicked off in the UK last summer, tickets sold out in a matter of 24 hours, and Middleton also played four sold-out shows at the Sydney Opera House in December.
“My American audience has always supported me and this tour is a brilliant opportunity for me to bring my world to life for them,” Middleton, who counts more than 16.5 million subscribers and nearly 10 billion views across his two YouTube channels, said in a statement. “I can’t wait to travel across the country and meet as many fans as possible.”
Added Tom Greenwood-Mears, head of live events for Endemol Shine UK: “The tour has been really well received in the UK and Sydney, by kids and parents alike, so we are extremely excited to take it to Dan’s largest market, the US.”
Tickets are available here, and you can check out the tour announcement on Middleton’s channel below:
Filed under: News by Geoff Weiss Comments Off on Dan ‘DanTDM’ Middleton Is Bringing His Sold-Out Tour Stateside
YouTube celebrated its 12th birthday this week, but that’s not the only milestone the video site has passed. It has also announced that one billion of its videos now contain captions, giving users who are hard of hearing a wide range of viewing options.
YouTube first gave its users the ability to add captions to videos in 2006, one year after its launch. In 2009, it added to that offering by introducing automated captions. Since then, the video site has accepted more text files from users while also upping the accuracy of its caption AI through machine learning and other processes. These days, automated captions — which are available in ten languages — are more exact than they have ever been, giving creators fewer opportunities to spin them for comedy.
As a result of these widely-proliferated and increasingly-accurate captions, it has been easier for hearing-impaired individuals to become valued members of the YouTube community. Creators like Tyler Oakley have ensured that all their videos are captions, while deaf creators like Rikki Poynter have gathered sizable audiences. In total, YouTube says, captioned videos receive 15 million views per day.
Filed under: Articles, Featured, News, YouTube by Sam Gutelle Comments Off on There Are Now More Than One Billion Captioned Videos On YouTube
A new study from influencer marketing platform Influence.co provides a fascinating glimpse into the kinds of deals that are proliferating across Instagram today, including the fact that the average price for a sponsored post on the platform is roughly $300.
According to the study, men charge more per sponsored post than women. And while the average post sets brands back $300, that number shoots up to $800 for Instagram influencers with more than 100,000 followers, AdWeek reports. Influencers with 1,000 followers or less — though they tend to be more engaged with their audiences — tend to earn around $100 per post.
The most well-paid influencers on Instagram, in terms of price per post, are modeling accounts, per AdWeek, which notes that models make on average $434 per sponsored post. Other lucrative niches include photographers, who make on average $385 per sponsored post, followed by food accounts ($326), pets ($320), fitness ($306), fashion ($217), beauty ($205), travel ($205), music ($201), and lifestyle ($172).
The study also took a look at the kinds of Instagram niches that tend to amass the most followers, with modeling, once again, taking the top spot. The average modeling Instagram account boasts 141,563 followers, followed by fitness creators (131,625), pets (117,259), beauty (83,786), fashion (61,419), food (59,761), photography (53,814), travel (38,274), lifestyle (27,699), and music (26,403).
For additional findings, check out AdWeek’s report right here.
Filed under: Instagram by Geoff Weiss Comments Off on Male Influencers Charge More Than Females For Sponsored Instagram Posts (Report)
As the controversy surrounding Felix ‘PewDiePie’ Kjellberg has reached a fever pitch this week, YouTube’s most-subscribed star — who was dropped by Maker Studios and saw his YouTube Red series axed in the wake of a string of anti-Semitic jokes on his channel — has remained relatively off the radar. But not for long.
Kjellberg returned to Twitter yesterday and acknowledged supporters while vowing to return with new videos soon. He joked that he paid h3h3productions’ Ethan Klein $50,000 to make a video in which Klein defended his friend and accused the media of manufacturing an outraged response to the jokes.
PewDiePie also retweeted a video from the YouTube comedy channel World Of The Orange, which spoofs a video produced by The Wall Street Journal chronicling Kjellberg’s offenses. The World Of The Orange video, however, jokes that not only is Kjellberg guilty of telling anti-Semitic jokes, but also of recruiting child soldiers to fight for warlords in the Democratic Republic of Congo, rigging the U.S. presidential election in favor of Donald Trump, and even cancelling the cult 2003 sci-fi series Firefly.
Though he’s now network-less, it doesn’t look like that’s going to keep Kjellberg away from YouTube for long. Kjellberg did release a (potentially prescheduled) video on Valentine’s Day after the scandal had broke — though it simply featured he and his girlfriend Marzia Bisgonin playing the video game Genital Jousting. Yesterday, Kjellberg tweeted that more videos were on the way.
This paper was written by Paul Covington (currently a Senior Software Engineer at Google), Jay Adams (currently a Software Engineer at Google), and Embre Sargin (currently a Senior Software Engineer at Google) to show other engineers how YouTube uses Deep Neural Networks for Machine Learning. It gets into some pretty technical, high-level stuff, but what this paper ultimately illustrates is how the entire YouTube recommendation algorithm works(!!!). It gives a careful and prudent reader insight into how YouTube’s Browse, Suggested Videos, and Recommended Videos features actually function.
An Engineering Paper On The YouTube Algorithm For Dummies
While it was not necessarily the intent of the authors, it is our belief the Deep Neural paper can be read and interpreted by and for YouTube video publishers. The below is how we (and when I say we, I mean me and my team at my shiny new company Little Monster Media Co.) interpret this paper as a video publisher.
In a previous post I co-wrote here on Tubefilter, Reverse Engineering The YouTube Algorithm, we focused on the primary driver of the algorithm, Watch Time. We looked at the data from our videos on our channel to try to gain insight into how the YouTube algorithm worked. One of the limiting factors to this approach, however, is that it’s coming from a video publisher’s point of view. In an attempt to gain some insight into the YouTube algorithm we asked ourselves and then answered the question, “Why are ourvideos successful?” We were doing our best with the information we had, but our initial premise wasn’t ideal. And while I stand by our findings 100%, the problem with our previous approach is primarily twofold:
Looking at an individual set of channel metrics means there’s a massive blind spot in our data, as we don’t have access to competitive metrics, session metrics, and clickthrough rates.
The YouTube algorithm gives very little weight to video publisher-based metrics. It’s far more concerned with audience and individual-video-based metrics. Or, in laymen’s terms, the algorithm doesn’t really care about the videos you’re posting, but it cares a LOT about the videos you (and everyone else) are watching.
But at the time we wrote our original paper, there had been nothing released from YouTube or Google in years that would shed any light onto the algorithm in a meaningful way. Again, we did what we could with what we had. Fortunately for us though, the paper recently released by Google gives us a glimpse into exactly how the algorithm works and some of its most important metrics. Hopefully this begins to allow us to answer the more poignant question, “Why are videos successful?”
Staring Into The Deep Learning Abyss
The big takeaway from the paper’s introduction is that YouTube is using Deep Learning to power its algorithm. This isn’t exactly news, but it’s a confirmation of what many have believed for some time. The authors make the reveal in their intro:
In this paper we will focus on the immense impact deep learning has recently had on the YouTube video recommendations system….In conjugation with other product areas across Google, YouTube has undergone a fundamental paradigm shift towards using deep learning as a general-purpose solution for nearly all learning problems.
What this means is that with an increasing likelihood there’s going to be no humans actually making algorithmic tweaks, measuring those tweaks, and then implementing those tweaks across the world’s largest video sharing site. The algorithm is ingesting data in real time, ranking videos, and then providing recommendations based on those rankings. So, when YouTube claims they can’t really say why the algorithm does what it does, they probably mean that very literally.
The Two Neural Networks
The paper begins by laying out the basic structure of the algorithm. This is the author’s first illustration:
Essentially there are two large filters, with varying inputs. The authors write:
The system is comprised of two neural networks: one for candidate generation and one for ranking.
These two filters and their inputs essentially decide every video a viewer sees in YouTube’s Suggested Videos, Recommend Videos, and Browse features.
The first filter is Candidate Generation. The paper states this is determined by “the user’s YouTube activity history,” which can be read as the user’s Watch History and Watch Time. Candidate Generation is also determined by what other similar viewers have watched, which the authors refer to as Collaborative Filtering. This algorithm decides who’s a similar viewer through “coarse features such as IDs of video watches, search query tokens, and demographics”.
To boil this down, in order for a video to be one of the “hundreds” of videos that makes it through first filter of Candidate Generation, that video must be relevant to the user’s Watch History and it must also be a video that similar viewers have watched.
The second filter is the Ranking filter. The paper goes into a lot of depth around the Ranking Filter and cites a few meaningful factors of which it’s composed. The Ranking filter, the authors write, ranks videos by:
…assigning a score to each video according to a desired objective function using a rich set of features describing the video and user. The highest scoring videos are presented to the user, ranked by their score.
Since Watch Time is the top objective of YouTube for viewers, we have to assume it’s the “desired objective function” referenced. Therefore, the score is based on how well a video, given the various user inputs, is going to be at generating Watch Time. But, unfortunately, it’s not quite that simple. The authors reveal there’s a lot more that goes into the algorithm’s calculus.
We typically use hundreds of features in our ranking models.
How the algorithm ranks videos is where the math gets really complex. The paper also isn’t explicit about the hundreds of factors considered in the ranking models, nor how those factors are weighted. It does cite the three elements mentioned in the Candidate Generation filter, however, (which are Watch History, Search History, and Demographic Inforomation) and several others including “freshness”:
Many hours worth of videos are uploaded each second to YouTube. Recommending this recently uploaded (“fresh”) content is extremely important for YouTube as a product. We consistently observe that users prefer fresh content, though not at the expense of relevance.
One interesting wrinkle the paper notes is that the algorithm isn’t necessarily influenced by the very last thing you watched (unless you have a very limited history). The authors write:
We “rollback” a user’s history by choosing a random watch and only input actions the user took before the held-out label watch.
For example, a user may watch a given video with high probability generally but is unlikely to click on the specific homepage impression due to the choice of thumbnail image….Our final ranking objective is constantly being tuned based on live A/B testing results but is generally a simple function of expected watch time per impression.
It’s not a surprise clickthrough rates are called out here. In order to generate Watch Time a video has to get someone to watch it in the first place, and the most surefire way to do that is with a great thumbnail and a great title. This gives credence to many creator’s claims that clickthrough rate are extremely important to a video’s ranking within the algorithm.
Ranking by click-through rate often promotes deceptive videos that the user does not complete (“clickbait”) whereas watch time better captures engagement [13, 25].
While this might seem encouraging, the authors go on to write:
If a user was recently recommended a video but did not watch it then the model will naturally demote this impression on the next page load.
These statements support the idea that if viewers are not clicking a certain video, the algorithm will stop serving that video to similar viewers. There is evidence in this paper that this happens at the channel as well. It states (with my added emphasis):
We observe that the most important signals are those that describe a user’s previous interaction with the item itself and other similar items… As an example, consider the user’s past history with the channel that uploaded the video being scored – how many videos has the user watched from this channel? When was the last time the user watched a video on this topic? These continuous features describing past user actions on related items are particularly powerful…
In addition, the paper notes all YouTube watch sessions are considered when training the algorithm, including those that are not part of the algorithm’s recommendations:
Training examples are generated from all YouTube watches (even those embedded on other sites) rather than just watches on the recommendations we produce. Otherwise, it would be very difficult for new content to surface and the recommender would be overly biased towards exploitation. If users are discovering videos through means other than our recommendations, we want to be able to quickly propagate this discovery to others via collaborative filtering.
Ultimately though, it all comes back to Watch Time for the algorithm. As we saw at the beginning of the paper when it stated the algorithm is designed to meet a “desired objective function,” the authors conclude with “Our Goal is to predict expectedwatch time,” and “Our final ranking objective is constantly being tuned based on live A/B testing results but is generally a simple function of expected watch time per impression.”
This confirms, once again, that Watch Time is what all of the factors that go into the algorithm are designed to create and prolong. The algorithm is weighted to encourage the greatest amount of time on site and longer watch sessions.
To Recap
That’s a lot to take in. Let’s quickly review.
YouTube uses three primary viewer factors to choose which videos to promote. These inputs are Watch History, Search History, and Demographic Information.
There are two filters a video must get through in order to be promoted by way of YouTube’s Browse, Suggested Videos, and Recommended Videos features:
Candidate Generation Filter
Ranking Filter
The Ranking Filter uses the viewer inputs, as well as other factors such as “Freshness” and Clickthrough Rates.
The promotional algorithm is designed to continually increase watch time on site by continually A/B testing videos and then feeding that data back into the neural networks, so that YouTube can promote videos that lead to longer viewing sessions.
Still Confused? Here’s An Example.
To help explain how this works, let’s look at an example of the system in action.
Josh really likes YouTube. He has a YouTube account and everything! He’s already logged into YouTube when he visits the site one day. And when he does, YouTube assigns three “tokens” to Josh’s YouTube browsing sessions. These three tokens are given to Josh behind the scenes. He doesn’t even know about them! They’re his Watch History, Search History, and Demographic Information.
Now is where the Candidate Generation filter comes into play. YouTube takes the value of those “tokens” and combines it with the Watch History of viewers who like to watch the same kind of stuff Josh likes to watch. What’s left over is hundreds of videos that Josh might be interested in viewing, filtered out from the millions and millions of videos on YouTube.
Next, these hundreds of videos are ranked based on their relevancy to Josh. The algorithm asks and answers the following questions in fractions of a second: How likely is it that Josh will watch the video? How likely is it the video will lead to Josh spending a lot of time on YouTube? How fresh is the video? How has Josh recently interacted with YouTube? Plus hundreds of other questions!
The top ranked videos are then served to to Josh in YouTube’s Browse, Suggested Videos, and Recommended Videos features. And Josh’s decision on what to watch (and what not watch) is sent back into the Neural Network so the algorithm can use that data for future viewers. Videos that get clicked, and keep the user watching for long periods of time, continue to be served. Those that don’t get clicks may not make it through the Candidate Generation filter the next time Josh (or a viewer like Josh) visits the site.
Conclusion
Deep Neural Networks for YouTube Recommendations is a fascinating read. It’s the first real glimpse into the algorithm, directly from source(!!!), that we’ve seen in a very long time. I hope we continue to see more papers like it so publishers can make better choices about what content they create for the platform. And that’s ultimately why I write these blogs in the first place. Making content suited for the platform means creators will generate more views, and therefore more revenue, which ultimately means we can make more and better programming and provide more entertainment for the billions of viewers who rack up significant Watch Time on YouTube each and every month.
Matt Gielen is the founder and CEO of Little Monster Media Co., a video agency specializing in audience development on YouTube and Facebook. Founded in the summer of 2016 Little Monster has already helped dozens of clients big and small grow their audiences. Formerly, Matt was Vice President of Programming and Audience Development at Frederator Networks where he oversaw the building of the audiences for Cartoon Hangover, Channel Frederator and the Channel Frederator Network.
