Manoel Horta Ribeiro

Come work with Homa!! She’s amazing!!

Trying an experiment in good old-fashioned blogging about papers: dallascard.github.io/granular-mat...

Home - The Accelerator Speeding Research on the Information Environment We Need Faster, More Efficient Research The information environment has far-reaching impacts, shaping our mental health, our response to everyday event...

This work was led by Abdurahman Maroouf, who did a fantastic job over the summer (!)

With: Kevin T Greene, Stefan Feuerriegel, @jshapiro.bsky.social

It was also made possible by this amazing initiative, the Research Accelerator (researchaccelerator.org).

Putting it together:

Short-form video platforms act as a structural driver of mobile usage.

They don’t just change what people watch; they change how much and how often they use their phones: longer total use, shorter breaks, more entrenched checking routines.

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.

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.

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

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%.

We use a matched, stacked DiD design:
- Treatment: people who adopt TikTok
- Control: people who adopt Instagram or Facebook in the same period

Outcomes:
- total daily mobile duration
- number of sessions
- average time away from phone (TAP)

We use passively collected, individual-level data from a large U.S. mobile panel (N=1,764), tracking every app and website people used on their phones.

We focus on a period when TikTok offered short-form video and Instagram/Facebook did not! Short-form video was the key diff!

Short-form video raises concerns about compulsive use and well-being, but most evidence is correlational.

We ask a simple question: What happens to people’s overall phone behavior when they start using a short-form video platform?

Do they just swap one app for another?

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...

⚖️ Measuring Scalar Constructs in Social Science with LLMs

with rising (and established) stars in Computational Social Science

@haukelicht.bsky.social
@rupak-s.bsky.social
@patrickwu.bsky.social
@pranavgoel.bsky.social
@elliottash.bsky.social
@alexanderhoyle.bsky.social

arxiv.org/abs/2509.03116

🚨 I'm recruiting PhD students in Computer Science at Johns Hopkins University for Fall 2026. If you're interested in AI, HCI, and designing better online platforms and experiences, apply to work with me!
More info: piccardi.me

Have been waiting for this paper to come out ever since Hongyao told me about it -- more academics should be taking advantage of such open data to answer important questions

wild :o

There’s plenty of evidence for political bias in LLMs, but very few evals reflect realistic LLM use cases — which is where bias actually matters.

IssueBench, our attempt to fix this, is accepted at TACL, and I will be at #EMNLP2025 next week to talk about it!

New results 🧵

I argue that if we consider these three points, we find that labeling with LLMs is neither trick nor treat. Treated as measurement instruments, their value lies in forcing us to confront uncertainty we once ignored; not in completely eliminating it.

I cluster work in this area broadly into three waves: the “wow” phase (e.g., Gillardi’s PNAS paper), the “how do we do this right?” phase (e.g., Egami’s DSL), and the “the boat is on fire” wave (e.g., Baumann’s LM hacking).

Large language models are quietly transforming how social scientists label data. In dozens of new studies, undergrad coders and Turkers have been replaced by GPT-5 or Gemini 2.5 (or whatever new model just arrived). What began as a convenience is becoming a methodological shift.

Labeling Data with Language Models: Trick or Treat? Large language models are now labeling data for us.

The debate over “LLMs as annotators” feels familiar: excitement, backlash, and anxiety about bad science. My take in a new blogpost is that LLMs don’t break measurement; they expose how fragile it already was.

doomscrollingbabel.manoel.xyz/p/labeling-d...

Graze is doing great work in supporting the entire ATProto Ecosystem, and their latest newsletter is a great illustration. Also featuring an interview with @sjgreenwood.bsky.social about Paper Skygest (bsky.app/profile/pape...)

One postdoctoral Research Position Deadline: November 15th, 2025

Join us as postdoc at the Inequality Discourse Observatory at the University of Konstanz: stellen.uni-konstanz.de/jobposting/7...
We will do epic research between Linguistics and Computational Social Science at the Cluster of Politics of Inequality. Feel free to DM if you have any questions.

None of this is `hard'—great material already exists (Brady Neal on causality, Moritz Hardt on benchmarks, etc.). What's missing is mindset: causality, regression, and experimental design must become core to how we train computer scientists—not optional extras.

I elaborate on what I think should be taught. It boils down to (at least) four things:
1 causality: how to pose and identify effects
2 regression: as a tool for inference, not prediction
3 benchmarks: as measurements, not trophies
4 experiments: with rigor, power, and ethics

Success is measured by benchmarks, not by robustness or causal clarity. Yet more and more papers now make causal claims --- from HCI to NLP, ML to Security and Privacy.

Why the contrast? Because the two fields treat empiricism in opposite ways.

Econometrics was forged in the crucible of skepticism. Every paper is a defensive war against omitted variables, selection bias, etc. Yet, CS (and ML) was built on demonstration, not falsification ...

I'd posit a similar, flipped version of the law for ML:

> When an economist reads (and understands) an empirical machine learning study done after 2022, the probability that they will think of an objection that the researcher has failed to take into account is close to one.

Henderson’s first law of econometrics reads:

> When you read an econometric study done after 2005, the probability that the researcher has failed to take into account an objection that a non-economist will think of is close to zero.

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...