Skip to content
Get in touch

Research

We show the work.

We study how people talk, argue and make up their minds, and we publish what we find with the data. Every estimate with its interval.

What we study

Six questions about how people decide.

We work in three disciplines on one subject: how people talk, argue and make up their minds.

Conversation modeling

Who said what, to whom, and what it changed. We model conversations as they unfold, from the words on the page to the people in the room.

Systems engineering

The machinery under every reading: language-based controls from the web to the server, built to give the same answer every time, on less compute.

Interaction design

How design, media and cues steer what people do, and how to build interfaces that sharpen a person's thinking instead of doing it for them.

Point at a question to see what we look for.

Reasoning

How people weigh a claim against the facts.

1 student paper

Persuasion

Who an argument wins over, and why.

3 student papers

Attention

What noise costs people, and who ends up paying.

2 student papers

Metacognition

How people think about their own thinking.

Write the first paper.

Interaction

How design, media and cues steer what we do.

2 student papers

Engagement

Which clicks are real, and what people recall.

2 student papers

The instrument

Labels were the ceiling.

Aristotle named ethos, pathos, logos and kairos. Two thousand years later, Robert Cialdini named six principles of influence. A label tells you a lever is present; it cannot say how hard it presses, on whom, or when. Fields stall at the label stage until an instrument arrives.

  1. SignalThe message, in order: text by position, audio by time, video by frame.
  2. Rhetorical MassHow much was said: claims, entities, structure, fallacies. The same text always gives the same mass.
  3. Persuasive VelocityHow well it was said, across eight channels, down to leaves that each carry a score and its reasons.
  4. Momentum and forceWhat moved. Force is what it takes to change the course of a conversation.
  5. The personaWho is listening: who they are, what moves them, their state of mind, their setting.
  6. Persuasive ResonanceDid it land? Four probabilities, one person, one message, one moment.

What sets it apart

Persuasive Resonance

Did it land? One person, one message, one moment: four probabilities that always sum to one.

See a readout

Stateful Personas

A persona that remembers. Who is listening carries from one message to the next: who they are, what moves them, their state of mind and their setting.

Configure one in the interactive demo

Objective-based Scoring

Every score is computed against a goal someone stated, so a number always answers a question: better at what, and for whom.

Persuasive Velocity

Eight channels.

How well something was said, read on eight channels, each down to leaves that carry a score and their reasons.

EthosDo I believe the source?
PathosDoes it reach me through feeling?
LogosDo the steps connect?
KairosIs this the right moment, at every scale?
StyleHow vividly is it put?
ClarityDoes it feel easy to grasp?
FocusDoes it capture and hold attention?
AwarenessDo I see the move as a move? The channel Pulp cares about most.

Persuasive Resonance

Did it land?

At the top of the instrument a message meets a person, and the readout reports four probabilities that always sum to one.

An illustration, not a result. Every number in a real readout carries its stated reasons, and every score is computed against a goal someone stated.

persuasion.wiki

Every term we measure, defined.

The persuasion ontology: every term Pulp uses to measure persuasion, defined in real and plain words, with how each one connects to the others.

Open persuasion.wiki

What we're finishing

Three papers nearly done, and the models underneath.

Implementing

Valid results from AI responses

We pair AI responses with a smaller human sample, so the estimate stays unbiased. A peer-reviewed method from 2026, new to our work.

Building

Graph neural network foundations

Models that learn from who talks to whom: the base our machine learning builds on.

Finalizing

Persuasive resonance

How closely a message fits the person hearing it.

Finalizing

Persuasive force

How hard an argument pushes, and what it moves.

Finalizing

Stateful persona models

Personas that remember, from one message to the next.

Long shadows of people walking, seen from above on a pale street.
On the way to the next talk.Photo: Dibakar Roy

The record

Where the work has been.

Seminars, roundtables and classes. Invite us, and we bring the data.

Invite us to speak

ASCCriminology · nextASASociologyUFSeminarsEmoryBusiness analyticsINFORMSOperations research
Five rooms, from criminology to operations research. Next: Chicago, November 2026.
  • American Society of Criminology

    Coming up

    Annual meeting, Chicago, November 2026: new social media research, a companion to our fentanyl policy roundtable.
  • American Sociological Association

    Presented

    2026 roundtable: fentanyl policy and the election cycle.
  • University of Florida

    Given

    Seminars.
  • Emory University, MSBA program

    Given

    A speaking event.
  • INFORMS

    Accepted

    2026 Annual Meeting.

Data stories

New

The poll and the open conversation disagree. The wording explains most of it.

Read it

Coming soon

From Stone to Screen

Coming soon

How people ask, and what it costs to understand them

Coming soon

Understand first: how Pulp cuts the electricity AI uses

Pulp

Color mode

Roll into a random page on Pulp. A few hidden pieces can only be accessed by taking a chance.

Get in touch

Color mode

Kept on this device only.

What are you trying to understand?

So we know who to write to.
An address we can reply to, like name@example.com.
A question, a dataset, an argument that keeps going in circles.

Get in touch
Roll into a random page on Pulp. A few hidden pieces can only be accessed by taking a chance.