YouTube monetization analysis | Hacker News Reader

YouTube monetization analysis

docs.google.com

354 pointsm0ck158 comments

I am almost certain another factor is clickthrough from "undesirable" referral links. ReviewBrah [0] (The Report of The Week) is the most family friendly channel with high subscribers that I can think of. He reviews fast food and has a unique fashion style and personality. Very clean.

He has become a popular meme and is regularly linked to from 4chan. I struggle to find a single thing that could warrant demonetisation (having looked through the youtube guidance), yet a large chunk of his videos are automatically demonetised.

[0] https://www.youtube.com/channel/UCeR0n8d3ShTn_yrMhpwyE1Q

Of course, this would mean that YT has handed 4chan a laser cannon to demonetize videos at their discretion.

Someone will read these comments and it will be posted on 4chan within a few minutes, if it hasn't been already.

We might as well start posting suggestions for targets right away.

So you're telling me there's a way we might demonetize "kinder playtime", "hailey's magical playhouse", "funtoys", and "toys unlimited"?

THAT is a noble cause! Gimme the sixty-five. I'm on the job!

Could you elaborate on what these channels are and why you would want them demonetized? Are their videos actually sponsored by Disney et.al.?

https://gizmodo.com/youtubes-creepy-kid-problem-was-worse-th...

Basically, it ranges from use of copyrighted characters, attempts to game the autofill system, said characters doing weird and non-child-friendly shit, and actual exploitation of children.

MK ultra training.

There's a community here talking about this issue: https://www.reddit.com/r/ElsaGate/

The only other factor I can think of with ReviewBrah is that he's often negatively reviewing products from huge companies (fast food chains) that might advertise on YouTube. Would Burger King want to run an ad before he reviews chicken fries or whatever?

There are a huge number of other food reviewers that don't have the same issues with demonitization, even if they have slightly less ad-friendly content. That said, it's a good suggestion; it's much easier to find a way to shoot down your suggestion than it is to think of it.

> it's much easier to find a way to shoot down your suggestion than it is to think of it.

words of wisdom, I would do well to remember them.

Criticising products is totally fine, otherwise all video game or movie reviewers would be demonized. Demonetization is for "upsetting" content from a brand perspective (terrorism, catastrophe, sex, violence, ...), but the models get it wrong from time to time.

If it were the case, would it be an anti-trust issue?

Or it's possible that people from those sources report his videos.

But my own videos get demonetized almost instantly and none of mine have weird referral links as far as I can tell.

Good point, it very well could be. However, I'm inclined to think that this isn't the case, solely from how much positive feedback he gets, as well as how little negative comments he gets. I mean, finding a negative comment about ReviewBrah on any of the Chans is nigh impossible.

Predicting the behavior of 4chan seems nigh impossible to me.

I think you're over-generalizing from one example. I follow a few differnt channels that are totally harmless and boring, and they all complain about demonitization.

It seems like the algorithm just defaults strongly towards demonitization. Maybe it will learn and get better, but at the moment lots of things are getting demonitized.

To me, it sounds like what's really happening is that the amount of content has vastly outstripped the amount of available, ahem, monetization.

I want to believe, but then compare it with actual TV...

Another hypothesis I've heard that would explain the same effect is that youtube groups uploaders with the preferences of their viewers. If someone has lots of viewers who also watch videos of undesirable topics, that uploader will be considered undesirable as well.

Is there a name for this type of "guilty by association" algorithm?

K-means clustering.

This is one popular clustering algorithm. I think "clustering algorithms in general" would be a more accurate answer.

I didn't know there were others. I heard about this from a journal entry somewhere on DeviantArt about 10 years ago.

The algorithm does not have viewer's other watched videos as an input. Perhaps it should, but it does not.

I became familiar with ReviewBrah a couple weeks ago when browsing broadcasts on my shortwave radio and found his radio show. He sounded like an old-timey broadcaster and I imagined him to be a Jewish gentlemen and in his 50s.

Every single video, without exception, that I've uploaded in the past few months has been demonetized within a day or two. Then I request manual review and they're monetized again within another day or two.

My content is clean and all of it is programming/tech related (programming tutorials and some Raspberry Pi stuff).

So if there is an algorithm it's really naive at this stage or it doesn't apply until you request manual review.

In my experience, Youtube demonetizes based on metadata. So any tags/descriptions with keywords that YT views as “controversial” the system will demonetize regardless of actual content in the video.

I uploaded a video yesterday that's already demonetized. It's part of a series to write an RSS feed reader in Python (specifically the part about adding static files to the server).

Tags: python, rss, feedparser, feed reader, news aggregator, flask, css, static files

It doesn't seem "controversial" to me.

Try without the word "news", perhaps?

Perhaps "news"?

I'm wondering if "flask" is getting flagged (as in relation to alcohol.

Or news, because you know - "fake news"

Maybe, but the series before that had nothing to do with flask or news.

For instance a video in a series about writing a Go web app had these tags: golang, http, server, redis, gorilla mux, go-redis, web, application, tutorial, howto

And it was demonetized. Every single video I've uploaded seems to have been until I request manual review (which takes a couple of days).

gorilla is a racist term...

The algorithm doesn't understand context.

Do you know which excluded_ads labels it was tagged with?

For the video I uploaded yesterday, in the source for the video I can see:

"excluded_ads": "46=14_14; 59=14_14; 76=2_2_1,2_2_4; 102=1_1,1_3,2_1,2_3; 106=1_1,1_2_1,1_3,2_1,2_2_1,2_2_4,2_3,14_14,17_2_1,17_2_4"

EDIT: This may just be the generic variable in their source code, I'm not sure how to get the value specific to my video.

You should try making the same video with a parser written in Golang and see if that get's demonetized as well.

The series before this one was about Golang and they all got demonetized too.

I'm not sure which is funnier, his comment or your deadpan response.

Feedp arse r maybe?

What's a link to your channel? You've at least got me interested in checking it out.

https://www.youtube.com/channel/UC6tqFmkqznYJTzidBerHA8g

The video from yesterday (08 - CSS and static files) is still demonetized. If that gives you any context.

I'm 99.9% sure it's nothing to do with your tags and everything to do with some sort of speech to text analysis google is doing on your video.

In my case, I have a demonitized tutorial video on how to download and upgrade my software.

I was a bit surprised since it was a basic boring screen share video. Nothing controversial either, until I looked at the auto generated close captioning.

I said "Prosper202" but Google heard "pr0st!tut3" (I edited the actual word just in case HN also has auto block filters) But as you can see, it's obviously a not safe for brands keyword.

I've manually fixed the closed captioning with the correct words, but as of now the video is still demonetized.

So I think there's some sort of blacklist of words said in videos that will auto demonetize videos.

In your case, after watching your video with closed captioning I notice you use the word "explicit" a lot in your video. Once again given the context, it's totally safe. But there's also a very unsafe context that I doubt google is willing to automatically monetize without some sort of manual review. (https://support.google.com/youtube/answer/6162278?hl=en)

You can request a manual review of the affected videos, but only if your video has 1000 views in the past 7 days. For most small channels wrongly flagged, there's no real fix (for the old video).

The only real suggestion to test out is to not say explicit in your next video.

I'm also interested in hearing from others with demonetized videos, that may be suffering from the same auto captioning and blacklisted words issue.

That's a good point. I've seen my captioning be WAY off from what I actually said.

But when I request a manual review the videos do seem to get monetized again within a few days... Even with less than 1000 views.

This is correct.

Videos are demonetized mostly on a blacklist of bad words. Run text to speech on the video before upload (preferably upload the audio only on a test channel and download the captions), and edit out all words which have any negative connotation whatsoever.

Beware - the blacklist contains a lot of words which are only bad in a certain language or a certain region or slang you might be unaware of. Use urban dictionary to find the bad words... It seems you can let a few slip through, but if a bad word appears too many times, it causes demonetization.

That seems like an absurd step to require for every upload. Their algorithm should really give me the benefit of the doubt after having required manual review dozens of times and always having my content remonetized.

With "flask", "python", and "feed" the NLP is assuming you are some kind of alcoholic snake wrangler?

Some users are publishing the video as a private video first, get demonetized, ask for the review, the video is monetized again and user can publish it as a public video.

This doesn't help people posting time-sensitive content such as news or reviews. Those who used to fill this niche who have been targeted by the algorithms are seeing upwards of 90% loss in revenue despite having all videos re-monitized after manual review.

Are you sure that's possible? I've heard that a video needs to be averaging a certain number of videos (something like 1000/week) before they'll do the manual review.

My source is this video: https://www.youtube.com/watch?v=IM-gFU74UzA. This could possibly be false.

I make programming videos too, and a similar thing happens to me. The first day I release a video it's demonetized, and even if I don't request a manual review the video will be remonetized a couple days later.

Interesting. Maybe YouTube needs to adjust their algorithm for that category or something.

I'd bet this behavior is their algorithm "working as intended".

Aka, advertisers saying they don't want to waste ad spend on worthless programming videos. So they get demonetized for the first few days to reduce the impact of any ad spend on the times that would be most likely to get views, then turn it on so the drip of ad spend can treat the videos like problem children.

I'm more inclined to think that the algorithms are just a bit naive at this point. Surely some advertisers out there do want to target that market.

The notice on my videos is: Limited or no ads due to content identified as not suitable for most advertisers[1]. Your video remains fully playable and is eligible to earn subscription revenue from YouTube Red.

[1] https://support.google.com/youtube/answer/6162278?hl=en

Sure, just curious that the videos would be monetized after the first few days. From what I've heard the first few days of a video basically makes/breaks a lot of videos as far as monetization goes.

My guess is at this early stage of deployment, videos flagged by the algorithm to be demonetized (or a sample thereof) are automatically put into some manual review queue, for training and validation, which takes a few days to process through the volume.

Udemy does a lot of advertising.

Keep your eye on their 10Q statements next quarter. If it is what you say, it should turn up in the financials.

If I had to guess, videos consisting of mostly text (screencasts/slide shows) are considered likely to be bad videos (spammy junk and make-money-fast scams) and programming stuff is getting caught in the crossfire

The other day I stumbled upon this project:

https://github.com/Chocobozzz/PeerTube

It's a federated tube that uses ActivityPub and WebTorrent. I think this platform should be invested in and creators should consider moving. The interests of YouTube and its creators will never be entirely aligned.

That's a cool project.

However, to make videos with quality content at a reasonable frequency, many of these creators depend on YouTube for income or at least partial income.

Any P2P / federated service would have to find an agreeable way to A) send ads in a safe and private way to each viewer B) stop people from manipulating the system to game ads while accurately determining if an ad was viewed.

Tools like Patreon can help fill this gap.

Also, there's no reason creators couldn't use "dumb ads" where they read ad copy during their videos, and sell the spots themselves. Seems to be the main revenue source in the podcast world.

It feels weird to be an advocate for something like dumb advertising (maybe "content-targeted advertising" is a better name?), but dumb ads and Patreon really are a breath of fresh air relative to the status quo of web content.

Yes, it's one thing for serial content like a podcast or talk show, it's another for more artistic endeavors where it would lesson the artistic credibility to have an ad ingrained into the video.

It's also not a suitable format for those who create many small 1-5m videos.

Additionally, it prevents you from retroactively changing what ad gets sent with the video, which can lead to conflicts arising over severed business ties, and limit your overall ability to make residual profit from older videos.

> Yes, it's one thing for serial content like a podcast or talk show, it's another for more artistic endeavors where it would lesson the artistic credibility to have an ad ingrained into the video.

> It's also not a suitable format for those who create many small 1-5m videos.

I don't think having platform-handled dynamic ads (ie YouTube) differs from these dynamics in any meaningful way. The difference that matters is that with dumb ads the creator has complete control.

> Additionally, it prevents you from retroactively changing what ad gets sent with the video, which can lead to conflicts arising over severed business ties, and limit your overall ability to make residual profit from older videos.

This is a feature (of content-targeted advertising), not a bug. If a business model relies on tracking and targeting unknowing users who aren't aware that they can resist being tracked and targeted, maybe it was a crappy business model in the first place.

> This is a feature (of content-targeted advertising), not a bug. If a business model relies on tracking and targeting unknowing users who aren't aware that they can resist being tracked and targeted, maybe it was a crappy business model in the first place.

This is a false dichotomy.

There exists a large, partly unexplored space between hard embedded pre-recorded ads and targeted advertising that makes use of tracking.

Hmmm, that's a possibility I haven't considered. I suppose because I'm having trouble imagining what that in-between space looks like.

If an ad is targeted at a user, doesn't that necessarily mean they were tracked to some degree (i.e. some data is known about them, and an ad is being shown to them based on whatever that known data is)? Conversely, if nothing is known about the user (they haven't been tracked at all), how can an ad be targeted at them (besides knowing they are viewing a webpage on a certain topic, but then we're back at being content-targeted)?

The only possibility I can think of that's somewhat in-between, is a site tracking a user across their own site (first-party only), and targeting ads based on that collected data, rather than contracting that out to Google who can follow you across many sites. Is this what you have in mind? Other than this, I'm having trouble thinking of possibilities that exist between content-targeted and user-targeted via tracking. It's interesting to consider though.

The parent isn't arguing against content-targeted advertising, just pointing out that it doesn't have to be baked immutably into the content the way YouTube/Podcast sponsorship messages are.

One example of this is the https://www.projectwonderful.com/ model of auctioning ad slots the same way you would sell e.g. a billboard. The only targeting in this model is based on demographics of the content's audience at large (e.g. advertising Bobcat in a Box on Explain XKCD).

When Linus Media Group sells advertising segments (an example of baked-in advertising), they don't provide user data. The advertisers purchase segments with the knowledge of the kind of crowd Linus Media Group attracts.

Linus isn't going to open a video talking about women's pantyhose. He will open a video talking about an accessory, peripheral, or computer part that a viewer is actually more likely to buy. This is very effective advertising.

Since most of LMG's content is medium and long format and doesn't rely on complete artistic integrity in each of its videos as part of the value, baked-in advertising is acceptable and understandable.

While baked-in advertising still doesn't work for other kinds of creators, it doesn't mean some form of targeted advertising can not be done in an ethical fashion.

It's not enough. Patreon only works for certain audiences in its current state, and additionally lacks certain integration tooling that would make it more transparent.

Any form of advertising would likely be unacceptable on a platform like this, as would any attempt to enforce copyright or limit the redistribution or reuse of content. Creators would just have to find another way to make money and accept that they have no practical right to "ownership" over the content they upload, regardless of legality.

To clarify: I don't believe the set of people willing to adopt a FOSS distributed peer-to-peer video streaming platform and the set of people who believe advertising to be tolerable would likely overlap enough to provide a better or more stable market than the existing web already provides. And advertising on the web is already breaking down and likely doomed to fail as a concept.

> I don't believe the set of people willing to adopt a FOSS distributed peer-to-peer video streaming platform and the set of people who believe advertising to be tolerable would likely overlap enough to provide a better or more stable market than the existing web already provides.

> And advertising on the web is already breaking down and likely doomed to fail as a concept.

How did you reach these two conclusions?

From observing the hostility towards advertising on the web grow and become mainstream through the adoption of ad blockers, and ad-driven sites (like Youtube) seeming more and more desperate to actually find profitability through advertising. It seems to me that we're approaching a watershed where advertising on the web just stops being effective.

And at least, on the web, sites can enforce some degree of control over content and have a monetization strategy in place. With a decentralized platform, it's more difficult to integrate and guarantee the efficacy of advertising because it's more difficult to control the platform and user experience.

I use ad-blockers to prevent arbitrary parties from running arbitrary code on my computer. Not because I detest seeing a 5-10 second ad every 20-60m worth of videos or seeing a small text block or banner ad or sponsor list off to the side.

People use adblockers not only to block unauthorized code but because advertising has become an avenue for fingerprinting and tracking, and thus, longterm, a tool of socioeconomic oppression. People are also fed up with ridiculous loading times and payload size.

I have no problem with the way certain hosts choose to conduct advertising on their websites and I try not to block advertising on their websites.

There will be stability one day in the form of a truce between advertisers and end users, where lightweight, self-hosted static content that isn't fetched remotely clientside or laden with scripts will find a way to make money in the general market again.

They can use stuff like Patreon or STEEM/other tipping cryptocurrencies (that should be baked into the platform with an easy to click upvote button).

Before you laugh, I've seen people make $1,000 a day consistently like this. And this sort of community is still tiny (fewer than a million people). Can you imagine if 100 million or a billion people could upvote content like that?

I think something like STEEM is the future of content monetization.

Similarly, https://d.tube/ uses IPFS network to retrieve videos.

On the upside of this, I've seen lots more creators including sponsors in the actual video. (This video brought to you by Blue Apron) seemingly with lots of creative control over how they present it. This seems like a much more sustainable method. Ad blockers can't do anything about it, and they are safe from random de monitization.

On the other hand it's more work for the creator, and for people just starting it may be impossible to convince anyone to sponser them.

Hm is anyone aware of a third-party marketplace for placing YouTube Creators with advertisers?

I'm curious how this will net out. Facebook for example is starting to crack down on 3rd party deals with Instagram influencers[1]. So the official policy now is that you need to run things as branded content (how well it is being enforced is another story). My guess is Youtube will have a similar policy, and provide a toolset for brands to go through to setup these partnerships, and of course give Youtube a cut while probably pushing the advertiser to spend on behalf of the influencer to actually get any reach worth caring about.

[1] https://www.facebook.com/policies/brandedcontent/

Does YouTube let sponsors/advertisers opt-in to advertising on "demonetized" videos at all? For example, let's say I really want to advertise on all of a certain channel's videos.. can I do that? Surely if I'm happy to pay money to do so, they should allow that to happen regardless of if big-name brands like it or not.

This has been my thought as well. Adding a "sub-par" video level between "any ads" and "no ads" to allow less strict brands to advertise anywhere seems like it would help both supply and demand of ads. Maybe Fischer Price doesn't want to advertise on profanity-laden "let's plays" of gory video games, but beer brands probably don't care as much.

This exists already. "Demonetized videos" are more accurately described as partially monetized videos.

There is a tier of completely demonetized videos, but these are pretty rare to the extent that a lot of creators didn't realize there was this separate distinction.

Somewhat...

https://www.polygon.com/2017/10/10/16453306/youtube-monetiza...

>A YouTube representative confirmed to Polygon that there are a number of “whitelisted” partners who are allowed to run partner-run ads on the website.

>That means they aren’t flagged by YouTube and used external sale forces to sell the advertisement for them.

  Wow, this file is really popular! Some tools might be unavailable until the crowd clears.

Even Google Docs is susceptible to the hacker news effect.

That is mostly a Google thing. I've seen small subreddits activating the same thing in Google Drive. I would say that the threshold might be as small as 1000 people trying to access it.

I think it is right to „soft“-censor in this way (i.e. the information is accessible but not promoted). Reality without sorting and filters would be inhuman because market incentives to grab people’s attention would maximally exploit our tribal instincts and thereby completely destabilize our societies. This is already happening and it needs to stop immediately.

Who decides what to filter? How are they elected?

Private people appointed by the company in accordance to CSR principles established by the board of directors.

There will always be a problem because people promote for vanity and status as well as economics. You can't get more human than that.

The next 5 years will be tech companies panicking and trying to impose social control vs. people becoming more aware of when they are being manipulated by other users (and censored by platforms) and opting out of platforms entirely and/or moving to less controlled platforms. Internet media will be seen like smoking and candy bar consumption are seed today. Some people will eat candy bars every day because they don't care about their health. Others will choose to be healthier and it will happen by increasing awareness rather than heavy handedness.

You don't have to go that far to justify demonetization. It's a way to encourage higher quality content in contrast to low quality clickbait and sensationalism. That's good for the users and also in Youtube's long-term interest. On a side note, that's not censorship at all, neither hard nor soft censorship, it's just a kind of automated editorial work.

>It's a way to encourage higher quality content in contrast to low quality clickbait and sensationalism.

Maybe its higher quality, but you will essentially get the same type of content you already get on cable. Its content that has the potential to be popular, is bland enough that advertisers don't mind seeing their logo on the content, and is conducive to people buying things that the advertisers are selling.

YouTube is going the way of being just another cable provider. They flirted with the "community" aspect for a while, but in the end decided they care much more about advertisers. Which is fair enough, IMO.

Agreed. I mean, Russia already verifiably tried to influence the 2016 election, right?

Maybe the algorithm can be adjusted to not promote content that aligns with Russian interests. That way we can ensure people are only seeing the perspectives we want them to.

That puts Youtube as a significant political power though.

Interestingly enough, there was thread saying Youtube was censoring searches of "navalnyi" a critic of Putin. https://news.ycombinator.com/item?id=15806393

https://www.youtube.com/watch?v=n3H8D2LrLHc&feature=youtu.be

This a video from the author with some commentary.

Who is Karlaplan? And why choose Google Docs to share information which Google may dislike and remove?

I was wondering if they might not be baiting Google into doing that. If they start removing content critical of them from Google Docs that's another big story.

Google should monetize Google Docs in situations like this, i.e. more than a dozen viewing, all with the same referer URL.

Regarding the thesis, it seems to me that unless you have read the advertiser guidelines then you are always going to have your viewing manipulated in ways you do not understand. The recommendations are really 'how best to monetize you'. So with music you may end up with mainstream artist recommends, not because the songs are inherently popular, it is just that they can be monetized.

Therefore, the best way to explore the back catalogue of an artist is to upload your own video and to then see what tracks by your favourite musicians you can set to it. So even within normal YouTube there is a 'better' search engine, better for you but maybe not so monetizable even though that is what the widget is about.

Solid game theory on both sides. The Demonitization bot and its shenanigans are well known in the Youtube content creator community. They know they aren't fooling anyone and this data is going to be "out there" along with countless anecdotes from content creators who reliably upload content only to have it get "demonetized" during the most profitable portion of a video's lifespan. (the first 48 hours)

If Google pulled the doc they'd likely get a lesson in the Streisand effect, and they likely know not to try. Posting it on Google docs is actually smart bait by the author in my opinion-- intentional or otherwise.

One thing I haven't figured out from the section 2:

Those tweets claim that demonetized videos don't reach anyone on the homepage. Do they reach anyone on the Subscriptions page?

For me, the Subscriptions page pretty much replaced the homepage, and it would be bad for me as a consumer if the channels I subscribe to don't end up on that page once they upload new content.

Also, I'm mildly annoyed by the lack of consistency with spelling of the word "YouTube" in this analysis. It's neither "Youtube" nor "youtube" (both of those spelling are used multiple times), it's YouTube.

An interesting read. However, when running a search engine we discovered there is a second order relationship between page traffic and advertising, that is click fraud.

There is a 'yuuuuuge' click fraud problem on the Internet where a shady advertising reseller and a botnet collude to put ads on things and do what ever action is needed to get them to pay out (click (CPC) or view (CPI), never execute a transaction (CPA)). It is also disproportionately spread over 'low volume' inventory. As a result if you're page or video is on the bubble (which means its not really that popular or what not) if it gets demonetized you may notice a loss of 'views' because the bots have nothing to click any more. Not saying this is the issue here, just saying it is an issue.

That said, early on in the AdSense for Content program it was pretty clear that if you were not ranking on Google's index and then added AbC ads to your page suddenly boom! you had a much better rank. I don't know if that is still the case but it was a scheme we saw people using to try to get their pages to rank higher. The temptation is always there if your both the supplier of the recommendation and you're able to benefit from recommending some things and not other things. All things being equal you go with the ones the might make you money.

What is "demonetization"? I've never heard of this before.

Does it mean YouTube stops inserting ads in your videos? As a viewer, that sounds pretty great to me. It sounds like they also stop suggesting your video to random viewers, which is understandable if they're not making money off it.

Surely the real problem is that ads are toxic, the ad-supported content model is toxic, and YouTube's automatic recommendations are horrible (see the recent article on terrifying race-to-the-bottom videos for kids: https://medium.com/@jamesbridle/something-is-wrong-on-the-in...)

Edit to clarify: yes, of course I realize that content creators need the money. I'm just saying that this is a bad system that often rewards bad content and punishes important-but-uncomfortable content. Unfortunately it's almost the only system in town. I would love to see e.g. Patreon rivaling YouTube in revenue.

Edit edit: and ads aside, relying on YouTube (Facebook, etc) to tell us what to watch is terrible. We should instead do our own crowd-sourced curation, aka recommend good things to each other. And we need to defend ourselves against the likes of Facebook and Twitter insinuating their own filters between us and our friends.

The report not only talks about that. It also talks about videos that, because of being demonetized, they also appear less in YT searches and related content.

In other words, if you video has ads, then the probability of appearing in searches and related content is higher, thus you, as used, are more exposed to ads.

>In other words, if you video has ads, then the probability of appearing in searches and related content is higher

So users that permit ads on their video get a virtual "fast lane" on the Comcast internet service. You know, I'm sick and tired of these internet service providers violating Net Neutrality principles.

Oh, wait. It's YouTube? Oh, that's fine then./s

Youtube not providing extra promotion for content is different from Comcast charging extra to even access that content.

Not permitting ads from the user's end doesn't even influence the ranking and likelihood of showing up in recommendations (as far as we know). It's just demonetization from Youtube's side that has that effect.

You do realise you're comparing a company that makes its money from advertising on top of an otherwise video sharing platform, with a company you pay a fixed monthly fee to access the internet, right?

>a company that makes its money from advertising on top of an otherwise video sharing platform

I know you meant YouTube, but that describes Comcast cable TV service as well. YouTube is a competitor. Should Comcast be forced to deliver a competitor's content?

>with a company you pay a fixed monthly fee to access the internet

Some parts of the internet. You don't get the "whole" internet. You've had limited internet and fast/slow lanes for decades.

http://www.newsgroupreviews.com/comcast-newsgroups.html

https://www.xfinity.com/support/internet/list-of-blocked-por...

https://wiki.p2pfoundation.net/Asymmetrical_Internet_Access

>What is "demonetization"?

YouTube deciding your video is ineligible for earning ad revenue. So the creator gets nothing (or close to nothing, not sure). I'm not sure if it actually stops running ads on the videos since I've been using an ad blocker for ages.

Wow, I would certainly have assumed that if they show ads, they must share the revenue.

If your video gets demonetized no (or much fewer) ads get shown. If ads are shown, revenue is shared between Youtube and the creator.

Youtube content creators rely on ads to make their living. Monetization (ads on videos) are managed automatically by an algorithm. There are a lot of false positive where people spend weeks working on a video and end up getting no revenue for it.

I watch a lot of videos on world history. It happen very often that those YouTubers have their revenue cut off because the video is automatically flagged as "war", "weapons". It's ridiculous since those wars and weapons are those of the ancient world.

Worst than that, recently one big user had his account outright deleted for talking about ninjas and their weaponry in a video uploaded around 10 years ago. The algorithm went through his previous videos and flagged his entire account as undesirable.

See this petition where it took 35k+ users for his account to be restored. https://www.change.org/p/google-inc-reinstate-the-thegnthran...

Seems to me content creators who get upset at their treatment by Youtube have made a serious mistake; they believe they are Youtube's customers. They are not. Neither are the people watching the videos. The advertisers are the customers, just like most other Google services. You are the product. If you don't like it, you need to switch to a platform where you are the customer and the provider cares about you on that level. Google never will.

I don't think they're under that misunderstanding. However, they correctly believe that they are the hook which pulls in viewers - the product - to sell to advertisers.

Take away those hooks by removing their incentive to post videos, and you pull in fewer viewers, resulting in fewer ad sales to YouTube's customers.

Sadly, the most affected are those who cover niche categories, which means YouTube can afford to fuck it up and not feel any significant pain. (Yes, I consider 90% loss in revenue for their bait, err, content creators to be worthy of the phrase "fucking it up").

>Yes, I consider 90% loss in revenue for their bait, err, content creators to be worthy of the phrase "fucking it up"

I'm not certain Google are "fucking it up," or at least, that this effect isn't intentional. How much revenue do those niche creators generate for Google? How much first party content do they include, that media companies might object to? Google may consider them to be a risk not worth the reward if their intent is to optimize their algorithm to prefer first-party commercial content, in order to increase the value of their platform to advertisers.

The existing system clearly hasn't been working for Google, so to me it seems more likely that the apparent failures of its algorithms are part of an attempt to forcibly pivot the platform while maintaining good PR. I could be wrong, though.

It comes down to what do you (or what should Google) value more:

Varied content covering a broad range of topics, pulling in broad and varied audiences

Google's short-term bottom line.

Personally, I value the former, which is why I think that Google is doing it wrong. Focusing exclusively on the bottom line doesn't help society, doesn't even really help companies in the long run. If you fire your entire R&D team, you'll goose the bottom line, at the cost of your company's future. If you (effectively) fire your content creators, you endanger the future of your video platform.

Everyone yells "Patreon" as the solution for content creators, but this is a bad thing for Google. It removes their control over the content creators by removing the biggest form of "stickiness" to the platform.

Twitch started out as a platform for gaming live streams. Youtube is demonetizing gaming videos. It's creating conditions ripe for an exodus of gamers (creators and viewers) from YouTube.

Ironic that they're using Google docs platform to publish the research. Maybe the censorship effect is not advertent?

Perhaps it is a dare for Google to make the incredibly foolish move of deleting the content or taking action. Even if technically against the terms of use to reverse engineer, taking action against the account would bring more negative attention to Google.

I would still make a backup, just in case

Brave browser is a potential way to fight YouTube's censorship.

http://www.breitbart.com/tech/2017/11/17/brave-browser-lets-...

You are linking to Breitbart? Steve Bannon's yellow rag?

No, he is linking to Steve Bannon's hungry ghost.

I like how upon opening the doc I see for the first time ever "Wow, this file is really popular! Some tools might be unavailable until the crowd clears.Try again Dismiss" HN stressing the internet.

The script and data used for this analysis is available here: https://drive.google.com/drive/folders/10gLKGH1fqCp39x_UcEb9...

And here is the original video from the creator with all the links: https://www.youtube.com/watch?v=n3H8D2LrLHc&feature=youtu.be

If you are interested in replicating this or extensions of this work, get in touch! You can reach via the email in the profile, here, or http://twitter.com/metaviv.

It's true, I no longer see my gun channels in my recommended videos

Serious question: what is HTBQ+?

Googling doesn't seem to turn anything up.

It seems to be a variant on LGBTQ. Except I guess with "Homosexual" replacing "Lesbian and Gay".

Correct, in Sweden we call it HBTQ+, and "Karlaplan" is a location in Stockholm. Seems likely whoever wrote this is from Sweden.

The article only uses that term in the abstract, and doesn't mention anything related to LGBT elsewhere. Hard to take that claim seriously.

Must be a reference to this:

https://en.wiktionary.org/wiki/HBTQ

>>We used youtube’s official “DATA API v3” to get the top 25 related videos for 100k videos. We then scraped the related videos for information relating to monetisation status, views, etc. The corpus containing information of the scraped videos were then analysed in detail.

Anyone knows any such list readily available? I have been wanting to Scrape 100k video URL videos for my own content classification. Any suggestions will be appreciated.

Have you considered the Internet Archive?

Anyone where to get started with content classification would do well to consider Hands-On Machine Learning with Scikit-Learn and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems written by the former head of video classification at YouTube (https://amzn.com/B06XNKV5TS).

source: https://news.ycombinator.com/item?id=15772174

I brought the book for Black Friday. Its on my december readling list. I will look through internet archive. Thanks!

I imagine another flag is the amount of time viewers watch your video, in an attempt to demonetize clickbait videos.

I never understood why Youtube didn't have uploaders select various categories for videos. It would do a number of things:

* allow them to demonetize / ban / block videos for abuse of categorization

* train machine learning on the various categories

It's a lot more socially palatable for Youtube to say "Hey guys, we HAVE a blood and violence category, but you didn't use it for this video. So we'll punish you for that."

Instead, we get programming videos getting demonetized / blocked for god only knows what reason. That makes people made, and is bad press.

this would be an extremely dirty dataset, as there is no incentive to not put your videos in as many categories as possible. if you can only have one, people will still put it in the most popular category a large fraction of the times I'm sure, not the correct one

> there is no incentive to not put your videos in as many categories as possible.

We are talking about how to incentivize people here... why suddenly decide that there's magically "no incentive" for something?

Youtube is free to also de-incentivize people for using the wrong category.

Which they would determine using... an automated algorithm which decides which category a video should be in, and comparing it to the category chosen. At which point...

it's not a training set anymore, but a machine learning production task

Well, it's both a challenging production task (which Google is great at) and a learning from streaming data task, which Google also has some experience with e.g. news. The latter is certainly a interesting challenge, but many researchers are already working on it.

> At which point...

They have more data than they do now for machine learning, and a better PR story.

i.e. uploaders can't be mad about the categories, because the categories are chosen by the uploader.

Uploaders can be mad about Youtube double-checking the categories and getting it wrong... which is less likely to happen if they have better data for machine learning.

What, exactly, is the down side of that?

Ultimately the determination is still made by the machine learning process, so you're describing extra work to provide an interface of dubious value that will be used more to misrepresent video content than to provide useful signals, and it seems that customer support related to this would increase dramatically.

I think Google relies overmuch on questionable ML in most areas, but in this case, the alternatives are either ridiculously expensive, or easily exploitable.

The implicit goal of most YouTubers is to get views, as that's how they make money. Putting your video in a smaller category/in not the max amount of category reduces your exposure, reducing your potential views. The incentive of having a lesser chance of being flagged feels tiny compared to that

yes, it might be a better idea to let viewers of videos choose tags/categories and vote on these categories, than to allow the uploader.

That seems like a reasonable idea to me but I think part of the answer is that a mind boggling amount of videos are uploaded to Youtube every second. Tracking categorization abuse seems straightforward in theory but good luck automating that reliably over millions of videos especially when you know that some very dedicated people will try their best to game the algorithm in their favor.

People will start using the "blood and violence" category exclusively if it means they're less likely to be demonetized, rendering it useless. Or the other way around: if this category means they get fewer views they'll try their best to be always nearly avoid it, except of course it can get very subjective and some people are going to be flagged "blood and violence" even though they argue it's not that violent and channel XYZ did worse and didn't get tagged etc...

When it come to moderating Youtube, and given the ridiculous amount of content hosted there, you shouldn't think "how would I do it" but rather "how would I design an algorithm that would do that". And suddenly it becomes a lot less obvious. You can't teach algorithms common sense (yet).

> Tracking categorization abuse seems straightforward in theory but good luck automating that reliably over millions of videos

Is Youtube (a) automating video categorization now, or (b) not automating video categorization?

> People will start using the "blood and violence" category exclusively if it means they're less likely to be demonetized, rendering it useless

No... it means that the advertisers can determine whether or not they want to monetize ads in that category.

So there would be no "demonetize EVERYTHING", just categories that advertisers can choose to place ads on, or not.

I think you're assuming that a different system would work exactly the same as the system works today. That isn't the point...

Other commenters speculate that there is a technical reason why YouTube doesn't allow content producers to choose their monetization categories but I bet it's more likely to be a business reason. I'm sure it would be technically challenging and possibly for not much reward (for Google), however, I bet it's even harder to sell advertisers on ad packages that include Google ads when any producer can opt out of any group of ads. It would make it much harder to tell advertisers how many eyes their ads would actually make it in front of.

People would game the system. I do music, but everyone knows that other categories pay higher.

It also would depend on the audience you naturally attract. Attracting 19 year old girls will do better than 65 year old men.

YouTube does actually have video categories that the uploader selects, although historically it has led to people mis-categorizing their videos on purpose.

For example, before gaming was a category that was embraced by YouTube, many popular gaming creators would categorize their videos under Comedy instead of Gaming, because the Gaming category did not have a spot on the front page. This also led to a fights between gaming YouTubers over mis-labelling under lesser used categories just to get the top spot and end up on the front page.

As far as I can tell the categories are as follows: Autos and Vehicles, Comedy, Education, Film & Animation, Gaming, Howto & Style, Music, News & Politics, Nonprofits & Activism, People & Blogs, Pets & Animals, Science & Technology, Sports, Travel & Events.

Absolutely. Flickr had a pretty good system.

I think the reason YT avoids this is for fear of monetization loss as fewer things get exposed to wide audiences so there is less "vitality", and YT loses advert potential.

If they cared about users and creators, they'd take a page from Flickr.

I thought this was about the demonization algorithm that youtube comment bots use to cyberbully people.

> This will in turn lead to censorship of political ideologies, HTBQ+, mental health awareness, suicide awareness and prevention, etc.

Sounds like a slippery slope argument.

Mate. https://en.wikipedia.org/wiki/Slippery_slope#Non-fallacious_...

My Subaru has four wheels just like a Ferrari.

Mental health awareness channels already have their videos demonetized. Whatever strategy Google is using, it's afflicting every community on youtube regardless of topic covered, whether they are left or right, etc.

Reply on news.ycombinator.com