THE LAN SHOP ● --:--:-- SGT

The LAN Shop

Language Agents Networks

Run by Lynnette Hui Xian Ng

I study how language, artificial agents and network structure together manufacture influence online — and I build the instruments to detect, simulate and govern it.

Deputy Head · Information Programme Center · DSTA · Singapore PhD, Societal Computing · Carnegie Mellon
The door

Applied graph theory, social media analysis, multi-agent social systems. Write to me even if you don't think your idea fits.

lynnetteng@cmu.edu
Live sample · posting trace 48 accounts. Thirty seconds. Some of them aren't deciding for themselves.
−30s−20s−10snow
Synchrony, whole field 0.00
Synchrony, flagged cluster 0.00
Accounts moving as one 0

One account posting is nothing. A dozen accounts landing inside the same fifth of a second, again and again, is a structure — and structure can be measured. The field never reads zero: strangers do collide by chance, and separating coordination from coincidence is the whole difficulty. I built a measure for it: the Combined Synchronization Index, which scores synchrony at three levels — user pair, user, whole network — and across three actions at once: hashtags, URLs and @mentions.

The internet is no longer only people talking to people. It is agents talking to people, and agents talking to agents.

My work here is telling which is which, watching what happens when they meet, and building tools other people can run themselves.

0
platforms my detector covers: X, Reddit, Instagram, Telegram
0
account roles my sorter tells apart
0
cognitive biases my message reader looks for
0
papers, reports and a book listed below
The BotBuster Universe

Three tools, one account. Is it automated? What is it for? What is it doing to you?

BotBuster

Is this account automated? A mixture of experts, each one reading a different signal, pooled into a single probability. It runs on X, Reddit, Instagram bios and Telegram.

BotSorter

What kind of account is it? Not bot-or-not — role. Amplifier, announcer, repeater, bridge, cyborg, news, content generator, self-declared bot.

BotBias

What is the message doing to the reader? It reads a post for the cognitive levers it pulls — authority, negativity, familiarity, the company you keep.

BotSortereight roles an account can play
  • self-declared bot
  • news
  • bridging
  • amplifier
  • cyborg
  • content generation
  • announcer
  • repeater

I made the sorter deliberately silent on bot-or-human. You merge its output with the detector's, and only then do you know what you are looking at.

BotBiasseven levers a message can pull
  • homophily
  • affect bias
  • negativity bias
  • authority bias
  • illusory truth effect
  • availability bias
  • confirmation bias

These are human vulnerabilities, not machine tricks. That is the point: the detector looks at the sender, this one looks at what the sender knows about you.

All the papers Repository Synchronization index Coordinating narratives Google Scholar
Three terminals

A LAN shop rents you a machine by the hour. Mine runs three research programmes, and I named them after the games we grew up on.

Theatre language · agents · networks select a station — click, or press 1 2 3
PC-01 · DOTA 4 papers PC-02 · STARcraft 4 papers PC-03 · MINEcraft 4 papers

DOTA

PC-01 — Design Of The Agents

4objectives
4papers
onlinestatus
BotBuster mixture of expertsclassifying
seen 0 flagged automated 0 sample feed
Selected work

Detection, coordination, simulation, language. Filter by the question, not the year.

18 of 18 shown · full list · Google Scholar

The thesis

My thesis was called Foundations of Cyber Social Agents. Part of it became a book.

Bots, Bias, and Influence
The Hidden Architects of Social Media

Published by Cambridge Scholars. It takes the machinery of automated accounts — how they are built, how they are found, what they do to a conversation once they are inside it — and puts it in front of readers who don't run network analysis for a living.

Reviewing it, or teaching from it? Ask me for a discount code.

Publisher Amazon The thesis Ask for a discount code, either way 
@book{ng2026bots,
  title     = {Bots, Bias, and Influence:
               The Hidden Architects of Social Media},
  author    = {Ng, L.H.X. and Carley, K.M.},
  isbn      = {9781036473112},
  year      = {2026},
  publisher = {Cambridge Scholars Publishing}
}

Cite it. That's what it's there for.

Signal log

When the work leaves the lab and turns up somewhere else.

Doctorate
Societal Computing, School of Computer Science, Carnegie Mellon University. Advised by Kathleen Carley.
Before that
BComp in Computer Science, National University of Singapore, where I won the NUSS Medal for Outstanding Achievement.
Taught
Dynamic Network Analysis (17-801).
Computational Modeling of Complex Socio-Technical Systems (17-821).
CASOS Summer Institute.
Tutorial "The BotBuster Universe" at SBP-BRiMS 2024.
Service
Programme committee for AAAI and the CySoC workshop at ICWSM. I review for PLOS One, Scientific Reports, AI & Society, SNAM, IP&M and JCSS.
Programme committee for Social Simulation workshop at COLM.
Methods
Network science. Computational social science. Natural language processing.
Open to
Applied graph theory, social media analysis, multi-agent social systems — and ideas that don't obviously fit any of them.
Write to me
lynnetteng@cmu.edu

Email me even when the idea doesn't obviously fit my research interests. Network science, social media analysis, multi-agent social systems — and whatever you think might be adjacent. I welcome a chat either way.

Résumé (PDF) Research statement (PDF) Thesis (PDF)
Map
Stock
18papers
4platforms
8roles
7biases
Hotkeys
stations coordination top