The end game isn’t embedded stream, it’s attestation

I've heard things along these lines, but always as an aside. Can you expand on this?

Like, YouTube would only work in a browser environment which has been approved by Google? and which has promised not to be running in headless mode? And which promises to have validated that there is at least one authentic retina looking at the screen?

What is a reasonable end game situation to be working on countermeasures for, and what is tinfoil haberdashery?

The end game here is to simply turn all personal computing devices into locked down ecosystems that the user has no real control over, basically what iOS already is (and Android isn't too far off from).

All code being executed on the device, from the bootloader all the way to the browser or youtube app, must be approved by the manufacturer, with an opaque black box security chip inside the device verifying that this is true and providing a signed message as proof that has to be sent to Youtube servers before videos are served. This completely eliminates the issue of both adblockers and third-party frontends.

After that, impression tracking just needs to start covering some of that gap of air between your brain and the screen. Let's throw in some attention-detecting cameras so you can't skip the ads by looking away from them.

They'll let you skip commercials, but not without shouting the brand name.

https://www.techradar.com/news/sony-patent-would-have-you-ye...

They'll use ChatGPT (in spoken-audio form, with speech recognition) and you'll have to tell the agent what the ad was for, and why the product is so great, to prove you watched it and really understood the marketing message.

What mute buttons were made for.

Drink the verification can.

I suppose I should expect my phone to start refusing to run gcc at some point then?

Extremely difficult? No. SponsorBlock already removes in-stream ads. If things get more complicated, ML will be employed. It's an arms race but so far things seem to be on our side. Long-term, I don't know.

They could just block buffering this video until the time of the ad is finished on serverside, as they know how long the ad is.

You still can block ads but you must wait if you see a black window or an ad stream.

We can just buffer manually and wait. It's an unsolvable problem without DRM, and even then...

This whole approach would massively increase the cost of distribution due to how CDN networks scale.

A silent black screen would still be a win. Ads are disinformation.

Yep, I'm tired of fighting the arms race with twitch ads, so I just mute the video now and do something else for a few minutes.

I would honestly be perfectly happy with a black screen for 30 seconds as a compromise to ads. I only leave the twitch webpage because the visuals bother me almost just as much as hearing the stupid sales pitch or the goofy "lol we're so cheeky and random" jokes made non-stop by insurance company ads.

I'd rather a silent black screen than an ad.

I can't understand people using SponsorBlock. Basically you want people creating videos for free?

Me watching a sponsor segment isn't going to make any money for the video creator. They're being paid by the sponsor to place the segment in their video; there's no way for the sponsor to track who watches it. The segment is just a waste of my time.

This is separate from YouTube itself tracking who watches the video, and for how long (and if they skip any), and if they watch any ads, and then giving a bit of money to the creator for the ad views.

Is this true? I'm not sure if sponsors ask for this but I think metrics are available to creators to show how many people watch during different portions of the video, so they could, including post-payment based on the results.

See audienceWatchRatio under https://developers.google.com/youtube/analytics/metrics

Example: "A one-minute video is watched 100 times. Half of the viewers stop watching after 15 seconds and the rest watch the entire video. None of the viewers watch any parts of the video more than once. In this case, this metric's value would be 1 for time buckets in the first quarter of the video, and its value would be 0.50 for the remainder of the video."

So someone could ask for ratios before, after, and during the ad segment to see if it's worth sponsoring them again. On average they can tell how many people skip the segment, how many stop watching entirely, etc. You can also see this on the UI where it shows peaks above the timeline with other users clicking past sponsored content.

Yes, but if SponsorBlock or any other solution like that starts to become popular, sponsors will stop paying creators.

Like it or not, the whole way "free" services on the internet work with ad-blocking, and always have, is that the people blocking ads are taking advantage of technological measures to avoid them, so that the actual cost of serving content is paid (by ad-viewing time) by all the suckers who don't know how to block ads, or don't want to for some reason. It's basically like that with sponsor segments, though here I think it's more like a television or newspaper ad: the sponsor is simply hoping enough people will watch the segment and go buy their stuff (since it's personally recommended by the content creator!!) to make their sponsorship investment worthwhile.

How is it fair for creators to take my premium money AND force me to watch sponsor segments? That's double dipping. YouTube should have a system for marking segments so premium users don't have to watch those segments. Premium users pay with premium money and free users pay by watching YouTube ads + sponsor segments. Fair in my book

So, don't? Nobody is forcing you to do anything. Just don't watch or don't pay premium. Jeez, since when does the world owe you anything?

The only way SponsorBlock will become popular is if a huge fraction of YouTube users switch to 3rd-party clients. I have a really hard time seeing that happening; that would be like most Mac and Windows users all switching to Linux.

I just uninstalled the Youtube app. Now I use the website on my phone, and Sponsorblock works without issue.

Why do I need the app for anyway?

If I understand you correctly, you're using YouTube in a browser on your phone. SponsorBlock only works on your phone because you've installed an "app": a browser extension that implements it within the browser.

Most people don't do this.

More people ought to do this. It's the best way right now to defeat spurious decisions by platforms. Reddit's app is off my phone. I use its terrible website, so I don't go there anymore. Youtube has a good website, and until they enshittify it, it's the best way to experience Youtube on a phone.

I totally agree: using the website (through Firefox + uBo) is generally the best way, because the user has the most control. A proprietary app doesn't give you this.

There are more than 2 options

Amusingly, a person could pay for YouTube Premium (thereby, allegedly, giving the creators more money) and still use sponsorblock to block the double-dipped inline ads.

Most people I watch on YouTube are either not sponsored or on Nebula, so I just watch there instead.

Sponsorblock doesn't take away sponsorship money from creators. Sponsors have no way of knowing how many people use sponsorblock. Oh I also pay for premium. So the creators get their cake and I eat ad-free content. Win win

People don't come to YouTube for ads/integrations. When I click on the video, I only want the actual content. Ad segments aren't the only way to monetize anyway.

> Basically you want people creating videos for free?

Noone "wants" anything here. It's just more convenient.

Since ads are personalised, this would produce a new stream per view which would come with a computational overhead. The stream would have to be rendered on the spot and not in advance, which puts further pressure on compute resources during peak times. That would decrease margins, maybe so much it isn't feasible at all.

(Disclaimer: I don't know that much about video codecs or editing.) I'm surprised they don't just break the video at key frames and insert the ads that way. From my little use of open-source video editing tools, it seems like this is pretty low-CPU to do, and if (like YouTube) you don't care about a good user experience and are happy to just insert ads at seemingly-random places instead of between natural breaks in the program (as with TV shows), it should work.

You are probably right, eg Losslesscut in Linux does exactly that: no transcoding, no overlays nor any editing other than cutting and moving around sequences from the same video. Editing and exporting such videos is blazingly fast.

Adding a bit technical context (for readers) to your comment on inserting ads at "random places": many codecs use the concept of an initial frame [1] which would place technical restrictions on where the ad could be inserted. If Youtube believes that ad timing is too important, they might not accept technical restrictions on the timing.

[1] https://en.wikipedia.org/wiki/Video_compression_picture_type...

Creators can mark specific sections in the video where they want an ad to play. I assume this dynamism also plays a contributing role.

YouTube reencodes uploaded videos, so YouTube controls where the keyframes are at playback time.

> and if (like YouTube) you don't care about a good user experience

Other than ads, which is the business model and keeps the content flowing... does YouTube not care about UX? I generally find the UX pretty good. I can always watch a video near-instantly, on nearly every device I own, search is astonishingly good even when my Apple TV mis-hears my searches, and it's one of the biggest video streaming sites in the world.

As a user I think the UX overall is pretty great. Sure there are things I'd tweak, but I can see reasons for them.

Removing downvotes?

Randomly rewarding the latest 'feature' so you get a load of shorts of videos you've already seen, random votes for things, etc.

I've never used the voting. It's a quality signal and they are notoriously difficult to get right. Netflix infamously switched from stars to like/dislike and everyone hated it, and yet it provides better signal and recommendations become better and therefore you get better content surfaced that you'll enjoy more. I'd guess a similar thing happened with YouTube.

Zooming out there's typically an ecosystem benefit, because otherwise why would they do it. That ecosystem benefit isn't always about money, if no one watched videos there would be no business. In fact a business that is often much better than a business with no users that isn't, so user happiness and experience tends to be prioritised highly in products/companies like this.

A video with 1000 upvotes and 1 downvotes tells me something

A video with 1000 upvotes and 2000 downvotes tells me something.

A video with 1000 upvotes doesn't tell me anything.

I can still downvote, I just don't get told the downvote total.

If they decided the system was broken and decided to replace it, I could understand. But they haven't changed the system, they've just removed the utility it did have.

You’re right about the YouTube too, they also started with rating videos x/5 stars

>does YouTube not care about UX? I generally find the UX pretty good

You think it's "pretty good" when a video is interrupted at some random point in time (usually while someone is mid-word) to play some annoying video at a much, much louder volume than the video? That is not what I consider a "good user experience" at all.

You can chain MP4 containers without re-rendering anything. It's extremely quick and computationally inexpensive. Of course at such a scale it might still cost a lot in the absolute sense, but I'm pretty sure it would be balanced by the increased revenue

I understand Youtube uses a variety of codecs and encoding parameters to serve videos to a multitude of platforms and clients. Maybe there are restrictions on some clients that force Youtube to pick a codec for them which makes ad embedding unprofitable. Circumventing embedded videos would then be as trivial as faking a particular client signature.

They could certainly prepare a stream for every video with inline ads that's used as a fall back when personalised ads won't load. Certainly showing a non-targeted ad is better than showing no ad, right?

If you make the generic ads worse than personalized ads (show more, show them more frequently, or just get the most annoying ads) you'll actually get people opting into personalized ads as the lesser of 2 evils.

I would be very surprised if this isn't either in progress or done and just waiting for the right opportunity to roll it out.

Aren't Ads dependent on cookies, location, search history etc., such as it is, so remain flown in?

Simple: no benefit to serving ads you can’t count. YT doesn’t care about serving ads, they care about ad revenue.

What is about embedding ads within the primary video stream itself that suggests they would not be countable?