YouTube ranks “wholesome and funny” comments higher?
twitter.com
twitter.com
while this might seem annoying when one is used to the up/down vote mechanism I honestly think incentivizing people to check content for themselves is a significant improvement.
I think a lot of websites should remove signals and tone down recommender systems in favor of making people just watch or read. It's one of the things I still like about HN. The greyed out downvoted comments and order aside there's no real visual cues or gamified mechanisms here.
I installed addons so I could see the ratio before I even clicked on the video. It’s saved me so much time. (Unfortunately it only worked on YouTube so when I searched on Google I still stumbled onto bad videos and wasted a lot of time)
Yeah. In an ideal world "howto videos for dangerous activities" or even just "howto videos in general" would have very special policies applied to them. There's a lot of people who have no idea what they're talking about who like to make howto videos, and it's super important to surface critical information about those.
Which is a good demonstration of one of the fundamental problems of trying to make a general content-agnostic distribution platform. A general platform run according to contemporary business practices will tend towards one-size-fits-all policies, but that works about as well as one-size-fits-all pants.
It's slightly inconveniencing the majority in order to protect the minority from harassment. If anything, it shows YouTube has the balls to do what's unpopular.
If anything, it is the most lazy measure they could came up with, but even with that, I simply don’t buy it and I suspect the reasons behind are others.
Googs tried to make YouTube a more corporate friendly place
"Don't read the YouTube comments" was a common phrase, and for good reason. Nowadays, I wouldn't say YouTube comments are especially funny or insightful, but they are certainly tolerable. I imagine having a list of epithets, slurs, and phrases (and accounts that use them) to de-rank would have gone far enough to help fix the nightmare that was the comment section.
I find that a lot of them are something like 'lol', or 'love the way X did Y' / an unfunny meme (with thousands of thumbs ups) / something else along those lines. I find myself scrolling quite a long way to find anything insightful about what I watched, and just for something I can take with me in general.
The only video I found that sorta matches this is Chanel 5's ComplexCon video: https://youtu.be/jy9x09iCATA which has mostly meme replies.
Mentour Pilot made a video about a 747 crash https://youtu.be/Y50saxfTqQA and the top reply is from a pilot with a lot of the other replies being from pilots as well.
And the last video I looked at was one titled "All 8 Species of Bear" https://youtu.be/7DERN0R3AbM which I thought would definitely have trash comments because it seemed to be from one of those "Top #" click bait channels. But the top replies are bear stories, species information, and some zoologists. No idea how accurate the info is, however.
The livestreams were particularly bad because there'd be a comment box live-updating at breakneck pace right next to the video. I recall a description of the livestream of that SpaceX launch of the Tesla Roadster into space that was particularly striking- Here's this placid video feed of Starman calmly floating through space after escaping planet Earth juxtaposed with the incoherent screeching of the millions of angry apes stuck on it.
FB is still really bad, but in my experience it really depends on whether it's arguing with internet randos vs people you're actually friends -- or friends-of-friends -- with. The former is still horrible, but that's because you're nearly anonymous when it comes to someone across the country. I think it's not as bad as some internet comments though, people are more hesitant to say the really blatantly racist or murder-y stuff on something attached to their real name. It still happens, but not as often.
When people are arguing with those they actually know on FB, I find they hold back a lot more. They can still be mean, but they don't go as extreme.
> When people are arguing with those they actually know on FB, I find they hold back a lot more. They can still be mean, but they don't go as extreme.
This is what I mean by it's a societal/human problem though, that there is something to hold back... I say that is the root of it.
* Humans largely evolved in small tribes/bands. We just weren't built for communicating with and relating to potentially thousands or millions at a time. Not a surprise then if there are some 'bugs' lurking.
* Even sans the internet, while you may not have had this particular problem, you definitely had some others. I'm reminded of a part of The World Until Yesterday where the author talks about little kids in a non-state tribe spontaneously celebrating around the corpse of an enemy tribe member that had been killed. By modern internet standards, all of those little kids are sociopaths, or maybe the adults that raised them are all sociopaths, because according to many people today, any sort of obvious 'darkness' in human personality, like a tendency for violence, makes you a sociopath. I think the reality is that yeah, parts of our peronalities are fucked up, but that's essentially normal. It may even be a feature, rather than a bug.
Anyway yeah, you could view it as a part of human nature that's just revealed by new technology.
I will say that I don’t really feel like I’m missing out on much not seeing them.
My instinct is that the progress has been primarily because of the better tooling and moderation capabilities, just because the larger companies always promise that ML will take care of their large-scale moderation processes without much concrete evidence. But I also suspect that the answer is somewhere between the two approaches, and it would be great if YT were able to improve the wider ecosystem by sharing some of their strategies.
When you optimize for engagement (e.g. Facebook’s sorting), you get toxicity. When you optimize for approval, you get funny and wholesome stuff.
This does vary by channel of course, but if you're talking about a popular channel that shows up for non-logged in users like LockPicking Lawyer then the comments will consist of the same tired and lame but highly upvoted jokes in every thread. Maybe if you scroll down far enough you'll get an anecdote about the company that manufactured the lock or something, but never above the "lol, it just falls apart when LPL looks in its direction!" comments.
Also those "Who's listening to <random song> in 2028?" or "Timeless classic" on every old song (even kind of garbage ones, after all, emotional attachment doesn't really discriminate much).
This inane garbage is way better than the racist garbage that preceded it, though.
Personally, I tend to have a good experience with YouTube comments, but I'm definitely familiar with the memes you're talking about on certain videos.
I have noticed that they've improved from "zero-value abuse" to just "zero-value". I assume because they've gotten better at burying the abuse. Which is... something.
I'm starting to believe that whole NPC bots meme now (which is also pretty annoying, sorry).
And "wholesome" itself is a meme that is pretty annoying. It tends to drag "inane and superficial" along with it. Feels like we get so tired of walking past human tragedies camped on the sidewalk all day long that when we get home we sit there like coke-addicted mice slapping the "wholesome" button to get our Soma so that we forget what we just saw.
Wasn't the initial engagement on Reddit the founders spamming under different names?
I've been on various internet forums across the years and also on IRC.
I've haven't been on 4chan or whatever that thing is called, I think it's meant to be a completely unmoderated forum?
In any case, forums ranged from horrible to amazing. IRC, same.
The original, totally unmoderated version of Youtube comments was the worst dumpster fire I've ever seen. Stupidity, hate, racism, you name it, it was there. It probably had one of the worst communities in the history of communities.
Then about 5 (10?) years ago they added some kind of ranking system that improved things. It went from being maybe the worst mainstream community on the planet to somewhere in the bottom 100, I'd say. Quite an improvement :-))
So if they improved it again, maybe they'll reach Reddit main subreddits levels soon :-p
"This is the best conversation between two robots that I’ve ever witnessed."
The other top comments are just friendly and positive.
I noticed many (most?) of the top comments are made by accounts that on a closer look are promoting or even selling what must be some kind of scam or MLM. Would be fair to call them submarine comments.
Several of them were even verbatim duplicates of other comments on the same video. I got the feeling that a lot of this is automated, and not by the people supposed to be running the platform.
Just open up the YT front page in a clean unpersonalized session and go into those Minecraft/Fortnite/whatever that tend to end up high in the rankings and you’ll probably see what I mean.
I don’t know how many of these are robots vs knowing karma whores, but it’s definitely a pervasive problem.
It's a loooong way away from the comment section of just a few years ago, which was almost always full of rage, poorly-written comments, politics, and inside jokes.
[1] Sample of my homepage: https://i.imgur.com/IrmWii4.png
"I'm Indian/Pakistani, but hats off to Pakistan/India for creating this masterpiece. Music unites us all. Love to my brothers in Pakistan/India".
I mean, I appreciate the sentiment, but it feels odd that it's expressed in the comments section of a random YouTube video!
Another weird sort of comment is:
"May your loved one be blessed and live a hundred years. One like = one blessing."
Again, an embarrassingly high number of likes. Do people actually think a YouTube "like" is going to generate a blessing for them?!
You never know! Like this comment to get like blessing insurance. Like some of my other posts to get higher coverage. We protect you from bad luck from not liking posts.
This seems to be a common refrain, despite little evidence to back it up. Just because the dislike button doesn't render a count doesn't mean that it isn't being counted and used to weight comments.
From there it's probably fair to say that way more younger people watch lighter stuff than oldsters there, and there are probably a lot more younger folks using YouTube than oldsters.
These aren't people making the things "wholesome and funny". They are criminals in teddy bear outfits controlling what people hear. Modern day clowns.
i'm here to tell you that the cutesy shit drives certain people away.
'Funny' isn't some objective quality that things can be neatly sorted into with any level of accuracy, except in cases of the worst most generic/juvenile types of humor that are guaranteed to get a chuckle from 'some idiot' out there.
The parody show 'Ow! My Balls!' in the movie 'Idiocracy' comes to mind.
I believe this was when youtube was growing.
> it's trying to build an experience that doesn't actively turn people away
I believe youtube was growing at this time.
I personally don't remember it being that toxic. But even if it were whose responsibility should it be to deal with the toxicity? The platform's or the content creator's?
There's a saying that's oft repeated "dont read the comments section". That's a strategy in line with the approach I'd endorse. We shouldn't be censoring information just because its "toxic", people need to find ways to withstand or ignore or deal with the criticism that exists in the world.
People develop weak immune systems if they are protected from everything, it's not just a biological thing. The American mind is as coddled as their biological immune system. We should be teaching people how to deal and withstand criticism whether valid or caustic. Otherwise we get this mad world where no one can offend anyone and you can't even tell clinically obese people their body is not healthy.
Wrong way to frame the incentives. "Who benefits if the toxicity is dealt with?" is the relvant question.
(... and the answer is "the platform owner, assuming they attract more viewers with honey than acid." So, a predictable outcome if the premise holds).
But moralistically I'm still at odds with the coddling.
There's a lot of people willing to trade their hard-earned valuables (coin, time, what have you) for a little gentleness and softness. I don't see that as a moral ill.
The world is sharp, cold, apathetic, and dangerous. People know. That doesn't imply people need to live in that part all the time if circumstances do not force them to.
In fact, one could argue most of the last ten-odd-thousand years of building societies has been a slow crawl towards figuring out how to make something less sharp, cold, apathetic, and dangerous for people to spend their days in.
Toxic racists in YouTube comments did not provide any value to anyone, but they did create an environment that encouraged others to behave in toxic ways and they diminished the enjoyment many others derived from the platform.
They made their presence known. Thats valuable in and of itself. If people disagree with them and they are the minority, other commenters can then throw incendiary feedback at them. Others can then watch the debate and learn how to take apart these toxic rascists using nothing but words.
> diminished the enjoyment
Ignorance is bliss, I'll give you that much. Most take the blue pill for a good reason.
Yeah, because arguing with loud racists is a good use of time.
It is, hypothetically, possible that one will find fringe opinions there more often because those on the fringe tend to be detached from other uses of time (i.e. if you're unemployed, you have a lot of time to argue with strangers online, for example).
Then you're probably not remembering it right. Its toxicity levels were off the charts. I'd go out of my way to block the comments just so I couldn't see them when a video would be loading and have them ruin my day.
> But even if it were whose responsibility should it be to deal with the toxicity? The platform's or the content creator's?
Well, as they say, "my house, my rules". It's their house :-)
> People develop weak immune systems if they are protected from everything, it's not just a biological thing.
> We should be teaching people how to deal and withstand criticism whether valid or caustic.
Nobody can withstand internet criticism. We weren't built for this. We were built for a tribe of apes around us, most of which would leave us alone to not disturb the social hierarchy. There's no psyche on this planet that can withstand a torrent of crap thrown onto it from all over the world. See articles about Facebook moderators for more details (those poor souls!).
Most jurisdictions around the world have laws against verbal assault (threatening others with actual harm) or verbal harassment, for this very reason. Verbal interactions can lead to very real effects (people have been driven to suicide, for example).
Some people we interact with are just cuckoo and if we can't get out of there quickly there needs to be a way to keep the situation under control.
We sometimes like to see ourselves as little Sherlock Holmes' - humans on a trajectory of ever more self-improvement and breathtaking intellect. All data is good data, all speech is good speech, because it feeds the ever improving intellect. Misinformation is at worst a temporary blip that more communication will correct. We're little Googles that slurp up data and spit out ever improving judgements.
But we're not that rational, and not that great at filtering information. We have biases that are trivially exploitable at scale, and we're rather good at building machines or even merely memes to exploit those.
We need to start accepting human cognitive weaknesses, and that includes doing a bit of informational gardening - pruning out the weeds before they choke everything we value. You don't feed ML algorithms maliciously distorted training data and hope for them to nevertheless magically converge to something useful, do you? (And no, GAN ML doesn't do that either yet - but please correct me if I'm behind the times - because the detectors aren't really adversarial but rather specific training targets and thus a detector from one GAN can't detect another's fakes - and in any case, it's all still based on trusted training data - which is exactly what humans don't have).
I don't believe this problem is fixable until we recognize that freedom of speech in an era of mass manipulation via social media is intrinsically dangerous. We all understand the upsides and have been well indoctrinated against risks of government ministries for truth, but the reality is that we're all agents in an immensely complex system where we want certain convergence properties, yet are kind of assuming that specific hyper simplistic policies will achieve those goals - and assuming that almost as gospel. I don't buy that that's going work. I don't think it actually _ever_ worked, which is why we have stuff like fraud statutes, and _do_ penalize lying in all kinds of other scenarios. Even more critically, social _norms_ imposed heavy extra-legal punishment on frauds and other informational polluters. We just turned extreme freedom of expression into an article of patriotism to show we were team america; and to distinguish ourselves from team soviet or team monarchy - but that was (fortunately!) merely skin deep because various legal restrictions actually remained and in any case social punishments matter more than legal ones - and being seen as a fraud or liar _used_ to be a quick way to becoming a pariah.
But without speech-restricting social norms against deception, and with a public attitude that sometimes even _embraces_ falsehood as a symbol of pride for ones rights - well, then truth is optional; and quixotically the _whole point_ of freedom of expression disappears, being that it's a tool to discover and disseminate new insights and burn away corruption via transparency. Speech is irrelevant when the only people listening are those that share your opinion anyhow. And where social norms remain, we've pulled their teeth by accepting that employees have no responsibility for a corporation's actions unless they're very directly linked (e.g. I wouldn't blame a facebook employee personally for facebook-spread misinformation, yet I would if they spread it themselves), yet corporations are increasingly everpresent as tools to mediate daily interactions.
Until we face up to the reality that core founding principles we hang some of our identities on are _actively harmful_ in some scenarios, we're not going to be able to consider how to improve ourselves. Instead, we'll tell ourselves comforting bedtime tales of how people can teach themselves to fend these things off, and how team free speech always wins in the end. Until, one day, it doesn't.
> teach themselves to fend these things off, in fact I think its the only thing that works.
“But you are telling me, Susan, that the ‘Society for Humanity’ is right; and that Mankind has lost its own say in its future.”
“It never had any, really. It was always at the mercy of economic and sociological forces it did not understand — at the whims of climate, and the fortunes of war. Now the Machines understand them; and no one can stop them, since the Machines will deal with them as they are dealing with the Society, — having, as they do, the greatest of weapons at their disposal, the absolute control of our economy.”
“How horrible!”
“Perhaps how wonderful! Think, that for all time, all conflicts are finally evitable. Only the Machines, from now on, are inevitable!”
And the fire behind the quartz went out and only a curl of smoke was left to indicate its place.
"""
Asimov, Isaac. "I, Robot". pp. 147-148. Available online: http://ekladata.com/-Byix64G_NtE0xI4A6PA1--o1Hc/Asimov-Isaac...
I get the feeling (and that's fine!) you're uninterested in detailed argumentation whether in support of your position or analysis of mine; I won't try to drag you into a flame war - but I simply don't understand your response. I must have said something that triggered this belief; if you're willing to point out what I said that could be read that way, I'd appreciate it. After all, being misunderstood is a waste everyone's time.
My immediate reaction was "You could... But why do you want to? In what scenario is it going to benefit your fledgling social network to chug toxic waste?"
... apparently, it didn't.
It’s a no-win if you demand that comments go largely unmoderated. Because we had that and it made them genuinely useless to the point where creators turned them off and people installed add-ons to remove them.
That's better than one that is useful, but useful for only one affiliation.
I find the wholesome and funny comments to be relatively uninteresting and dull 90% of the time, but it’s leaps and bounds better than the unmitigated angry mud slinging that used to be there.
These comics unfortunately both highlight the need for toxicity and dissenting opinions.
In the first, moon landing conspiracists have come out with great suspicious findings including the fact that the "moon rock" given to the dutch is actually just a peice of petrified wood [1]. Given the political need to appear more technologically advanced during the cold war its a very real possibility. I'm not saying the moon landings were fake, but to quash any disagreement if the intent is to be as truthful as possible would be a mistake.
In the second, the user gets valuable feedback that he's a moron, which he would not if he were censored. He would only dive deeper into anger if he found out he was silenced.
I don't let this "toxicity" get to me, but that means I'm not outsourcing a machine to protect me from it. Not only does facing more toxicity build up my own immune system and make me stronger, I'm not dancing with like minded people circle jerking a helicopter into the clouds of confirmation biases. Doesn't matter what the attack is, its actually good to have some of it around. Learning to deal with it makes people more resilient. We shouldn't be trying to get machines to do it for us, we should be making it the standard rhetoric that people should lean into criticisms consuming it to strengthen themselves.
Moron.
Another analogy is that we need to have managed forest fires, instead of trying to quash every little fire at every turn which only results in a catastrophic blaze when one turns up that is out of control.
If you’re stupid and can’t provide for yourself pissing others off is a quick way to starve.
I hate bullies as much as the next person on HN, but I'm deeply unsettled by youtube's recent efforts in the direction of manufacturing consent.
On the other hand it is nice that their comment section isn't as toxic as it once was.
We have restricted the constitutional rights of corporations in other cases e.g. utilities, though. And we can do it again.
Discrimination by companies toward the public is covered mainly by Civil Rights Act (or the Americans with Disabilities Act in the case of discrimination based on disability).
Okay, but that's not the first amendment. That's the Civil Rights Act, IIRC. And I think it'd be difficult to prove discrimination in this case.
There may have been the word shit in my comment. But it was polite and constructive. Meh. Makes me not want to bother and I don't.
Sorry about the flippant tone, but I hope we've learned that any system like this operating behind closed doors can be and will be used for manipulation and (much more direct) profit-seeking.
And even before that happens, the idea itself is full of cultural bias, and so brings in cultural discrimination and the imposition of one culture's idea of wholesome/funny on the whole world, by its very nature.
This is a massive improvement on what came before. Maybe it could be better yet, but I haven't seen anything better myself for websites in a similar situation.
Likes and similar, fine. Maybe evolve that. Maybe do the whole Dark Souls thing where you can only leave gestures and stock messages. Sure, you'll still get "try finger, but hole" and "amazing chest ahead", but that's still streets ahead of Youtube's communication.
Obviously YouTube probably values having comments, but beyond them I’m betting creators do too. How do you respond to them wanting to keep comments?
You can create vanishing messages even without youtube providing the service. Write a comment on a piece of toiletpaper and flush it down.
It probably also means people being more engaged with said youtuber if they feel like they can talk to them.
The homogeneity of the internet is becoming so tiresome.
You’re also trapped in your own eternal September, doomed to remember your own personal “early” internet as being so much better because it was all new to you despite the fact that it was someone else’s “oh my god it’s all the same now.”
No you couldn't. Because back then, even 10-15 years after 1995, not everyone went to the same 3 websites, and there were more than 3 companies on the web that owned everything. Furthermore, there was no concept of upvotes and downvotes to literally encourage crafting your comments to appeal to the widest audience / lowest common denominator. (This concept is no longer specific to Reddit by the way; nearly every comment section on every website utilizes the same type of deal.)
Back then if you wanted to speak your mind or say something, you could just type it out and everyone would see it. Today, if you want to say something, you need to craft your comment in such a way to make it worth it: sure, you could say exactly what you want. But that might only get one upvote. Or, God forbid, it might get downvoted! Then nobody sees it; furthermore, the rating your receive could potentially work against you and make the hivemind turn further against your argument. Because lots of people downvoted it. Or not many people upvoted it. Which means it's bad / not good.
It's a perverse incentive. The result is... usually, milquetoast homogeneous repetition.
Absolutely nobody:
OP: I'm TiReD oF THe SamE joKEs!1!
(On a serious note, I totally agree with you.)
Nobody:
X: Y
Does not compute for me though as "nobody saying nothing" would mean that literally everyone would say at least something, making it anything but unprovoked. But at least I can stop trying to make sense of it now.
Of course it's always fine to link to an archived copy from the comments, and we intend eventually to build software support for this.