9 ms·

The problem with at least Facebook and Twitter (and possibly Google) in this analysis is that their attention-grabbing model requires them to analyse and interf

The problem with at least Facebook and Twitter (and possibly Google) in this analysis is that their attention-grabbing model requires them to analyse and interfere with the data which is coming through their systems. The only way not to do that would be for FB timelines and Twitter feeds to be strictly sequential feeds which at some point in the past they may have been but they are not now. Once you start basing your business model on selectively making some content more visible, you have already crossed a rubicon between a strictly neutral common carrier and a publisher. (There is an error in the quote where they use publisher when they mean common carrier, a publisher is precisely an organisation that does have full editorial control of what they put out which is the opposite of what they intend from the rest of the quote.

chr1 7y ago
They could make the algorithm customizable and let the user to control it. That would be much better than the current blackbox where you get uninteresting stories promoted because of misclicking once, and have no way to tell the algorithm about its mistakes.
- luckylion 7y ago
  > They could make the algorithm customizable and let the user to control it. It often gets it wrong, but that's the general idea, isn't it? You don't customize it by checking some boxes or adding some keywords, you customize it by giving attention to some results more than to others. Obviously, there's the added layer that they aren't optimizing for your satisfaction but for your business value to them.
jyrkesh 7y ago
Exactly. At face value, I'm firmly in favor of what Ravikant was saying, but even the simple example of a spam filter proves how it's impossible in practice: without basic spam filtering, something like Twitter would be absolutely unusable in the face of endless crypto and malware spam. At some level, Twitter/FB/et al have to do SOMETHING to shape/filter the volume of traffic they receive.
- rayiner 7y ago
  I don't think calls to regulate Twitter/FB would get much traction if all they were doing was spam filtering. To see it from a different direction: imagine the shit show that would happen if Google started "moderating" gmail content, even though it currently does very aggressive spam filtering.
  - tomp 7y ago
    Indeed. Spam filters, porn filters, swearword filters... all of those have precedents (in mainstream, non-social/internet media) and are perfectly acceptable by most of society. In addition, notice that even GMail doesn't actually filter spam. It just hides it - it's an assistant, not arbiter of what you should see. It's more than happy for you to adjust the filters (for yourself personally). Edit: GMail also does prioritization as of recently - but again, the purpose of that is to help users not filter content. Again, those algorithms are also adjustable for each user separately, they just have some "smart" defaults.
    - tambourine_man 7y ago
      Kind of. You can't turn their anti-spam off, you can't exclude addresses from it and, worse for me, Gmail can hard bounce if it doesn't like something in your server/IP/content in a way that's not fully documented or reproducible.
      - stubish 7y ago
        You can now. You can create a filter matching whatever criteria you want (including from:foo@example.com or a catch all), and specify matches 'never get flagged as spam'. Its not obvious, but it is there. And the usual limitations of filters apply, such as not being able to match arbitrary headers.
        krageon 7y ago
        Things can get flagged outside of your mailbox as so spammy that they will never arrive. This is how every spam filter works that I have seen to date and google is no exception.
ahnick 7y ago
I think it depends on how the recommendation algorithms are implemented. (I have no idea about how Facebook and Twitter recommendation algorithms work) Let's say for example you have a community contributed labeling system that allows users to label content. A recommendation algorithm might then be written that simply pulls the top 10 highest number of viewed pieces content that have the same labels as a piece of content the viewer just viewed. In this case, there is nothing in the algorithm that evaluates the content itself. The evaluation was done by users of the software, not the software itself.
- wmf 7y ago
  Simple "neutral" algorithms are generally easy to game. If you choose an algorithm that you already know will be gamed for harassment or whatever then you're essentially choosing that outcome.
  - ahnick 7y ago
    The labeling algorithm is just a relatively naive example. More complex "neutral" algorithms can be designed. Mitigations for abuse that are "neutral" themselves may be added before/after initial implementation. (e.g To extend the labeling example, a moderation system could be added that allows users to flag inappropriate labels.) Also, I don't think it is always possible to foresee in what ways a system/algorithm might be abused; therefore, it might be necessary to address said abuse once it reveals itself.
BlueTemplar 7y ago
People warned that this would happen when Facebook (YouTube, Steam, etc.) started sliding in this direction. And years before Facebook was even created, a middle way was also proposed where a carrier like Facebook wouldn't be legally responsible for the content they carried, unless they started to select and/or moderate that content, at which point they would lose the carrier status, becoming a simple broadcaster, which would be legally responsible for everything that they would help to broadcast.
- motivic 7y ago
  > unless they started to select and/or moderate that content, at which point they would lose the carrier status But since the homepage feed (or any medium really) displays contents in a certain order, some selection must take place. Typically some algorithm (usually a recommender system together with some business logic) is used to determine which contents from all that's available to you are actually shown to you and in what order. Bias seems to be an unavoidable part of the design to me.
  - roenxi 7y ago
    There are shades of grey. However, we have leaked footage of a Google co-founder saying (at an all-hands meeting no less) that the outcome of a democratic election conflicts with Google's values. There is a lot of room for interpretation there, but there are signals from Google in particular (eg, donation streams; leaked video; the occasional scandal bubbling out) that their management might be seeing the world through a partisan lens. Abstract ideas of unavoidable bias are only of academic interest; the right wing of politics is justified in seeing Google as a direct political threat. That would not be justified if Google had a strict "no political talk, no political campaigning, we are the Switzerland of the internet" style policy for their workplace.
    - m10i 7y ago
      Would you mind providing a source for this leak?
davidw 7y ago
The problem is also that they are advertisers, and I'll be damned if I want ads for my company appearing next to "Illinois Nazis" or ISIS or whatever other garbage. There's another problem: it becomes a race to the bottom if it's nothing but trolls and garbage content.
- mcny 7y ago
  > The problem is also that they are advertisers, and I'll be damned if I want ads for my company appearing next to "Illinois Nazis" or ISIS or whatever other garbage. I don't understand the problem. Why can't Toyota or Proctor and Gamble say it only wants to bid on a certain white list of channels or videos? The only downside is a smaller reach for the advertiser and a smaller "pie" for the "content" company (be it Instagram or YouTube). I think the main problem is the "content providers" be it Instagram or YouTube want to maximize the ad revenue and any suggestion I have such as strictly reverse chronological timeline by default goes against the idea of "growth hacking". I would personally love to be a fly in the wall in Netflix or Spotify headquarters. Because in my naive mind, Netflix makes MORE money when people subscribe to Netflix but never watch anything. But then if people watch Netflix a lot, they are more likely to talk about what they watch to others and encourage others to subscribe to Netflix? Any company probably wants to make sure they make more money the more customers they have? Kind of sounds obvious but I don't think it is the case with every company. I'd imagine there are some businesses where there is an ideal size and if you are bigger, the additional customers kind of follow a law of diminishing returns? I can't think of any examples but would love to hear if anyone reading this can...
- flukus 7y ago
  And the problem this creates for society is that we censor content that isn't advertiser friendly. We've already seen where this leads with broadcast television, lowest common denominator trash that is safe for advertisers, escaping that advertiser optimum made youtube good in the first place. I subscribe to a lot of history channels so I've seen the effects of this filtering directly, history contains a lot of talk about nazis and hitler and you can't even mention the big H without being demonetized. But this is just another example of how advertising is a cancer on society.
philwelch 7y ago
I don't think it's been convincingly argued that ranking or recommendation algorithms necessarily represent any sort of editorial bias. The purpose of these algorithms is to rank content based either on a prediction of what the individual user would want to see based on their existing history or on which content is more broadly engaging in general. In other words, if YouTube decides I really like to watch Ben Shapiro owning the libs with FACTS and LOGIC, that doesn't mean YouTube is making an editorial decision to promote that content, it just means YouTube has profiled me as some sort of conservative. If YouTube profiled me differently, they would be equally happy to recommend TYT videos or whatever. There may be some editorial bias introduced if the userbase of a given social media site is themselves skewed in a particular direction. For instance, there seems to be some skew in the content that shows up on my Tumblr dashboard that seems fairly irrespective of anything I've subscribed to. But even that isn't necessarily damning for Tumblr, because they're not necessarily making an editorial decision to promote that content.
- ikeyany 7y ago
  These platforms (these meaning the ones with the most market cap) need to be advertiser-friendly in order to grow. They inherently require editorial bias.
  - philwelch 7y ago
    I'm not an expert on advertising, but I suspect these concerns are often overblown. There are YouTubers who get demonetized--presumably because their content isn't "advertiser-friendly"--who still manage to get individual sponsorship deals. If these companies were serious about being a "common carrier", they would invest more money into developing an advertising market where advertisers could pick and choose the content they wanted to advertise and less money into demonetization.
    - afiori 7y ago
      demonetized video still get ads, it is a two tier system where ad space on demonetized videos is lower in price.
tomp 7y ago
No, if they instead worked to develop better user moderation tools, it would work just as well as it does now, just without their (appearance of) liability. Things like word filters, subscribable block lists, etc.
iamsb 7y ago
This is where twitter screwing over third party developers comes in to play. I would have preferred if twitter main feed was sequential and then there were thriving third party alternatives to it. Twitter could have easily monetised by taking a slice of ad revenue from third party developers, or by charging API fees to them.
dvfjsdhgfv 7y ago
> The problem with at least Facebook and Twitter (and possibly Google) in this analysis is that their attention-grabbing model requires them to analyse and interfere with the data which is coming through their systems. Not to mention a completely different topic - ads for scams. When I create a new ad on FB, they keep it in review for at least a couple of hours. But every day I see dozens of ads that are obviously dishonest. Some financial scams, ads for some fishy schemes, some of them badly translated. When you are a publisher and accept an ad, obviously it's very tempting to claim all responsibility is with the advertiser (it is not, although details vary between jurisdictions). But even if you didn't have any responsibility legally, it still feels terribly wrong from the ethical point of view to get rich by helping scammers cheat some poor chaps of their money.
- mlang23 7y ago
  Not on a daily basis, but I see ads for scams on youtube as well. Different topic, but the strangest thing I witnessed on youtube was ads for audible during a stolen audible book... So the owner of the content payed the thief (indirectly, but still) for placing an ad during performance of the stolen content. And youtube couldn't care less, as long as money moves...
uchman 7y ago
He oversimplified the issue. Phone companies don't make money through ads. FB/Twitter/Google make money by "getting into the weeds". They chose a business model with much more reward (hence much more risk). They're basically running ads businesses. You post about shoes, they sell you shoes. You talk about unemployment, they sell you a candidate ? Gitlab/Microsoft et al have slightly less dependent business models that aren't dependent on ads (for the most part).
pishpash 7y ago
Publishers don't exercise editorial control? Come on now.
evrydayhustling 7y ago
This business model is the reason that FB, T and G are dominant platforms worth discussing. Products that are purely transactional are easy to compete with, splitting up markets and driving down prices. This is why true "common carriers", like ISPs, have fought desperately against net neutrality and to capture more customer data -- they want in on the value capture that comes from shaping content, because they are rapidly being commoditized as carriers of bits. IMO this is a moat in the eye of a lot of libertarian tech philosophy. Most software products become competitive by building networks -- with end users (e.g. social networks), with developers (e.g. open source), or indirectly via content (from media aggregators to spam filtering). Once you're in the business of curating a network, you are implementing values whether you acknowledge them or not.

chr1 7y ago

They could make the algorithm customizable and let the user to control it. That would be much better than the current blackbox where you get uninteresting stories promoted because of misclicking once, and have no way to tell the algorithm about its mistakes.

luckylion 7y ago
> They could make the algorithm customizable and let the user to control it. It often gets it wrong, but that's the general idea, isn't it? You don't customize it by checking some boxes or adding some keywords, you customize it by giving attention to some results more than to others. Obviously, there's the added layer that they aren't optimizing for your satisfaction but for your business value to them.

jyrkesh 7y ago

Exactly. At face value, I'm firmly in favor of what Ravikant was saying, but even the simple example of a spam filter proves how it's impossible in practice: without basic spam filtering, something like Twitter would be absolutely unusable in the face of endless crypto and malware spam. At some level, Twitter/FB/et al have to do SOMETHING to shape/filter the volume of traffic they receive.

rayiner 7y ago
I don't think calls to regulate Twitter/FB would get much traction if all they were doing was spam filtering. To see it from a different direction: imagine the shit show that would happen if Google started "moderating" gmail content, even though it currently does very aggressive spam filtering.
- tomp 7y ago
  Indeed. Spam filters, porn filters, swearword filters... all of those have precedents (in mainstream, non-social/internet media) and are perfectly acceptable by most of society. In addition, notice that even GMail doesn't actually filter spam. It just hides it - it's an assistant, not arbiter of what you should see. It's more than happy for you to adjust the filters (for yourself personally). Edit: GMail also does prioritization as of recently - but again, the purpose of that is to help users not filter content. Again, those algorithms are also adjustable for each user separately, they just have some "smart" defaults.
  - tambourine_man 7y ago
    Kind of. You can't turn their anti-spam off, you can't exclude addresses from it and, worse for me, Gmail can hard bounce if it doesn't like something in your server/IP/content in a way that's not fully documented or reproducible.
    - stubish 7y ago
      You can now. You can create a filter matching whatever criteria you want (including from:foo@example.com or a catch all), and specify matches 'never get flagged as spam'. Its not obvious, but it is there. And the usual limitations of filters apply, such as not being able to match arbitrary headers.
      - krageon 7y ago
        Things can get flagged outside of your mailbox as so spammy that they will never arrive. This is how every spam filter works that I have seen to date and google is no exception.

ahnick 7y ago

I think it depends on how the recommendation algorithms are implemented. (I have no idea about how Facebook and Twitter recommendation algorithms work) Let's say for example you have a community contributed labeling system that allows users to label content. A recommendation algorithm might then be written that simply pulls the top 10 highest number of viewed pieces content that have the same labels as a piece of content the viewer just viewed. In this case, there is nothing in the algorithm that evaluates the content itself. The evaluation was done by users of the software, not the software itself.

wmf 7y ago
Simple "neutral" algorithms are generally easy to game. If you choose an algorithm that you already know will be gamed for harassment or whatever then you're essentially choosing that outcome.
- ahnick 7y ago
  The labeling algorithm is just a relatively naive example. More complex "neutral" algorithms can be designed. Mitigations for abuse that are "neutral" themselves may be added before/after initial implementation. (e.g To extend the labeling example, a moderation system could be added that allows users to flag inappropriate labels.) Also, I don't think it is always possible to foresee in what ways a system/algorithm might be abused; therefore, it might be necessary to address said abuse once it reveals itself.

BlueTemplar 7y ago

People warned that this would happen when Facebook (YouTube, Steam, etc.) started sliding in this direction. And years before Facebook was even created, a middle way was also proposed where a carrier like Facebook wouldn't be legally responsible for the content they carried, unless they started to select and/or moderate that content, at which point they would lose the carrier status, becoming a simple broadcaster, which would be legally responsible for everything that they would help to broadcast.

motivic 7y ago
> unless they started to select and/or moderate that content, at which point they would lose the carrier status But since the homepage feed (or any medium really) displays contents in a certain order, some selection must take place. Typically some algorithm (usually a recommender system together with some business logic) is used to determine which contents from all that's available to you are actually shown to you and in what order. Bias seems to be an unavoidable part of the design to me.
- roenxi 7y ago
  There are shades of grey. However, we have leaked footage of a Google co-founder saying (at an all-hands meeting no less) that the outcome of a democratic election conflicts with Google's values. There is a lot of room for interpretation there, but there are signals from Google in particular (eg, donation streams; leaked video; the occasional scandal bubbling out) that their management might be seeing the world through a partisan lens. Abstract ideas of unavoidable bias are only of academic interest; the right wing of politics is justified in seeing Google as a direct political threat. That would not be justified if Google had a strict "no political talk, no political campaigning, we are the Switzerland of the internet" style policy for their workplace.
  - m10i 7y ago
    Would you mind providing a source for this leak?

davidw 7y ago

The problem is also that they are advertisers, and I'll be damned if I want ads for my company appearing next to "Illinois Nazis" or ISIS or whatever other garbage. There's another problem: it becomes a race to the bottom if it's nothing but trolls and garbage content.

mcny 7y ago
> The problem is also that they are advertisers, and I'll be damned if I want ads for my company appearing next to "Illinois Nazis" or ISIS or whatever other garbage. I don't understand the problem. Why can't Toyota or Proctor and Gamble say it only wants to bid on a certain white list of channels or videos? The only downside is a smaller reach for the advertiser and a smaller "pie" for the "content" company (be it Instagram or YouTube). I think the main problem is the "content providers" be it Instagram or YouTube want to maximize the ad revenue and any suggestion I have such as strictly reverse chronological timeline by default goes against the idea of "growth hacking". I would personally love to be a fly in the wall in Netflix or Spotify headquarters. Because in my naive mind, Netflix makes MORE money when people subscribe to Netflix but never watch anything. But then if people watch Netflix a lot, they are more likely to talk about what they watch to others and encourage others to subscribe to Netflix? Any company probably wants to make sure they make more money the more customers they have? Kind of sounds obvious but I don't think it is the case with every company. I'd imagine there are some businesses where there is an ideal size and if you are bigger, the additional customers kind of follow a law of diminishing returns? I can't think of any examples but would love to hear if anyone reading this can...
flukus 7y ago
And the problem this creates for society is that we censor content that isn't advertiser friendly. We've already seen where this leads with broadcast television, lowest common denominator trash that is safe for advertisers, escaping that advertiser optimum made youtube good in the first place. I subscribe to a lot of history channels so I've seen the effects of this filtering directly, history contains a lot of talk about nazis and hitler and you can't even mention the big H without being demonetized. But this is just another example of how advertising is a cancer on society.

philwelch 7y ago

I don't think it's been convincingly argued that ranking or recommendation algorithms necessarily represent any sort of editorial bias. The purpose of these algorithms is to rank content based either on a prediction of what the individual user would want to see based on their existing history or on which content is more broadly engaging in general. In other words, if YouTube decides I really like to watch Ben Shapiro owning the libs with FACTS and LOGIC, that doesn't mean YouTube is making an editorial decision to promote that content, it just means YouTube has profiled me as some sort of conservative. If YouTube profiled me differently, they would be equally happy to recommend TYT videos or whatever. There may be some editorial bias introduced if the userbase of a given social media site is themselves skewed in a particular direction. For instance, there seems to be some skew in the content that shows up on my Tumblr dashboard that seems fairly irrespective of anything I've subscribed to. But even that isn't necessarily damning for Tumblr, because they're not necessarily making an editorial decision to promote that content.

ikeyany 7y ago
These platforms (these meaning the ones with the most market cap) need to be advertiser-friendly in order to grow. They inherently require editorial bias.
- philwelch 7y ago
  I'm not an expert on advertising, but I suspect these concerns are often overblown. There are YouTubers who get demonetized--presumably because their content isn't "advertiser-friendly"--who still manage to get individual sponsorship deals. If these companies were serious about being a "common carrier", they would invest more money into developing an advertising market where advertisers could pick and choose the content they wanted to advertise and less money into demonetization.
  - afiori 7y ago
    demonetized video still get ads, it is a two tier system where ad space on demonetized videos is lower in price.

tomp 7y ago

No, if they instead worked to develop better user moderation tools, it would work just as well as it does now, just without their (appearance of) liability. Things like word filters, subscribable block lists, etc.

iamsb 7y ago

This is where twitter screwing over third party developers comes in to play. I would have preferred if twitter main feed was sequential and then there were thriving third party alternatives to it. Twitter could have easily monetised by taking a slice of ad revenue from third party developers, or by charging API fees to them.

dvfjsdhgfv 7y ago

> The problem with at least Facebook and Twitter (and possibly Google) in this analysis is that their attention-grabbing model requires them to analyse and interfere with the data which is coming through their systems. Not to mention a completely different topic - ads for scams. When I create a new ad on FB, they keep it in review for at least a couple of hours. But every day I see dozens of ads that are obviously dishonest. Some financial scams, ads for some fishy schemes, some of them badly translated. When you are a publisher and accept an ad, obviously it's very tempting to claim all responsibility is with the advertiser (it is not, although details vary between jurisdictions). But even if you didn't have any responsibility legally, it still feels terribly wrong from the ethical point of view to get rich by helping scammers cheat some poor chaps of their money.

mlang23 7y ago
Not on a daily basis, but I see ads for scams on youtube as well. Different topic, but the strangest thing I witnessed on youtube was ads for audible during a stolen audible book... So the owner of the content payed the thief (indirectly, but still) for placing an ad during performance of the stolen content. And youtube couldn't care less, as long as money moves...

uchman 7y ago

He oversimplified the issue. Phone companies don't make money through ads. FB/Twitter/Google make money by "getting into the weeds". They chose a business model with much more reward (hence much more risk). They're basically running ads businesses. You post about shoes, they sell you shoes. You talk about unemployment, they sell you a candidate ? Gitlab/Microsoft et al have slightly less dependent business models that aren't dependent on ads (for the most part).

pishpash 7y ago

Publishers don't exercise editorial control? Come on now.

evrydayhustling 7y ago

This business model is the reason that FB, T and G are dominant platforms worth discussing. Products that are purely transactional are easy to compete with, splitting up markets and driving down prices. This is why true "common carriers", like ISPs, have fought desperately against net neutrality and to capture more customer data -- they want in on the value capture that comes from shaping content, because they are rapidly being commoditized as carriers of bits. IMO this is a moat in the eye of a lot of libertarian tech philosophy. Most software products become competitive by building networks -- with end users (e.g. social networks), with developers (e.g. open source), or indirectly via content (from media aggregators to spam filtering). Once you're in the business of curating a network, you are implementing values whether you acknowledge them or not.