#recommendationengines — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #recommendationengines, aggregated by home.social.
-
MegaLoL:
"Beijing has said the US spin-off of TikTok to be sold to American investors in a deal orchestrated by President Donald Trump will use parent company ByteDance’s Chinese algorithm.
Wang Jingtao, deputy head of China’s powerful cyber security regulator, on Monday said US and Chinese officials had agreed a framework that included “licensing the algorithm and other intellectual property rights”.
He said ByteDance would “entrust the operation of TikTok’s US user data and content security”, without elaborating.
Trump, who is expected to finalise the deal when he speaks to China’s President Xi Jinping on Friday, has again extended the deadline for the US app to be divested from its Chinese owner.
(...)
An Asia-based investor of ByteDance said the new US TikTok entity would use at least part of the Chinese algorithm but train it in the US on American user data.“Beijing’s bottom line is a licensing deal,” said the investor. “Beijing wants to be seen as exporting Chinese technology to the US and the world.”
“It’s the ultimate Taco trade,” said one US adviser close to the deal, referring to the acronym “Trump always chickens out”. “After all this, China keeps the algorithm.”
https://www.ft.com/content/550e4680-89e7-4b59-bb5d-2064cd6799c7
#USA #Trump #SocialMedia #TikTok #China #Algorithms #RecommendationEngines #ByteDance
-
MegaLoL:
"Beijing has said the US spin-off of TikTok to be sold to American investors in a deal orchestrated by President Donald Trump will use parent company ByteDance’s Chinese algorithm.
Wang Jingtao, deputy head of China’s powerful cyber security regulator, on Monday said US and Chinese officials had agreed a framework that included “licensing the algorithm and other intellectual property rights”.
He said ByteDance would “entrust the operation of TikTok’s US user data and content security”, without elaborating.
Trump, who is expected to finalise the deal when he speaks to China’s President Xi Jinping on Friday, has again extended the deadline for the US app to be divested from its Chinese owner.
(...)
An Asia-based investor of ByteDance said the new US TikTok entity would use at least part of the Chinese algorithm but train it in the US on American user data.“Beijing’s bottom line is a licensing deal,” said the investor. “Beijing wants to be seen as exporting Chinese technology to the US and the world.”
“It’s the ultimate Taco trade,” said one US adviser close to the deal, referring to the acronym “Trump always chickens out”. “After all this, China keeps the algorithm.”
https://www.ft.com/content/550e4680-89e7-4b59-bb5d-2064cd6799c7
#USA #Trump #SocialMedia #TikTok #China #Algorithms #RecommendationEngines #ByteDance
-
Netflix loves to recommend two kinds of shows to me:
1. Shows I have already watched start to finish.
2. Shows I started, didn't like, marked "not for me", and stopped watching.There's no customer-centric explanation for it.
They get my monthly payment regardless of how much I do or don't watch. Are these useless recommendations intended to suppress viewing time in order to maximize profit per subscriber free (or something)? 1/n
-
"The algorithmic recommender systems that select, filter, and personalize experiences across online platforms and services play a significant role in shaping user experiences online. These systems largely determine what users see, read, and watch, fueling debates around their potential to amplify harmful content, foster societal division, and prioritize engagement over user well-being. In reaction, some policymakers have turned to blanket bans on personalization or to the promotion of chronological feeds. But there are many better alternatives. Suggesting that users must choose between today’s default feeds and chronological or non-personalized feeds creates a false choice.
This report, prepared by the KGI Expert Working Group on Recommender Systems, offers comprehensive insights and policy guidance aimed at optimizing recommender systems for long-term user value and high-quality experiences. Drawing on a multidisciplinary research base and industry expertise, the report highlights key challenges in the current design and regulation of recommender systems and proposes actionable solutions for policymakers and product designers.
A key concern is that some platforms optimize their recommender systems to maximize certain forms of predicted engagement, which can prioritize clicks and likes over stronger signals of long-term user value. Maximizing the chances that users will click, like, share, and view content this week, this month, and this quarter aligns well with the business interests of tech platforms monetized through advertising. Product teams are rewarded for showing short-term gains in platform usage, and financial markets and investors reward companies that can deliver large audiences to advertisers."
https://kgi.georgetown.edu/research-and-commentary/better-feeds/
#SocialMedia #SocialNetworks #Algorithms #AlgorithmicRecommendation #RecommendationEngines
-
"The algorithmic recommender systems that select, filter, and personalize experiences across online platforms and services play a significant role in shaping user experiences online. These systems largely determine what users see, read, and watch, fueling debates around their potential to amplify harmful content, foster societal division, and prioritize engagement over user well-being. In reaction, some policymakers have turned to blanket bans on personalization or to the promotion of chronological feeds. But there are many better alternatives. Suggesting that users must choose between today’s default feeds and chronological or non-personalized feeds creates a false choice.
This report, prepared by the KGI Expert Working Group on Recommender Systems, offers comprehensive insights and policy guidance aimed at optimizing recommender systems for long-term user value and high-quality experiences. Drawing on a multidisciplinary research base and industry expertise, the report highlights key challenges in the current design and regulation of recommender systems and proposes actionable solutions for policymakers and product designers.
A key concern is that some platforms optimize their recommender systems to maximize certain forms of predicted engagement, which can prioritize clicks and likes over stronger signals of long-term user value. Maximizing the chances that users will click, like, share, and view content this week, this month, and this quarter aligns well with the business interests of tech platforms monetized through advertising. Product teams are rewarded for showing short-term gains in platform usage, and financial markets and investors reward companies that can deliver large audiences to advertisers."
https://kgi.georgetown.edu/research-and-commentary/better-feeds/
#SocialMedia #SocialNetworks #Algorithms #AlgorithmicRecommendation #RecommendationEngines
-
#SocialMedia #Algorithms #Instagram #RecommendationEngines #CoffeeShops: "Social media acumen requires awareness of each platform’s recommendation algorithm. Walsh observed that some companies may have great stories to tell, but they “are not attempting to keep up with these algorithmic patterns that will allow them to be visible to a larger audience”. Maybe they don’t post often enough, or they don’t keep up with shifts, such as Instagram promoting videos more than still images, a particularly stark change that occurred around 2022 as the platform attempted to mimic TikTok. Staying on top of what the algorithm demands is not easy, and even well-informed guesswork doesn’t always produce results. As Walsh told me: “We’ve put a lot of time and energy into creating beautiful content. But as a result of that algorithm, we find we’re not necessarily hitting as many eyeballs as we think we could or should, and sometimes that can be a little disheartening.”
“I hate the algorithm. Everyone hates the algorithm,” said Anca Ungureanu, the owner and founder of Beans & Dots, a coffee company in Bucharest, Romania, with its original location in a former printing plant. Her goal was to build “something that did not exist at that moment in Bucharest” – a space that was, at least aesthetically, non-local. It draws an international crowd; when someone searches Google for speciality coffee shops in Bucharest, Beans & Dots pops up. Ungureanu developed an Instagram account full of cappuccino snapshots and more than 7,000 followers, but grew frustrated when she felt that the platform was taking away her ability to access her audience through the feed. When her cafe started selling coffee online, Facebook and Instagram seemed to throttle its reach – unless it bought advertising and boosted the social media company’s own profits. It felt like algorithmic blackmail: pay our toll or we won’t promote you.
(...)
Other cafe owners I spoke to made the same complaint." -
#SocialMedia #Algorithms #Instagram #RecommendationEngines #CoffeeShops: "Social media acumen requires awareness of each platform’s recommendation algorithm. Walsh observed that some companies may have great stories to tell, but they “are not attempting to keep up with these algorithmic patterns that will allow them to be visible to a larger audience”. Maybe they don’t post often enough, or they don’t keep up with shifts, such as Instagram promoting videos more than still images, a particularly stark change that occurred around 2022 as the platform attempted to mimic TikTok. Staying on top of what the algorithm demands is not easy, and even well-informed guesswork doesn’t always produce results. As Walsh told me: “We’ve put a lot of time and energy into creating beautiful content. But as a result of that algorithm, we find we’re not necessarily hitting as many eyeballs as we think we could or should, and sometimes that can be a little disheartening.”
“I hate the algorithm. Everyone hates the algorithm,” said Anca Ungureanu, the owner and founder of Beans & Dots, a coffee company in Bucharest, Romania, with its original location in a former printing plant. Her goal was to build “something that did not exist at that moment in Bucharest” – a space that was, at least aesthetically, non-local. It draws an international crowd; when someone searches Google for speciality coffee shops in Bucharest, Beans & Dots pops up. Ungureanu developed an Instagram account full of cappuccino snapshots and more than 7,000 followers, but grew frustrated when she felt that the platform was taking away her ability to access her audience through the feed. When her cafe started selling coffee online, Facebook and Instagram seemed to throttle its reach – unless it bought advertising and boosted the social media company’s own profits. It felt like algorithmic blackmail: pay our toll or we won’t promote you.
(...)
Other cafe owners I spoke to made the same complaint." -
When I visit my son in Cleveland next month, we're going to Cedar Point for one day. To prepare, he sent me links to the official PoV videos for their roller coasters on YouTube. Now I fear the YouTube algorithm will be recommending roller coaster videos for me for the next 6 months. 😅 #YouTube #recommendationEngines
-
When I visit my son in Cleveland next month, we're going to Cedar Point for one day. To prepare, he sent me links to the official PoV videos for their roller coasters on YouTube. Now I fear the YouTube algorithm will be recommending roller coaster videos for me for the next 6 months. 😅 #YouTube #recommendationEngines
-
#YouTube #Algorithms #News #Media #Journalism #RecommendationEngines: "Huang and Yang used a data set of 1.7 million of YouTube’s “Up Next” recommended videos in 2019, using automated incognito browsing to eliminate any individual watch histories. They used network analysis, mathematical modeling, and Markov chains to determine the likelihood of news videos being recommended versus other topical categories.
They found that the topical filter bubble effect was stronger for most types of entertainment videos than for news (the stickiest topic in YouTube’s recommendations: cars), and that algorithmic redirection worked much more in entertainment videos’ favor, too. In other words, if you watch an entertainment video, you’re far more likely to be recommended the same genre of video than if you watch a news video.
The result is that as a user, you might start out watching news, but you’re likely to see more and more entertainment videos pop up as recommendations until you eventually watch one of them instead. On average, the researchers wrote, an entertainment video was three times more likely to be recommended than a news video, “indicating that no matter what users start with on YouTube, they are more likely to end up watching entertainment than news videos.”
Of course, recommendation algorithms don’t determine what people watch by themselves. Users can choose which of several recommended videos to click on, or what to type into the search bar, or whether to stop watching entirely. But Huang and Yang’s study isolates the influence of YouTube’s recommendation algorithm itself."
https://www.niemanlab.org/2024/06/how-youtubes-recommendations-pull-you-away-from-news/
-
#YouTube #Algorithms #News #Media #Journalism #RecommendationEngines: "Huang and Yang used a data set of 1.7 million of YouTube’s “Up Next” recommended videos in 2019, using automated incognito browsing to eliminate any individual watch histories. They used network analysis, mathematical modeling, and Markov chains to determine the likelihood of news videos being recommended versus other topical categories.
They found that the topical filter bubble effect was stronger for most types of entertainment videos than for news (the stickiest topic in YouTube’s recommendations: cars), and that algorithmic redirection worked much more in entertainment videos’ favor, too. In other words, if you watch an entertainment video, you’re far more likely to be recommended the same genre of video than if you watch a news video.
The result is that as a user, you might start out watching news, but you’re likely to see more and more entertainment videos pop up as recommendations until you eventually watch one of them instead. On average, the researchers wrote, an entertainment video was three times more likely to be recommended than a news video, “indicating that no matter what users start with on YouTube, they are more likely to end up watching entertainment than news videos.”
Of course, recommendation algorithms don’t determine what people watch by themselves. Users can choose which of several recommended videos to click on, or what to type into the search bar, or whether to stop watching entirely. But Huang and Yang’s study isolates the influence of YouTube’s recommendation algorithm itself."
https://www.niemanlab.org/2024/06/how-youtubes-recommendations-pull-you-away-from-news/
-
#SocialMedia #SocialNetworks #ContentModeration #Algorithms #RecommendationEngines #Messaging: "So you joined a social network without ranking algorithms—is everything good now? Jonathan Stray, a senior scientist at the UC Berkeley Center for Human-Compatible AI, has doubts. “There is now a bunch of research showing that chronological is not necessarily better,” he says, adding that simpler feeds can promote recency bias and enable spam.
Stray doesn’t think social harm is an inevitable outcome of complex algorithmic curation. But he agrees with Rogers that the tech industry’s practice of trying to maximize engagement doesn’t necessarily select for socially desirable results.
Stray suspects the solution to the problem of social media algorithms may in fact be … more algorithms. “The fundamental problem is you've got way too much information for anybody to consume, so you have to reduce it somehow,” he says."https://www.wired.com/story/latest-online-culture-war-is-humans-vs-algorithms/
-
#SocialMedia #SocialNetworks #ContentModeration #Algorithms #RecommendationEngines #Messaging: "So you joined a social network without ranking algorithms—is everything good now? Jonathan Stray, a senior scientist at the UC Berkeley Center for Human-Compatible AI, has doubts. “There is now a bunch of research showing that chronological is not necessarily better,” he says, adding that simpler feeds can promote recency bias and enable spam.
Stray doesn’t think social harm is an inevitable outcome of complex algorithmic curation. But he agrees with Rogers that the tech industry’s practice of trying to maximize engagement doesn’t necessarily select for socially desirable results.
Stray suspects the solution to the problem of social media algorithms may in fact be … more algorithms. “The fundamental problem is you've got way too much information for anybody to consume, so you have to reduce it somehow,” he says."https://www.wired.com/story/latest-online-culture-war-is-humans-vs-algorithms/
-
#SocialMedia #Facebook #Algorithms #RecommendationEngines #AI #Spam: "Facebook’s recommendation algorithms are promoting bizarre, AI-generated images being posted by spammers and scammers to an audience of people who mindlessly interact with them and perhaps don’t understand that they are not real, a new analysis by Stanford and Georgetown University researchers has found. The researchers’ analysis aligns with what I have seen and experienced over the course of months of researching and reporting on these pages, many of which have found a novel way to link to off-platform, AI-generated “news” sites that are littered with Google ads or which are selling low-quality products.
Last week the world was introduced to Shrimp Jesus, a series of AI-generated images in which Jesus is melded with a crustacean, and which have repeatedly gone viral on Facebook. The images are emblematic of a specific type of AI image being used by spammers and scammers, which I first wrote about in December but have repeatedly made the masses go “WTF” and “WHY?” when shared away from an audience of Facebook users who are seemingly unable to detect them as AI, or don’t care that they are AI. “WHAT IS HAPPENING ON FACEBOOK,” a viral tweet about Shrimp Jesus read." https://www.404media.co/facebooks-algorithm-is-boosting-ai-spam-that-links-to-ai-generated-ad-laden-click-farms/
-
#SocialMedia #Facebook #Algorithms #RecommendationEngines #AI #Spam: "Facebook’s recommendation algorithms are promoting bizarre, AI-generated images being posted by spammers and scammers to an audience of people who mindlessly interact with them and perhaps don’t understand that they are not real, a new analysis by Stanford and Georgetown University researchers has found. The researchers’ analysis aligns with what I have seen and experienced over the course of months of researching and reporting on these pages, many of which have found a novel way to link to off-platform, AI-generated “news” sites that are littered with Google ads or which are selling low-quality products.
Last week the world was introduced to Shrimp Jesus, a series of AI-generated images in which Jesus is melded with a crustacean, and which have repeatedly gone viral on Facebook. The images are emblematic of a specific type of AI image being used by spammers and scammers, which I first wrote about in December but have repeatedly made the masses go “WTF” and “WHY?” when shared away from an audience of Facebook users who are seemingly unable to detect them as AI, or don’t care that they are AI. “WHAT IS HAPPENING ON FACEBOOK,” a viral tweet about Shrimp Jesus read." https://www.404media.co/facebooks-algorithm-is-boosting-ai-spam-that-links-to-ai-generated-ad-laden-click-farms/
-
#Ireland #AI #GenerativeAI #Algorithms #RecommendationEngines: "Ireland cannot put its faith in "voluntary action by tech corporations" to safeguard young people online from the rise of generative artificial intelligence (AI) technologies, the Oireachtas children’s committee will hear today.
The Irish Council for Civil Liberties (ICCL) is set to speak to the committee on Tuesday regarding the safe use of AI by young people.
”Technology corporations have a very poor record of self-improvement and responsible behaviour, even when they know their technology is harmful, and even when lives are at stake,” Dr Johnny Ryan, director of the ICCL’s Enforce unit is due to tell the committee.
“Tech corporations will not save our children,” Dr Ryan is expected to say. "
-
#Ireland #AI #GenerativeAI #Algorithms #RecommendationEngines: "Ireland cannot put its faith in "voluntary action by tech corporations" to safeguard young people online from the rise of generative artificial intelligence (AI) technologies, the Oireachtas children’s committee will hear today.
The Irish Council for Civil Liberties (ICCL) is set to speak to the committee on Tuesday regarding the safe use of AI by young people.
”Technology corporations have a very poor record of self-improvement and responsible behaviour, even when they know their technology is harmful, and even when lives are at stake,” Dr Johnny Ryan, director of the ICCL’s Enforce unit is due to tell the committee.
“Tech corporations will not save our children,” Dr Ryan is expected to say. "
-
#Journalism #Media #News #BBC #RecommendationEngines #PublicBroadcasting: "The BBC is the world’s largest public service broadcaster. Every week it reaches more than 90% of the UK’s adult population and 489 million people worldwide. To ensure our audiences get the most engaging experience, our team develops recommender systems which aim to provide users with the most relevant pieces of content among the thousands the BBC publishes every day. All BBC output should serve the organization’s mission to “act in the public interest, serving all audiences through the provision of impartial, high-quality and distinctive output and services which inform, educate, and entertain.” Recommendations make no exception and, since they determine what our audiences see, they are in effect editorial choices at scale. How can we ensure that our recommendations are consistent with our mission and public service values, avoiding some of the harmful effects that might be associated with recommenders? In addressing this question, we identified two main challenges: 1) methodological challenges: public service values are hard to measure through specific metrics, therefore we have no clearly defined optimization function for our recommenders; 2) cultural/operational challenges: domain knowledge around public service values sits with our editorial staff, whereas data scientists are the recommendations specialists. We need to create a shared understanding of the problem and a common language to describe objectives and solutions across data science and editorial. Our paper describes the approach we devised to tackle these challenges, presenting a use case from our work on a BBC product, and reporting the lessons learned."
-
#Journalism #Media #News #BBC #RecommendationEngines #PublicBroadcasting: "The BBC is the world’s largest public service broadcaster. Every week it reaches more than 90% of the UK’s adult population and 489 million people worldwide. To ensure our audiences get the most engaging experience, our team develops recommender systems which aim to provide users with the most relevant pieces of content among the thousands the BBC publishes every day. All BBC output should serve the organization’s mission to “act in the public interest, serving all audiences through the provision of impartial, high-quality and distinctive output and services which inform, educate, and entertain.” Recommendations make no exception and, since they determine what our audiences see, they are in effect editorial choices at scale. How can we ensure that our recommendations are consistent with our mission and public service values, avoiding some of the harmful effects that might be associated with recommenders? In addressing this question, we identified two main challenges: 1) methodological challenges: public service values are hard to measure through specific metrics, therefore we have no clearly defined optimization function for our recommenders; 2) cultural/operational challenges: domain knowledge around public service values sits with our editorial staff, whereas data scientists are the recommendations specialists. We need to create a shared understanding of the problem and a common language to describe objectives and solutions across data science and editorial. Our paper describes the approach we devised to tackle these challenges, presenting a use case from our work on a BBC product, and reporting the lessons learned."
-
Platformed! How #Streaming, #Algorithms and Artificial Intelligence are Shaping #Music Cultures
Tiziano Bonini and Paolo Magaudda
(Palgrave Macmillan, 2024)
"Offers an up-to-date overview how the music industry has transformed in response to digitalization and online platforms
Shows how digital music platforms and the cultural value of music in today’s society are mutually constructed
Presents an analysis of emerging music technologies, like artificial intelligence and blockchain for music circulation."
-
Platformed! How #Streaming, #Algorithms and Artificial Intelligence are Shaping #Music Cultures
Tiziano Bonini and Paolo Magaudda
(Palgrave Macmillan, 2024)
"Offers an up-to-date overview how the music industry has transformed in response to digitalization and online platforms
Shows how digital music platforms and the cultural value of music in today’s society are mutually constructed
Presents an analysis of emerging music technologies, like artificial intelligence and blockchain for music circulation."
-
#EU #BigTech #RecommendationEngines #Personalization #Algorithms: "Another policy tug-of-war could be emerging around Big Tech’s content recommender systems in the European Union where the Commission is facing a call from a number of parliamentarians to rein in profiling-based content feeds — aka “personalization” engines that process user data in order to determine what content to show them.
(...)
The letter, signed by 17 MEPs from political groups including S&D, the left, greens, EPP and Renew Europe, advocates for tech platforms’ recommender systems to be switched off by default — an idea that was floated during negotiations over the bloc’s Digital Services Act (DSA) but which did not make it into the final regulation as it did not have a democratic majority. Instead EU lawmakers agreed to transparency measures for recommender systems, along with a requirement that larger platforms (so called VLOPs) must provide at least one content feed that isn’t based on profiling." -
#EU #BigTech #RecommendationEngines #Personalization #Algorithms: "Another policy tug-of-war could be emerging around Big Tech’s content recommender systems in the European Union where the Commission is facing a call from a number of parliamentarians to rein in profiling-based content feeds — aka “personalization” engines that process user data in order to determine what content to show them.
(...)
The letter, signed by 17 MEPs from political groups including S&D, the left, greens, EPP and Renew Europe, advocates for tech platforms’ recommender systems to be switched off by default — an idea that was floated during negotiations over the bloc’s Digital Services Act (DSA) but which did not make it into the final regulation as it did not have a democratic majority. Instead EU lawmakers agreed to transparency measures for recommender systems, along with a requirement that larger platforms (so called VLOPs) must provide at least one content feed that isn’t based on profiling." -
#RecommendationEngines #Algorithms #Ethics #HumanValues: "Recommender systems are the algorithms which select, filter, and personalize content across many of the world's largest platforms and apps. As such, their positive and negative effects on individuals and on societies have been extensively theorized and studied. Our overarching question is how to ensure that recommender systems enact the values of the individuals and societies that they serve. Addressing this question in a principled fashion requires technical knowledge of recommender design and operation, and also critically depends on insights from diverse fields including social science, ethics, economics, psychology, policy and law. This paper is a multidisciplinary effort to synthesize theory and practice from different perspectives, with the goal of providing a shared language, articulating current design approaches, and identifying open problems. We collect a set of values that seem most relevant to recommender systems operating across different domains, then examine them from the perspectives of current industry practice, measurement, product design, and policy approaches. Important open problems include multi-stakeholder processes for defining values and resolving trade-offs, better values-driven measurements, recommender controls that people use, non-behavioral algorithmic feedback, optimization for long-term outcomes, causal inference of recommender effects, academic-industry research collaborations, and interdisciplinary policy-making." https://dl.acm.org/doi/10.1145/3632297
-
#RecommendationEngines #Algorithms #Ethics #HumanValues: "Recommender systems are the algorithms which select, filter, and personalize content across many of the world's largest platforms and apps. As such, their positive and negative effects on individuals and on societies have been extensively theorized and studied. Our overarching question is how to ensure that recommender systems enact the values of the individuals and societies that they serve. Addressing this question in a principled fashion requires technical knowledge of recommender design and operation, and also critically depends on insights from diverse fields including social science, ethics, economics, psychology, policy and law. This paper is a multidisciplinary effort to synthesize theory and practice from different perspectives, with the goal of providing a shared language, articulating current design approaches, and identifying open problems. We collect a set of values that seem most relevant to recommender systems operating across different domains, then examine them from the perspectives of current industry practice, measurement, product design, and policy approaches. Important open problems include multi-stakeholder processes for defining values and resolving trade-offs, better values-driven measurements, recommender controls that people use, non-behavioral algorithmic feedback, optimization for long-term outcomes, causal inference of recommender effects, academic-industry research collaborations, and interdisciplinary policy-making." https://dl.acm.org/doi/10.1145/3632297
-
#EU #SocialMedia #TikTok #DSA #Algorithms #RecommendationEngines: "TikTok users in Europe will be able to switch off the personalized algorithm behind its For You and Live feeds as the company makes changes to comply with the EU’s Digital Services Act (DSA). According to TikTok, disabling this function will show users “popular videos from both the places where they live and around the world” instead of content based on their personal interests.
These changes relate to DSA rules that require very large online platforms to allow their users to opt out of receiving personalized content — which typically relies on tracking and profiling user activity — when viewing content recommendations. To comply, TikTok’s search feature will also show content that’s popular in the user’s region, and videos under the “Following” and “Friends” feeds will be displayed in chronological order when a non-personalized view is selected."
https://www.theverge.com/2023/8/4/23819878/tiktok-fyp-algorithm-eu-dsa-personalization-data-tracking
-
#EU #SocialMedia #TikTok #DSA #Algorithms #RecommendationEngines: "TikTok users in Europe will be able to switch off the personalized algorithm behind its For You and Live feeds as the company makes changes to comply with the EU’s Digital Services Act (DSA). According to TikTok, disabling this function will show users “popular videos from both the places where they live and around the world” instead of content based on their personal interests.
These changes relate to DSA rules that require very large online platforms to allow their users to opt out of receiving personalized content — which typically relies on tracking and profiling user activity — when viewing content recommendations. To comply, TikTok’s search feature will also show content that’s popular in the user’s region, and videos under the “Following” and “Friends” feeds will be displayed in chronological order when a non-personalized view is selected."
https://www.theverge.com/2023/8/4/23819878/tiktok-fyp-algorithm-eu-dsa-personalization-data-tracking
-
#SocialMedia #Algorithms #ContentModeration #RecommendationEngines: "Six observations on ranking by engagement:
1. Internet platforms rank content primarily by the predicted probability of engagement.
2. Platforms rank by engagement because it increases user retention.
3. Engagement is negatively related to quality.
4. Sensitive content is often both engaging and retentive.
5. Sensitive content is often preferred by users.
6. Platforms don’t want sensitive content but don’t want to be seen to be removing it."https://tecunningham.github.io/posts/2023-04-28-ranking-by-engagement.html
-
#SocialMedia #Algorithms #ContentModeration #RecommendationEngines: "Six observations on ranking by engagement:
1. Internet platforms rank content primarily by the predicted probability of engagement.
2. Platforms rank by engagement because it increases user retention.
3. Engagement is negatively related to quality.
4. Sensitive content is often both engaging and retentive.
5. Sensitive content is often preferred by users.
6. Platforms don’t want sensitive content but don’t want to be seen to be removing it."https://tecunningham.github.io/posts/2023-04-28-ranking-by-engagement.html
-
#SocialMedia #Twitter #Algorithms #ML #RecommendationEngines: "As social media continues to have a significant influence on public opinion, understanding the impact of the machine learning algorithms that filter and curate content is crucial. However, existing studies have yielded inconsistent results, potentially due to limitations such as reliance on observational methods, use of simulated rather than real users, restriction to specific types of content, or internal access requirements that may create conflicts of interest. To overcome these issues, we conducted a pre-registered controlled experiment on Twitter's algorithm without internal access. The key to our design was to, for a large group of active Twitter users, simultaneously collect (a) the tweets the personalized algorithm shows, and (b) the tweets the user would have seen if they were just shown the latest tweets from people they follow; we then surveyed users about both sets of tweets in a random order.
Our results indicate that the algorithm amplifies emotional content, and especially those tweets that express anger and out-group animosity. Furthermore, political tweets from the algorithm lead readers to perceive their political in-group more positively and their political out-group more negatively. Interestingly, while readers generally say they prefer tweets curated by the algorithm, they are less likely to prefer algorithm-selected political tweets. Overall, our study provides important insights into the impact of social media ranking algorithms, with implications for shaping public discourse and democratic engagement."
https://arxiv.org/abs/2305.16941 -
#SocialMedia #Twitter #Algorithms #ML #RecommendationEngines: "As social media continues to have a significant influence on public opinion, understanding the impact of the machine learning algorithms that filter and curate content is crucial. However, existing studies have yielded inconsistent results, potentially due to limitations such as reliance on observational methods, use of simulated rather than real users, restriction to specific types of content, or internal access requirements that may create conflicts of interest. To overcome these issues, we conducted a pre-registered controlled experiment on Twitter's algorithm without internal access. The key to our design was to, for a large group of active Twitter users, simultaneously collect (a) the tweets the personalized algorithm shows, and (b) the tweets the user would have seen if they were just shown the latest tweets from people they follow; we then surveyed users about both sets of tweets in a random order.
Our results indicate that the algorithm amplifies emotional content, and especially those tweets that express anger and out-group animosity. Furthermore, political tweets from the algorithm lead readers to perceive their political in-group more positively and their political out-group more negatively. Interestingly, while readers generally say they prefer tweets curated by the algorithm, they are less likely to prefer algorithm-selected political tweets. Overall, our study provides important insights into the impact of social media ranking algorithms, with implications for shaping public discourse and democratic engagement."
https://arxiv.org/abs/2305.16941 -
#SocialMedia #USA #Section230 #CDA #ContentModeration #RecommendationEngines: "When you do a search online, you ask, “Of the millions of websites out there, tell me the one, oh Google, or oh Bing, which you recommend to me as most relevant to my query.” So recommendation algorithms are what… Some people prefer a chronological feed. So I think that’s great, I think we should have those choices, but I actually do like the Facebook news feed and the Twitter feed, even the For You feed. The TikTok feed is of course entirely recommendation algorithms. It’s not really based largely on whom you follow, and certainly not on the chronological postings of whom you follow.
So recommendation algorithms are everywhere. We learned in this case that they’re used by Reddit and they’re used by… Wikipedia is using various algorithms in this process. Maybe not a recommendation, but doing lots of other… The dirty work behind the scenes is being done by automated algorithms. And if the way that the automated algorithm functions, sometimes accidentally promoting something wrong, maybe it’s some kind of terrible weapon or some self-harm that it might promote because it’s an automated algorithm, then if that leads to liability… boy, does that change the way that the internet works. And that’s why you saw an outpouring of briefs, as I said, from Reddit, which people think of as very much human-curated. Or Wikipedia, which again people think of as human-moderated and human-produced, to, of course, all the big tech platforms, including Microsoft, which wasn’t being sued, but still has services like GitHub and LinkedIn, which it said were at risk if this case proceeded in the way that the plaintiffs would’ve liked. So in this case, I think there was a lot at stake."
-
#SocialMedia #USA #Section230 #CDA #ContentModeration #RecommendationEngines: "When you do a search online, you ask, “Of the millions of websites out there, tell me the one, oh Google, or oh Bing, which you recommend to me as most relevant to my query.” So recommendation algorithms are what… Some people prefer a chronological feed. So I think that’s great, I think we should have those choices, but I actually do like the Facebook news feed and the Twitter feed, even the For You feed. The TikTok feed is of course entirely recommendation algorithms. It’s not really based largely on whom you follow, and certainly not on the chronological postings of whom you follow.
So recommendation algorithms are everywhere. We learned in this case that they’re used by Reddit and they’re used by… Wikipedia is using various algorithms in this process. Maybe not a recommendation, but doing lots of other… The dirty work behind the scenes is being done by automated algorithms. And if the way that the automated algorithm functions, sometimes accidentally promoting something wrong, maybe it’s some kind of terrible weapon or some self-harm that it might promote because it’s an automated algorithm, then if that leads to liability… boy, does that change the way that the internet works. And that’s why you saw an outpouring of briefs, as I said, from Reddit, which people think of as very much human-curated. Or Wikipedia, which again people think of as human-moderated and human-produced, to, of course, all the big tech platforms, including Microsoft, which wasn’t being sued, but still has services like GitHub and LinkedIn, which it said were at risk if this case proceeded in the way that the plaintiffs would’ve liked. So in this case, I think there was a lot at stake."
-
E X P L A I N A B I L I T Y, yEaH!!
#Explainability #SocialMedia #SocialNetworks #Algorithms #RecommendationEngines #SocialSciences: "Governance of algorithms will require reliable explanations for how things could go wrong. But neither technologists nor social scientists have the language or models to create these explanations.
To a technologist, an explanation is an account of why an algorithm took a specific action, compared with what it might have done otherwise. In the case of YouTube, computer scientists might be able to explain what part of a mathematical matrix contributed most to the algorithm’s recommendation of a terrorist video at one moment. But they aren’t easily able to suggest changes that would have prevented terrorist recruitment, because that would require a better understanding of human behaviour. That’s why computer scientists cannot say in advance whether a given design will become more harmful or less harmful over time.
Humanities researchers and social scientists, meanwhile, are working to develop algorithm-informed explanations of collective behaviour. A 2021 study of Twitter users showed that posts expressing moral outrage received more likes than did posts showing other emotions, and that this feedback caused people to express more outrage in the future. But studies such as this can’t currently explain how algorithms respond or what they contribute to these dynamics"
-
E X P L A I N A B I L I T Y, yEaH!!
#Explainability #SocialMedia #SocialNetworks #Algorithms #RecommendationEngines #SocialSciences: "Governance of algorithms will require reliable explanations for how things could go wrong. But neither technologists nor social scientists have the language or models to create these explanations.
To a technologist, an explanation is an account of why an algorithm took a specific action, compared with what it might have done otherwise. In the case of YouTube, computer scientists might be able to explain what part of a mathematical matrix contributed most to the algorithm’s recommendation of a terrorist video at one moment. But they aren’t easily able to suggest changes that would have prevented terrorist recruitment, because that would require a better understanding of human behaviour. That’s why computer scientists cannot say in advance whether a given design will become more harmful or less harmful over time.
Humanities researchers and social scientists, meanwhile, are working to develop algorithm-informed explanations of collective behaviour. A 2021 study of Twitter users showed that posts expressing moral outrage received more likes than did posts showing other emotions, and that this feedback caused people to express more outrage in the future. But studies such as this can’t currently explain how algorithms respond or what they contribute to these dynamics"
-
Today #Twitter released much of the code used for their recommendation algorithm https://blog.twitter.com/engineering/en_us/topics/open-source/2023/twitter-recommendation-algorithm
An machine-learning system relies on both an algorithm and training data so I wonder exactly what insights can be gained from what's been made public (I'm definitely not an expert in this area so I invite corrections and clarifications here). Regardless, it's an unusual level of transparency for a major social-media platform.
#SocialMedia #RecommendationEngines #RecommendationAlgorithms
-
#SocialMedia #Algorithms #RecommendationEngines: "I think a broader understanding of recommendation algorithms is sorely needed. Policymakers and legal scholars must understand these algorithms so that they can sharpen their thinking on platform governance; journalists must understand them so that they can explain them to readers and better hold platforms accountable; technologists must understand them so that the platforms of tomorrow may be better than the ones we have; researchers must understand them so that they can get at the intricate interplay between algorithms and human behavior. Content creators would also benefit from understanding them so that they can better navigate the new landscape of algorithmic distribution. More generally, anyone concerned about the impact of algorithmic platforms on themselves or on society may find this essay of interest.
I hope to show you that social media algorithms are simple to understand. In addition to the mathematical principles of information cascades (which are independent of any platform), it’s also straightforward to understand what recommendation algorithms are trained to do, and what inputs they use."
https://knightcolumbia.org/content/understanding-social-media-recommendation-algorithms
-
#SocialMedia #Algorithms #RecommendationEngines: "I think a broader understanding of recommendation algorithms is sorely needed. Policymakers and legal scholars must understand these algorithms so that they can sharpen their thinking on platform governance; journalists must understand them so that they can explain them to readers and better hold platforms accountable; technologists must understand them so that the platforms of tomorrow may be better than the ones we have; researchers must understand them so that they can get at the intricate interplay between algorithms and human behavior. Content creators would also benefit from understanding them so that they can better navigate the new landscape of algorithmic distribution. More generally, anyone concerned about the impact of algorithmic platforms on themselves or on society may find this essay of interest.
I hope to show you that social media algorithms are simple to understand. In addition to the mathematical principles of information cascades (which are independent of any platform), it’s also straightforward to understand what recommendation algorithms are trained to do, and what inputs they use."
https://knightcolumbia.org/content/understanding-social-media-recommendation-algorithms
-
#Music #Streaming #Algorithms #RecommendationEngines: "In commissioning this literature review as the first stage in that research, the CDEI asked us to include in our considerations how and to what degree existing research has addressed a number of issues relevant to the concerns above, with a focus on “algorithmically-driven music recommendation systems”, in particular:
the question of “bias” in music streaming algorithms: how might different groups of artists and consumers be affected by algorithms?
the question of diversity: positive and negative impacts of algorithms on musical diversity
questions of transparency, opacity and oversight
These are therefore the main issues we seek to address here. A distinctive feature of the review is that we seek to put academic computer science research and “critical” social science and humanities research (notably a sub-field known as critical algorithm studies) into dialogue with each other, to a much greater extent than has been evident in existing scholarship."
-
#Music #Streaming #Algorithms #RecommendationEngines: "In commissioning this literature review as the first stage in that research, the CDEI asked us to include in our considerations how and to what degree existing research has addressed a number of issues relevant to the concerns above, with a focus on “algorithmically-driven music recommendation systems”, in particular:
the question of “bias” in music streaming algorithms: how might different groups of artists and consumers be affected by algorithms?
the question of diversity: positive and negative impacts of algorithms on musical diversity
questions of transparency, opacity and oversight
These are therefore the main issues we seek to address here. A distinctive feature of the review is that we seek to put academic computer science research and “critical” social science and humanities research (notably a sub-field known as critical algorithm studies) into dialogue with each other, to a much greater extent than has been evident in existing scholarship."
-
@siderea I think what can be done locally would be a good start to reducing things like "don't show me the same boost more than once".
I'm certain that the extent of what can be done locally with reasonable cache sizes is a meaningful change to what we would see for anyone whose followings have more than a couple hundred posts + boosts in a day.
And the complexity escalates exponentially if we need to carefully manage local cache sizes to support even simple models. Probably the most that's feasible (while still complex) would be thresholds or weighting for "number of posts + boosts to see from each follow" combined with "rank by reaction velocity since posting" — which allows a client to recognize that it needs to request more from the server to fill that queue. But again, you can see how for more than a couple hundred follows, this rapidly hits gigabytes of local caching and requires list preparation ON the server.
For comparison: few would consider Netflix recommendations to be toxic — and yet the list of rows and each horizontal row of movies must be prepared and cached on Netflix servers ready for a client to request it. That's on the order of 10k titles, each with a short list of tags — much less data than Mastodon handles.
Does that give you an idea of the technical challenges?
#MastodonRanking
#AlgorithmDrivenTimelines
#ChronologicalTimelines
#MastodonDesign
#RecommendationEngines -
@siderea I think what can be done locally would be a good start to reducing things like "don't show me the same boost more than once".
I'm certain that the extent of what can be done locally with reasonable cache sizes is a meaningful change to what we would see for anyone whose followings have more than a couple hundred posts + boosts in a day.
And the complexity escalates exponentially if we need to carefully manage local cache sizes to support even simple models. Probably the most that's feasible (while still complex) would be thresholds or weighting for "number of posts + boosts to see from each follow" combined with "rank by reaction velocity since posting" — which allows a client to recognize that it needs to request more from the server to fill that queue. But again, you can see how for more than a couple hundred follows, this rapidly hits gigabytes of local caching and requires list preparation ON the server.
For comparison: few would consider Netflix recommendations to be toxic — and yet the list of rows and each horizontal row of movies must be prepared and cached on Netflix servers ready for a client to request it. That's on the order of 10k titles, each with a short list of tags — much less data than Mastodon handles.
Does that give you an idea of the technical challenges?
#MastodonRanking
#AlgorithmDrivenTimelines
#ChronologicalTimelines
#MastodonDesign
#RecommendationEngines