But no. You can trim your video. You can delete it entirely, and suffer the consequences. But you can't simply fix and replace a video that may have had erroneous content and videos which had been "patched" by annotations in the past will now appear without context (aside from descriptions perhaps).
That's some impressively lousy product stewardship, and I say this as someone who mostly disliked annotations as most people
here seem to.
What kind of argument is that when machine learning is extensively used to prosecute DMC and copyright violations? Why not detect video similarity as well and determine a cutoff under which a newly uploaded video is "too altered" and cannot be used to replace an existing video?
You don't need machine learning to do that. Even something like p-hash of every frame and if more than some threshold aren't the same, then it can't be replaced. I guess over the course of dozens of replacements you could replace the whole video, but you could always just store the p-hash of the whole original video, or even some sample of it, and always compare against that and not the most recent version.
Absolutely, I was just saying that a lot of effort is already put into pattern analysis for videos so that licensing partners can be pleased. It's not like it's technically impossible. The real reason is that Google doesn't have the incentive because it simply doesn't care about its users' happiness. It merely wants to lock them in. What other service will people migrate to when they are annoyed by YouTube?
It would if done naievely; one seemingly-simple fix for all the problems I've seen in this thread: have multiple video versions as an explicit part of the API. You can link to a specific version, youtube can clearly show a flag saying "there's a newer version of this video available, do you want to watch that instead?", it can show comments and likes for all versions in one place (with a badge saying "this comment was made on an old version" if appropriate), etc
In the context we're talking about, do you believe the comments on video A (10 minutes of maths lecture with a typo in one of the digrams) would be hugely different from the comments on video B (10 minutes of maths lecture with that typo corrected); so much so that it's worth throwing out all the early comments and starting from scratch? Or are you thinking of some other use-case?
> I don’t want to additionally have to pay attention to whether we watched the same video or not.
Good news for you then! Making "video version number" part of the API means that you can filter those out programatically too :)
You've contributed to the discussion, but your opening gambit was inflammatory and I respectfully suggest that you tone it down in future. Just thought you should know that I was close to ignoring you outright because of your tone.
They're saying you could upload a pirated episode of Game of Thrones, get 100k views, switch the content and title to "My Newest Vlog!", and keep the views.
How is that different from uploading half a pirated episode of Game of Thrones and replacing the other half with your newest vlog? People could try it, but I think they'd notice that lots of views can be a bad thing if most viewers react negatively.
If it happens anyway, view counts could be disaggregated into "total" and "since last update".
But you can do exactly that on your own website, and for some reason we don't have an epidemic of bait-and-switch blogs. Why would it be different on YouTube?
YouTube tracks views and other popularity / credibility statistics for videos. Your personal blog doesn't -- it's even easier to just put up "My Newest Vlog!" with a little number claiming to have 100,000 views. Why would you need to do a bait-and-switch?
> for some reason we don't have an epidemic of bait-and-switch blogs
I've seen a number of social network groups being sold for their audiences and then rebranded to sell questionable things. I've even seen something similar done to some YouTube channel I once subscribed to: its owner (new or not, can't really tell this time) changed its name/images/whatever, removed all the videos and started publishing things that I never wanted to be in my feed. Things will only get worse the moment YouTube allows to change videos under the same URL.
I'm led to believe this is also common on Reddit; someone (or a bot) will repost to karma farm, then the account with high karma is sold off. I'm not sure how much there is to gain from that since you don't see users' karma unless you specifically view their profile.
I don't think it is different? I think even allowing edits to the video content would open up the door to replacing a video with something unrelated but retaining the views & popularity etc.
I think that's part of the problem. Does a new video, which may have nothing to do with the originial video, really deserve the views and popularity of the old video?
Agreed, YouTube shouldn't allow arbitrary re-upload. If they did, then 1) upload a popular song, 2) disable comments, 3) wait for it to get millions of views and tens of thousands of likes (https://www.youtube.com/watch?v=XycBLF6kWuY), 4) re-upload the video to now be a product pitch for your startup, 5) show potential investors how many millions of people are enthusiastic for your product.
And if one's reaction is, "re-uploading a video should reset all views and likes to zero", then re-uploading now has little advantage over making a new video.
I have disabled annotations long, long time ago. I was tired being forced to click on those silly boxes to uncover the content beneath.
On the other hand I don't see why Google couldn't just disable adding new annotations but left old annotations in peace. Maybe there is some underlying technical reason why they don't want annotations.
Code debt and shrinking use. Considering it's only on desktop and only on old videos, I'd assume the number of uses of annotation per view is probably under 0.01% in terms of traffic, and I can also assume it adds quite a lot of complexity to their player code.
Personally, I think maybe they could flatten all annotations into a single non-interactable layer and keep that instead. That would require a lot less code and do most of the job it needs to.
I imagine it's for consistency. The fewer unused features they continue to support going forwards, the simpler it becomes to develop the websites, mobile apps, and set-top box experiences.
YouTube is already pretty fragmented as it is, with each device having a different experience. Since viewing is increasingly a mobile thing, and the mobile site sucks, I would be more than pleased to see some consistency coming across their platform.
tl;dr Google may have money and YouTube may be one of their main sites, but there's always more to development than "but c'mon, they can afford it" or "surely they have enough engineers".
I understand what you're saying but based on what we've heard from ex-Googlers about user interface/experience design within the company, that people who've not got the first idea about good design are thrust into the role of designing user interfaces and user-facing features, I think it's a minor miracle we have what we have already — and what we have isn't great.
Plus, look at the variety of users and devices on YouTube. From those who watch 5K video on 27" high-resolution monitors backed by grunty graphics and powerful processors, to cheap $29 Android phones that barely keep the battery from draining to zero within a couple of minutes just showing a 360p video, and all the devices in between.
Between Google's issues with user interface/experience design on a company level and the fact that YouTube needs to cater to so many different kinds of people and devices, there's clearly more than meets the eye when it comes to the hidden scale of YouTube.
Since Google basically prints money, I don't think money could be the issue. Since Google is overflowing with engineers, I don't think throwing more engineers at the problem would help. Rather, I think that removing annotations is a good sign that Google is trying to do something smarter: manage the scale of YouTube.
The removal of just one feature, like annotations, lifts a heavy burden from the main site team, the mobile site team, the mobile apps teams, and everything else. It brings platform consistency, meaning some actually new and innovative features can finally get some of that precious engineering time, rather than the inelegance of supporting what was always a janky feature that could only become jankier as devices became more diversified.
> Since Google basically prints money, I don't think money could be the issue. Since Google is overflowing with engineers, I don't think throwing more engineers at the problem would help.
That's why Google Reader is gone, I suppose. Google just couldn't save it, no matter how much money and engineers' time was spend on it. And we all know how much exactly it were.
> Rather, I think that removing annotations is a good sign that Google is trying to do something smarter: manage the scale of YouTube.
Consistently removing the features which is not popular among majority of users is not smart at all. That's the dumbest way to manage the scale actually.
Think of Google as an airship. Platforms and features have mass. They've done everything they can to rise without dumping platforms or features. Now they're shedding excess mass to continue rising.
This means streamlining platforms and trimming excess/old code.
I've seen 'code debt' and 'technical debt' mentioned more than a few times in this thread and I think these are why Google is so aggressive about dropping support for things. They don't have time, and the engineers don't want to be forced into someone else's pet project (face it, that's where a lot of these platforms and features come from).
> I considered working at Google last year after a recruiter reached out to me, but their decision to backtrack on their promise in China changed my mind
I get the feeling that Google and Alphabet as a whole is pretty broken on the inside. I will admit that even though I support what Google is doing to make YouTube's platform more scalable, they do plenty of other things to screw things up for not just themselves but loyal content makers and users — the suggestions algorithm, subscriptions that aren't really subscriptions, etc.
To borrow a famous quote, Google is like an ship with a hole in the bottom, leaking water, and they seem to be trying to get the ship pointed in the right direction.
Still, whilst the ship's not yet overrun, I can't blame the individual teams for trying to throw as much water back overboard as they can muster.
You can't change your video after uploading for the same reason you can't change your submission on websites like Reddit after submitting it. It would be too easy to bait-and-switch. I can understand that it's frustrating for honest uploaders but that makes total sense to me.
In general people just mention that there's an updated video in the title or description and you can redirect yourself there if you want. A bit lo-tech but it works well in practice IMO.
That would be disastrous for YouTube. Imagine embedding a video on your website only to find out later that it had been replaced with spam? So little gain for a huge abuse opportunity.
Yes. Why not? If the 1st Video was your upload to begin with why shouldn't you be able to replace your video with a different Video?
This is a very useful feature for people since it would give creators the ability to fix errors in their videos without creating the confusion that happens when creators try to do this currently.
Every single time when I saw a reupload it had tons of comments saying something to the effect of "Didn't I see this before?" or "Didn't you upload this Yesterday?" and the creator commenting X times: "Sorry it is a reupload to fix Audio/Editing/Facts"
Can you then replace a John Carmack tech talk with a Amway recruiting video? Yes, but there was nothing stopping you right now from uploading that Amway video and simply calling it "John Carmack tech talk" anyways.
But now I can watch a correctly titled "John Carmack tech talk", send the link to a friend, and they'll wonder why I'm trying to recruit them to Amway.
Right now the system is so that I have sent links to friends and they replied: "Can't find it"
Both are minor things, but I am in favor of creators being in control over their videos. Especially since it would get used to correct accidental misinformation and things like that.
I just don't see much harm in the ability even if the video creator could completely switch content.
I think that it is great that they do not give the ability to replace the video. If I am linking a video that I watched in the past to someone I expect it to contain what I watched, not something entirely different. Same for upvotes, what if someone replaces a video with content that I find misleading and that I would not have up voted?
But no. You can trim your video. You can delete it entirely, and suffer the consequences. But you can't simply fix and replace a video that may have had erroneous content and videos which had been "patched" by annotations in the past will now appear without context (aside from descriptions perhaps).
That's some impressively lousy product stewardship, and I say this as someone who mostly disliked annotations as most people here seem to.