8 ms·

The problem of web content is the search and recommendation systems have no opinion on skill, effort, scholarship or accuracy. It's become disconnected so much

The problem of web content is the search and recommendation systems have no opinion on skill, effort, scholarship or accuracy. It's become disconnected so much that metrics of quality are seen as arbitrary, unrelated attributes. There is literally zero reward for hard work. It's actually more likely to hurt than help you because original scholarship or novel insights is definitionally more dissimilar to existing things than low effort derivative spam. There needs to be a fundamental redress that combines with categorical human-based systems. Categorical not only in what they "like" but in their modalities of engagement and what the styles of flow mean. For example, if a 4 year old watches a video 150 times over 2 weeks, they're being a 4 year old. If a 40 year old does that exact same thing, then it's a completely different meaning. There's other modalities, do people complete it but take longer then the duration (maybe to look things up) or do they set it at 1.5 speed and scrub through it. There's ways of characterizing these interactions. Choosing to be naive about such things in our search and recommendation systems is at our own peril and it's a big part of the SEO spam and low quality avalanche that we are all awash in. Digging ourselves out of it will require us to somehow capture a human part of human beings we haven't appeared to digitally organize yet.

onion2k 4y ago
The problem of web content is the search and recommendation systems have no opinion on skill, effort, scholarship or accuracy. They do, but only through the proxy of 'likes' and 'followers'. This leads to a bigger problem with web content, especially with video - people conflate the quality of the video production with the quality of the content. Someone who can make slick, glossy videos full of effects will almost always have a far larger following than someone who makes videos with better content but worse visuals. The same goes for a well-designed website. People literally believe that the loading speed of a webpage has an impact on the quality of what it says. You could put 30 years into learning, understanding, and explaining a complex topic, but if you can't use AffectEffects or you haven't paid attention to your Core Web Vitals the chances are that YouTube and Google are going to rank you lower than someone who can.
- Helmut10001 4y ago
  It is funny, but most of these points almost equally apply to research output.
- mschuster91 4y ago
  > People literally believe that the loading speed of a webpage has an impact on the quality of what it says. Or if there is a progress bar [1], which was discussed on HN back then [2]. The top comment in the HN thread is so utterly ridiculous it hurts. [1] https://www.theatlantic.com/technology/archive/2017/02/why-some-apps-use-fake-progress-bars/517233/ https://www.theatlantic.com/technology/archive/2017/02/why-s... [2] https://news.ycombinator.com/item?id=16289380 https://news.ycombinator.com/item?id=16289380
- formerly_proven 4y ago
  On average production quality seems to correlate inversely with accuracy and actual knowledge on Youtube. It kinda makes sense, most Youtube channels seem to be one-man-shows, if the production quality is good, the dude doing it is basically an A/V person at that point and you're looking at P(A/V person AND knowledgeable in X) vs. P(knowledgeable in X).
  - bayindirh 4y ago
    Considering the channels I follow, I can't see the correlation you mention, or we have a different definition for "production quality".
    - formerly_proven 4y ago
      > Considering the channels I follow No way for that sample to be biased.
      - bayindirh 4y ago
        It can be completely biased. This is why my definition is probably different than yours in the first place. On the other hand, your sample may be also biased, so it goes both ways.
techpression 4y ago
As someone who does photography on the side this rings very true for me (regarding quality of work). Instagram has steadily been showing my content less and less (based on their own analytics) and I see less and less of who I follow in favor of content aggregation accounts (that I don’t follow). In the extreme case I’m actually more likely to see content by someone I follow via an aggregation account then the actual author (I do follow both in one rare occasion). Weirdly enough I find recommendation systems getting worse and worse when it comes to content on YT/Instagram, shouldn’t we be seeing progress, it would benefit all parties involved.
- kuu 4y ago
  > Weirdly enough I find recommendation systems getting worse and worse Most likely because spammers are getting better and better to exploit them. I'm not saying that they (YT/IG) couldn't do better, but it's a two players competition: the recommendation system vs spammers.
trh0awayman 4y ago
Recommendation systems just mirror what the users watch - including their age - and the reality is that 99% of people have horrible taste in just about everything. They can't tell the difference between a great film and a terrible film - they walk away thinking "I enjoyed that".
- jopsen 4y ago
  > they walk away thinking "I enjoyed that". I really enjoyed reading your comment :)
  - trh0awayman 4y ago
    My point wasn't to denigrate people for having poor taste or whatever. I don't care what people watch. And I'm not one to talk - I sometimes think of buying one of those McMansions from McMansion Hell. But my point was that the systems are working as intended, and that there is no real failure here - people are seeing a reflection of their peers. It was hard for me to understand it until I saw how my Zoomer cousin watched Youtube like it was a TV station. Youtubers aren't "gaming" any sort of system, just like SEO isn't really "gaming" Google (for the most part) - they're both gaming people. SEO tricks work because it gets clicks. Clickbait titles have worked for 10+ years now because it STILL gets clicks.
    - ghusbands 4y ago
      >>> 99% of people have horrible taste in just about everything > My point wasn't to denigrate people for having poor taste These are directly contradictory. "horrible taste" is a denigrating term.
      - michaelt 4y ago
        Perhaps he means "I don't denigrate people for having poor taste - because in my view, there's no such thing"
- jahnu 4y ago
planb 4y ago
I don't think that's the underlying problem. The youtube algorithm sees people searching for something, watching a video and then liking it. How should it not conclude that the content was useful for the watcher? Probably it even _was_ useful and gave them the info needed to solve their problem. So why did they consume this content on YouTube and not on Stack Overflow? Because that's the way they are used to searching for answers. Probably because they feel videos are in general easier to understand. Or they come from a background where YouTube videos usually provide the best quality answers (home improvement comes to mind immediately). For those of us that know how these videos are made, they feel like spam at best and a giant scam that rips off Stack Overflow at worse (though this use is allowed by the license), but I don't see what incentives Youtube should have to ban them.
- NonNefarious 4y ago
  "So why did they consume this content on YouTube and not on Stack Overflow? Because that's the way they are used to searching for answers" No, it's because they searched on a technical topic and the Stack Overflow hits ranked lower than this spam. There's a shitload of spam-bucket, search-trap sites that scrape Stack Overflow and re-host its text in a column full of ads. I mean... pages of hits of them. Now we have auto-generated videos doing it. Again, Google's vaunted algorithms don't do shit about it; even worse, Google REMOVED the ability to block specific sites from search results. Despicable.
  - spywaregorilla 4y ago
    Yeah. On top of that google seems to prefer giving videos, categorically. Youtube results are shown first for many, but not all, types of queries. Not entirely clear what the distinction is. But it's not like it alternates video, web link, video, implying that they're sorted in terms of match rate even ignoring gross SEO. It's just video first.
  - planb 4y ago
    100% agreed on the decline of Google's algorithm. Those SO rip off sites make me furious, too [1]! But people know they are clicking on a video link and do so not just because it's ranked higher than the text ones, but because they feel a video is easier to understand. [1] https://news.ycombinator.com/item?id=26436490 https://news.ycombinator.com/item?id=26436490
SilverBirch 4y ago
I agree with you that there's a problem, but I think your solutions are kind of naive. All the logic you talk about in your proposed solution can be done with data that youtube has. In almost all cases it has a basic idea of who you are and how you view the videos. It could implement what you suggest tomorrow. In fact I suspect that it probably does do something like you suggest. The core of the problem is more about incentives and resources than the actually algorithm. Your incentive is to find informative or entertaining content, you have about 10% of an FTE to achieve that. Youtube's incentive is monetize your attention and has ~2000 FTEs to achieve this. Creators are incentivized to try and maximize the amount of Youtube's monetization they get a share of about ~1,000,000 FTEs to achieve this. That last category is in an adversarial relationship with the first two categories and they have way more resources. Even if Youtube's incentive was aligned with yours - which it isn't - the content creators are in a battle to the death to optimize how much share of the money they get which means they're incentivized to do almost the exact opposite of what you want.
themitigating 4y ago
Isn't the sort of data or at least similar already used for recommendations? I agree a large portion of the internet is garbage but there's always Wikipedia.
christophilus 4y ago
> There is literally zero reward for hard work. There’s truth in what you say, but there are plenty of counter examples, too. The Smarter Every Day channel on YouTube, for example. That guy works his ass off, and that hard work is a big part of the reason he’s successful.
lr4444lr 4y ago
That's what PageRank was supposed to do, though. 4 year olds may be WATCHING in volume, but they're not LINKING content. Links are the lifeblood of the web, a point majorly abused in the Web2/SPA fad over the last 10 years or so.
jrh206 4y ago
YouTube have begun asking people for their opinion on videos in quick surveys. Every so often (for me, every 2 days or so) there is a blue prompt with a video you watched recently. "What did you think of this video?" You're asked for a rating from 1 to 5, and then you can provide a little more data to contextualise your rating. For instance, if you vote 5, you're presented with a list of adjectives like "heart-warming", "motivational", "entertaining", etc, and you can choose as many as you like. If you vote 1, you're asked "did you regret watching this video?" YouTube are optimising for user time on site, not time on video, so they use these surveys to steer people away from videos that you might binge but regret, and leave YouTube. Right now, at least, the surveys have an enormous impact on which videos are pushed by the algorithm. I use the surveys personally to tailor my recommendations (and try to push creators who I think deserve more views). That is, I basically answer them honestly.
Stunting 4y ago
"There is literally zero reward for hard work. It's actually more likely to hurt than help you because original scholarship or novel insights is definitionally more dissimilar to existing things than low effort derivative spam." I use social media quite effectively within my industry circles to promote myself and stay relevant. It's a grind, but 100% get a ROI in the form of brand awareness leading to more work opportunities. I disagree that the formula "make something good and reap rewards" has ever been effective outside of lucky breaks. If the reward of making something good is not enough for an individual, then extra effort is required in the form of market research, promotion, and sales. The stuff you're seeing seem to exploit the back half, but I have always found that making something good and doing any marketing at all has huge benefits.
dredmorbius 4y ago
It's the "Market for Lemons" paradox --- itself a form of Gresham's Law. High-effort and low-effort content receive ... roughly ... similar compensation as YouTube videos. Or, for that matter, much online Web content. Given the costs of high-effort media are vasstly greater, the market "prefers" low-effort, high-volume crap. A market solution would be to effectively bid far less for the crap. The problem is that assessing content quality is itself expensive and error-prone. This is amongst the reasons the publishing world evolved toward publishers and editors, who carry a long-term reputation and face reputational damage from fielding crap. In cases such as YouTube, Google Web Search, and the Amazon marketplace, this reputational harm has been escallating and catching up with the "publisher", in this case, Google's YouTube subsidiary, Google Web Search itself, and Amazon. Much of the dysfunction of skilled labour markets follows a similar pattern. https://viterbi-web.usc.edu/~shaddin/cs590fa13/papers/AkerlofMarketforLemons.pdf https://viterbi-web.usc.edu/~shaddin/cs590fa13/papers/Akerlo... https://en.wikipedia.org/wiki/The_Market_for_Lemons https://en.wikipedia.org/wiki/The_Market_for_Lemons
CWuestefeld 4y ago
The problem of web content is the search and recommendation systems have no opinion on skill, effort, scholarship or accuracy. I'm OK with the rest of what you say, but the portion about "effort" seems to be echoing Marx's labor theory of value. This has been debunked any number of ways. But I think for the population here, used to building things as efficiently as we can, it's perverse to try to reward someone for working harder, i.e., doing things the slow and inefficient way. Given a product equal in the other measures - skill, scholarship, accuracy, and probably some other variables as well, we should probably prefer the one that took less effort to achieve that bar.
TheAceOfHearts 4y ago
videogamedunkey had a great video making fun of the issue of content spam vs quality content [0]. It takes a ton of time and effort to create new content. Meanwhile some people just make low-effort reaction videos every day, getting way more views. [0] https://youtu.be/VZzZKuQUguk https://youtu.be/VZzZKuQUguk
dopa42365 4y ago
hard work != good content by default If something is highly popular, no matter how utterly trash it is in your opinion, they must be doing something correctly. The right content at the right time for the right audience.

onion2k 4y ago

The problem of web content is the search and recommendation systems have no opinion on skill, effort, scholarship or accuracy. They do, but only through the proxy of 'likes' and 'followers'. This leads to a bigger problem with web content, especially with video - people conflate the quality of the video production with the quality of the content. Someone who can make slick, glossy videos full of effects will almost always have a far larger following than someone who makes videos with better content but worse visuals. The same goes for a well-designed website. People literally believe that the loading speed of a webpage has an impact on the quality of what it says. You could put 30 years into learning, understanding, and explaining a complex topic, but if you can't use AffectEffects or you haven't paid attention to your Core Web Vitals the chances are that YouTube and Google are going to rank you lower than someone who can.

Helmut10001 4y ago
It is funny, but most of these points almost equally apply to research output.
mschuster91 4y ago
> People literally believe that the loading speed of a webpage has an impact on the quality of what it says. Or if there is a progress bar [1], which was discussed on HN back then [2]. The top comment in the HN thread is so utterly ridiculous it hurts. [1] https://www.theatlantic.com/technology/archive/2017/02/why-some-apps-use-fake-progress-bars/517233/ https://www.theatlantic.com/technology/archive/2017/02/why-s... [2] https://news.ycombinator.com/item?id=16289380 https://news.ycombinator.com/item?id=16289380
formerly_proven 4y ago
On average production quality seems to correlate inversely with accuracy and actual knowledge on Youtube. It kinda makes sense, most Youtube channels seem to be one-man-shows, if the production quality is good, the dude doing it is basically an A/V person at that point and you're looking at P(A/V person AND knowledgeable in X) vs. P(knowledgeable in X).
- bayindirh 4y ago
  Considering the channels I follow, I can't see the correlation you mention, or we have a different definition for "production quality".
  - formerly_proven 4y ago
    > Considering the channels I follow No way for that sample to be biased.
    - bayindirh 4y ago
      It can be completely biased. This is why my definition is probably different than yours in the first place. On the other hand, your sample may be also biased, so it goes both ways.

techpression 4y ago

As someone who does photography on the side this rings very true for me (regarding quality of work). Instagram has steadily been showing my content less and less (based on their own analytics) and I see less and less of who I follow in favor of content aggregation accounts (that I don’t follow). In the extreme case I’m actually more likely to see content by someone I follow via an aggregation account then the actual author (I do follow both in one rare occasion). Weirdly enough I find recommendation systems getting worse and worse when it comes to content on YT/Instagram, shouldn’t we be seeing progress, it would benefit all parties involved.

kuu 4y ago
> Weirdly enough I find recommendation systems getting worse and worse Most likely because spammers are getting better and better to exploit them. I'm not saying that they (YT/IG) couldn't do better, but it's a two players competition: the recommendation system vs spammers.

trh0awayman 4y ago

Recommendation systems just mirror what the users watch - including their age - and the reality is that 99% of people have horrible taste in just about everything. They can't tell the difference between a great film and a terrible film - they walk away thinking "I enjoyed that".

jopsen 4y ago
> they walk away thinking "I enjoyed that". I really enjoyed reading your comment :)
- trh0awayman 4y ago
  My point wasn't to denigrate people for having poor taste or whatever. I don't care what people watch. And I'm not one to talk - I sometimes think of buying one of those McMansions from McMansion Hell. But my point was that the systems are working as intended, and that there is no real failure here - people are seeing a reflection of their peers. It was hard for me to understand it until I saw how my Zoomer cousin watched Youtube like it was a TV station. Youtubers aren't "gaming" any sort of system, just like SEO isn't really "gaming" Google (for the most part) - they're both gaming people. SEO tricks work because it gets clicks. Clickbait titles have worked for 10+ years now because it STILL gets clicks.
  - ghusbands 4y ago
    >>> 99% of people have horrible taste in just about everything > My point wasn't to denigrate people for having poor taste These are directly contradictory. "horrible taste" is a denigrating term.
    - michaelt 4y ago
      Perhaps he means "I don't denigrate people for having poor taste - because in my view, there's no such thing"
jahnu 4y ago

planb 4y ago

I don't think that's the underlying problem. The youtube algorithm sees people searching for something, watching a video and then liking it. How should it not conclude that the content was useful for the watcher? Probably it even _was_ useful and gave them the info needed to solve their problem. So why did they consume this content on YouTube and not on Stack Overflow? Because that's the way they are used to searching for answers. Probably because they feel videos are in general easier to understand. Or they come from a background where YouTube videos usually provide the best quality answers (home improvement comes to mind immediately). For those of us that know how these videos are made, they feel like spam at best and a giant scam that rips off Stack Overflow at worse (though this use is allowed by the license), but I don't see what incentives Youtube should have to ban them.

NonNefarious 4y ago
"So why did they consume this content on YouTube and not on Stack Overflow? Because that's the way they are used to searching for answers" No, it's because they searched on a technical topic and the Stack Overflow hits ranked lower than this spam. There's a shitload of spam-bucket, search-trap sites that scrape Stack Overflow and re-host its text in a column full of ads. I mean... pages of hits of them. Now we have auto-generated videos doing it. Again, Google's vaunted algorithms don't do shit about it; even worse, Google REMOVED the ability to block specific sites from search results. Despicable.
- spywaregorilla 4y ago
  Yeah. On top of that google seems to prefer giving videos, categorically. Youtube results are shown first for many, but not all, types of queries. Not entirely clear what the distinction is. But it's not like it alternates video, web link, video, implying that they're sorted in terms of match rate even ignoring gross SEO. It's just video first.
- planb 4y ago
  100% agreed on the decline of Google's algorithm. Those SO rip off sites make me furious, too [1]! But people know they are clicking on a video link and do so not just because it's ranked higher than the text ones, but because they feel a video is easier to understand. [1] https://news.ycombinator.com/item?id=26436490 https://news.ycombinator.com/item?id=26436490

SilverBirch 4y ago

I agree with you that there's a problem, but I think your solutions are kind of naive. All the logic you talk about in your proposed solution can be done with data that youtube has. In almost all cases it has a basic idea of who you are and how you view the videos. It could implement what you suggest tomorrow. In fact I suspect that it probably does do something like you suggest. The core of the problem is more about incentives and resources than the actually algorithm. Your incentive is to find informative or entertaining content, you have about 10% of an FTE to achieve that. Youtube's incentive is monetize your attention and has ~2000 FTEs to achieve this. Creators are incentivized to try and maximize the amount of Youtube's monetization they get a share of about ~1,000,000 FTEs to achieve this. That last category is in an adversarial relationship with the first two categories and they have way more resources. Even if Youtube's incentive was aligned with yours - which it isn't - the content creators are in a battle to the death to optimize how much share of the money they get which means they're incentivized to do almost the exact opposite of what you want.

themitigating 4y ago

Isn't the sort of data or at least similar already used for recommendations? I agree a large portion of the internet is garbage but there's always Wikipedia.

christophilus 4y ago

> There is literally zero reward for hard work. There’s truth in what you say, but there are plenty of counter examples, too. The Smarter Every Day channel on YouTube, for example. That guy works his ass off, and that hard work is a big part of the reason he’s successful.

lr4444lr 4y ago

That's what PageRank was supposed to do, though. 4 year olds may be WATCHING in volume, but they're not LINKING content. Links are the lifeblood of the web, a point majorly abused in the Web2/SPA fad over the last 10 years or so.

jrh206 4y ago

YouTube have begun asking people for their opinion on videos in quick surveys. Every so often (for me, every 2 days or so) there is a blue prompt with a video you watched recently. "What did you think of this video?" You're asked for a rating from 1 to 5, and then you can provide a little more data to contextualise your rating. For instance, if you vote 5, you're presented with a list of adjectives like "heart-warming", "motivational", "entertaining", etc, and you can choose as many as you like. If you vote 1, you're asked "did you regret watching this video?" YouTube are optimising for user time on site, not time on video, so they use these surveys to steer people away from videos that you might binge but regret, and leave YouTube. Right now, at least, the surveys have an enormous impact on which videos are pushed by the algorithm. I use the surveys personally to tailor my recommendations (and try to push creators who I think deserve more views). That is, I basically answer them honestly.

Stunting 4y ago

"There is literally zero reward for hard work. It's actually more likely to hurt than help you because original scholarship or novel insights is definitionally more dissimilar to existing things than low effort derivative spam." I use social media quite effectively within my industry circles to promote myself and stay relevant. It's a grind, but 100% get a ROI in the form of brand awareness leading to more work opportunities. I disagree that the formula "make something good and reap rewards" has ever been effective outside of lucky breaks. If the reward of making something good is not enough for an individual, then extra effort is required in the form of market research, promotion, and sales. The stuff you're seeing seem to exploit the back half, but I have always found that making something good and doing any marketing at all has huge benefits.

dredmorbius 4y ago

It's the "Market for Lemons" paradox --- itself a form of Gresham's Law. High-effort and low-effort content receive ... roughly ... similar compensation as YouTube videos. Or, for that matter, much online Web content. Given the costs of high-effort media are vasstly greater, the market "prefers" low-effort, high-volume crap. A market solution would be to effectively bid far less for the crap. The problem is that assessing content quality is itself expensive and error-prone. This is amongst the reasons the publishing world evolved toward publishers and editors, who carry a long-term reputation and face reputational damage from fielding crap. In cases such as YouTube, Google Web Search, and the Amazon marketplace, this reputational harm has been escallating and catching up with the "publisher", in this case, Google's YouTube subsidiary, Google Web Search itself, and Amazon. Much of the dysfunction of skilled labour markets follows a similar pattern. https://viterbi-web.usc.edu/~shaddin/cs590fa13/papers/AkerlofMarketforLemons.pdf https://viterbi-web.usc.edu/~shaddin/cs590fa13/papers/Akerlo... https://en.wikipedia.org/wiki/The_Market_for_Lemons https://en.wikipedia.org/wiki/The_Market_for_Lemons

CWuestefeld 4y ago

The problem of web content is the search and recommendation systems have no opinion on skill, effort, scholarship or accuracy. I'm OK with the rest of what you say, but the portion about "effort" seems to be echoing Marx's labor theory of value. This has been debunked any number of ways. But I think for the population here, used to building things as efficiently as we can, it's perverse to try to reward someone for working harder, i.e., doing things the slow and inefficient way. Given a product equal in the other measures - skill, scholarship, accuracy, and probably some other variables as well, we should probably prefer the one that took less effort to achieve that bar.

TheAceOfHearts 4y ago

videogamedunkey had a great video making fun of the issue of content spam vs quality content [0]. It takes a ton of time and effort to create new content. Meanwhile some people just make low-effort reaction videos every day, getting way more views. [0] https://youtu.be/VZzZKuQUguk https://youtu.be/VZzZKuQUguk

dopa42365 4y ago

hard work != good content by default If something is highly popular, no matter how utterly trash it is in your opinion, they must be doing something correctly. The right content at the right time for the right audience.