Manoel Horta Ribeiro
@manoelhortaribeiro.bsky.social
Assistant Professor @ Princeton
Previously: EPFL 🇨🇭, UFMG 🇧🇷
Interests: Computational Social Science, Platforms, GenAI, Moderation
Previously: EPFL 🇨🇭, UFMG 🇧🇷
Interests: Computational Social Science, Platforms, GenAI, Moderation
But is this about short-form video or about TikTok?
We repeat the analysis in a later period when Instagram and Facebook also had short-form video. There, differences between TikTok adopters and others vanish—suggesting the effect is about the format, not the specific app.
We repeat the analysis in a later period when Instagram and Facebook also had short-form video. There, differences between TikTok adopters and others vanish—suggesting the effect is about the format, not the specific app.
November 17, 2025 at 3:42 PM
But is this about short-form video or about TikTok?
We repeat the analysis in a later period when Instagram and Facebook also had short-form video. There, differences between TikTok adopters and others vanish—suggesting the effect is about the format, not the specific app.
We repeat the analysis in a later period when Instagram and Facebook also had short-form video. There, differences between TikTok adopters and others vanish—suggesting the effect is about the format, not the specific app.
When does this extra usage happen?
Effects are concentrated during the day. We find no consistent evidence that short-form video increases nighttime mobile use beyond what other social media already does.
Effects are concentrated during the day. We find no consistent evidence that short-form video increases nighttime mobile use beyond what other social media already does.
November 17, 2025 at 3:42 PM
When does this extra usage happen?
Effects are concentrated during the day. We find no consistent evidence that short-form video increases nighttime mobile use beyond what other social media already does.
Effects are concentrated during the day. We find no consistent evidence that short-form video increases nighttime mobile use beyond what other social media already does.
Who is most affected?
Short-form video especially pulls in people who previously used their phones less:
Low-intensity users: ≈31% increase in total mobile duration
High-intensity users: ≈14% increase
Short-form video especially pulls in people who previously used their phones less:
Low-intensity users: ≈31% increase in total mobile duration
High-intensity users: ≈14% increase
November 17, 2025 at 3:42 PM
Who is most affected?
Short-form video especially pulls in people who previously used their phones less:
Low-intensity users: ≈31% increase in total mobile duration
High-intensity users: ≈14% increase
Short-form video especially pulls in people who previously used their phones less:
Low-intensity users: ≈31% increase in total mobile duration
High-intensity users: ≈14% increase
Main finding: short-form video platforms meaningfully amplify mobile use.
After adoption, total mobile duration increases by ≈17% (about +28 minutes/day for the average user), and the average time away from the phone (TAP) shrinks by ≈20%.
After adoption, total mobile duration increases by ≈17% (about +28 minutes/day for the average user), and the average time away from the phone (TAP) shrinks by ≈20%.
November 17, 2025 at 3:42 PM
Main finding: short-form video platforms meaningfully amplify mobile use.
After adoption, total mobile duration increases by ≈17% (about +28 minutes/day for the average user), and the average time away from the phone (TAP) shrinks by ≈20%.
After adoption, total mobile duration increases by ≈17% (about +28 minutes/day for the average user), and the average time away from the phone (TAP) shrinks by ≈20%.
TikTok and the likes transformed how we spend time on our phones. But do they increase mobile use, or just reshuffle what we already do?
We find that adopting TikTok increases total mobile usage and shortens breaks from the phone.
Preprint: osf.io/preprints/so...
We find that adopting TikTok increases total mobile usage and shortens breaks from the phone.
Preprint: osf.io/preprints/so...
November 17, 2025 at 3:42 PM
TikTok and the likes transformed how we spend time on our phones. But do they increase mobile use, or just reshuffle what we already do?
We find that adopting TikTok increases total mobile usage and shortens breaks from the phone.
Preprint: osf.io/preprints/so...
We find that adopting TikTok increases total mobile usage and shortens breaks from the phone.
Preprint: osf.io/preprints/so...
Computer Science is no longer just about building systems or proving theorems--it's about observation and experiments.
In my latest blog post, I argue it’s time we had our own "Econometrics," a discipline devoted to empirical rigor.
doomscrollingbabel.manoel.xyz/p/the-missin...
In my latest blog post, I argue it’s time we had our own "Econometrics," a discipline devoted to empirical rigor.
doomscrollingbabel.manoel.xyz/p/the-missin...
October 5, 2025 at 4:07 PM
Computer Science is no longer just about building systems or proving theorems--it's about observation and experiments.
In my latest blog post, I argue it’s time we had our own "Econometrics," a discipline devoted to empirical rigor.
doomscrollingbabel.manoel.xyz/p/the-missin...
In my latest blog post, I argue it’s time we had our own "Econometrics," a discipline devoted to empirical rigor.
doomscrollingbabel.manoel.xyz/p/the-missin...
Bonsai is modular and platform-agnostic, opening paths for integration beyond Bluesky. The paper details the backend design, study, and implications: arxiv.org/abs/2509.10776
September 16, 2025 at 1:24 PM
Bonsai is modular and platform-agnostic, opening paths for integration beyond Bluesky. The paper details the backend design, study, and implications: arxiv.org/abs/2509.10776
We implemented Bonsai on Bluesky and conducted a two-phase, multi-week study with 15 participants. This deployment allowed us to observe how people used intentional feedbuilding in practice, and how it compared to their experiences with engagement-driven defaults.
September 16, 2025 at 1:24 PM
We implemented Bonsai on Bluesky and conducted a two-phase, multi-week study with 15 participants. This deployment allowed us to observe how people used intentional feedbuilding in practice, and how it compared to their experiences with engagement-driven defaults.
In the ranking stage, Bonsai orders the curated content using criteria derived from the user’s stated intent—rather than predicted engagement—making the logic behind feed prioritization transparent and directly aligned with user goals.
September 16, 2025 at 1:24 PM
In the ranking stage, Bonsai orders the curated content using criteria derived from the user’s stated intent—rather than predicted engagement—making the logic behind feed prioritization transparent and directly aligned with user goals.
In the curating stage, users can apply natural language prompts (e.g., “focus on recent policy updates” or “exclude promotional posts”) to filter and organize the sourced content, ensuring the feed reflects users' goals / preferences. Each prompt is fed into an LLM that individually ranks content.
September 16, 2025 at 1:24 PM
In the curating stage, users can apply natural language prompts (e.g., “focus on recent policy updates” or “exclude promotional posts”) to filter and organize the sourced content, ensuring the feed reflects users' goals / preferences. Each prompt is fed into an LLM that individually ranks content.
In the sourcing stage, Bonsai gathers a wide pool of candidate posts aligned with user goals. Users can refine this stage by editing sources (adding or removing accounts, hashtags, or feeds) to shape where their feed draws content from.
September 16, 2025 at 1:24 PM
In the sourcing stage, Bonsai gathers a wide pool of candidate posts aligned with user goals. Users can refine this stage by editing sources (adding or removing accounts, hashtags, or feeds) to shape where their feed draws content from.
In the planning stage, users express their goals in natural language (e.g., “updates on AI policy” or “posts from close colleagues”). Bonsai translates these goals into structured representations that guide the subsequent sourcing, curating, and ranking of content by providing initial suggestions.
September 16, 2025 at 1:24 PM
In the planning stage, users express their goals in natural language (e.g., “updates on AI policy” or “posts from close colleagues”). Bonsai translates these goals into structured representations that guide the subsequent sourcing, curating, and ranking of content by providing initial suggestions.
With Bonsai, users can articulate what they want from their feeds (e.g., tracking research, staying informed on a policy area, or connecting with a community) and the system procedurally builds a feed that reflects those intentions in four steps, which we discuss below.
September 16, 2025 at 1:24 PM
With Bonsai, users can articulate what they want from their feeds (e.g., tracking research, staying informed on a policy area, or connecting with a community) and the system procedurally builds a feed that reflects those intentions in four steps, which we discuss below.
Bonsai sits within a broader debate on recommender systems. While TikTok or Meta optimize (mostly) for attention capture, Bonsai explores what feeds look like when personalized for user intent. Under our taxonomy, it explores the design space of "intentional" and "personalized" feeds!
September 16, 2025 at 1:24 PM
Bonsai sits within a broader debate on recommender systems. While TikTok or Meta optimize (mostly) for attention capture, Bonsai explores what feeds look like when personalized for user intent. Under our taxonomy, it explores the design space of "intentional" and "personalized" feeds!
Social media feeds today are optimized for engagement, often leading to misalignment between users' intentions and technology use.
In a new paper, we introduce Bonsai, a tool to create feeds based on stated preferences, rather than predicted engagement.
arxiv.org/abs/2509.10776
In a new paper, we introduce Bonsai, a tool to create feeds based on stated preferences, rather than predicted engagement.
arxiv.org/abs/2509.10776
September 16, 2025 at 1:24 PM
Social media feeds today are optimized for engagement, often leading to misalignment between users' intentions and technology use.
In a new paper, we introduce Bonsai, a tool to create feeds based on stated preferences, rather than predicted engagement.
arxiv.org/abs/2509.10776
In a new paper, we introduce Bonsai, a tool to create feeds based on stated preferences, rather than predicted engagement.
arxiv.org/abs/2509.10776
This echoes today’s debate on large language models. Bender & Koller’s famous “octopus test” argues LMs only manipulate form, never meaning. Like Searle’s man, they can be fluent without grounding in the world.
August 25, 2025 at 4:30 PM
This echoes today’s debate on large language models. Bender & Koller’s famous “octopus test” argues LMs only manipulate form, never meaning. Like Searle’s man, they can be fluent without grounding in the world.
Among many criticisms of this, one that jumps out is inefficiency. The room assumes infinite time + an endless rulebook. Dennett argued: intelligence requires speed. Wilkenfeld (2013) adds: understanding is compression.
August 25, 2025 at 4:30 PM
Among many criticisms of this, one that jumps out is inefficiency. The room assumes infinite time + an endless rulebook. Dennett argued: intelligence requires speed. Wilkenfeld (2013) adds: understanding is compression.
In the original Chinese Room Argument, A man is locked in a room and slips of paper with Chinese characters are passed in. He has a giant English rulebook that tells him how to respond. He follows it mechanically and passes fluent Chinese back out—without understanding a word.
August 25, 2025 at 4:30 PM
In the original Chinese Room Argument, A man is locked in a room and slips of paper with Chinese characters are passed in. He has a giant English rulebook that tells him how to respond. He follows it mechanically and passes fluent Chinese back out—without understanding a word.
I'm thinking a lot about the debate around whether LMs can "understand." Enter a new thought experiment I call "the lean Chinese room." IMO, it shows how the pressure of compression and speed blurs the line between symbol-shuffling and understanding.
doomscrollingbabel.manoel.xyz/p/the-lean-c...
doomscrollingbabel.manoel.xyz/p/the-lean-c...
August 25, 2025 at 4:30 PM
I'm thinking a lot about the debate around whether LMs can "understand." Enter a new thought experiment I call "the lean Chinese room." IMO, it shows how the pressure of compression and speed blurs the line between symbol-shuffling and understanding.
doomscrollingbabel.manoel.xyz/p/the-lean-c...
doomscrollingbabel.manoel.xyz/p/the-lean-c...
Yeah, I guess I was focused on a very short time window where everything was always called AI, but the pattern holds for something like DeepBlue. On the flip side...
August 19, 2025 at 7:55 PM
Yeah, I guess I was focused on a very short time window where everything was always called AI, but the pattern holds for something like DeepBlue. On the flip side...
Whether or not AIs “truly” understand is less important than how they reshape jobs, politics, and culture.
My new post: why the “Human of the Gaps” fallacy distracts us from governing real-world AI impacts.
doomscrollingbabel.manoel.xyz/p/the-retrea...
My new post: why the “Human of the Gaps” fallacy distracts us from governing real-world AI impacts.
doomscrollingbabel.manoel.xyz/p/the-retrea...
August 17, 2025 at 2:58 PM
Whether or not AIs “truly” understand is less important than how they reshape jobs, politics, and culture.
My new post: why the “Human of the Gaps” fallacy distracts us from governing real-world AI impacts.
doomscrollingbabel.manoel.xyz/p/the-retrea...
My new post: why the “Human of the Gaps” fallacy distracts us from governing real-world AI impacts.
doomscrollingbabel.manoel.xyz/p/the-retrea...
We then used annotated repurposing using LLMs, correcting for bias using DSL (naokiegami.com/dsl/). Altogether, we estimate that ~0.25% (≈3,400 channels) were repurposed within three months, together holding 44 million subscribers under the radar (!)
July 30, 2025 at 8:29 PM
We then used annotated repurposing using LLMs, correcting for bias using DSL (naokiegami.com/dsl/). Altogether, we estimate that ~0.25% (≈3,400 channels) were repurposed within three months, together holding 44 million subscribers under the radar (!)
But we go beyond characterizing this individual marketplace of channels! To detect repurposing at scale, we drew a quasi-random sample of 1.4 M YouTube channels from Social Blade, scraped metadata in, and pulled historical snapshots via the Wayback Machine!
July 30, 2025 at 8:29 PM
But we go beyond characterizing this individual marketplace of channels! To detect repurposing at scale, we drew a quasi-random sample of 1.4 M YouTube channels from Social Blade, scraped metadata in, and pulled historical snapshots via the Wayback Machine!
Sites like Fameswap recorded over $1M in YouTube-channel sales in six months, with avg listings at $5.4K/100K subs, and some fetching >$100 K. Once sold, 37% of these channels are repurposed to push crypto scams, political propaganda, gambling, or get-rich-quick schemes
July 30, 2025 at 8:29 PM
Sites like Fameswap recorded over $1M in YouTube-channel sales in six months, with avg listings at $5.4K/100K subs, and some fetching >$100 K. Once sold, 37% of these channels are repurposed to push crypto scams, political propaganda, gambling, or get-rich-quick schemes