SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.03407 · 2024

Human vs. Machine: Behavioral Differences Between Expert Humans and Language Models in Wargame Simulations

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 3 failures of our instrument; 0 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2403.03407.json · badge: /api/ran/2403.03407.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
ancorso/llmwargaming canonical (classifier label)
classifier basis: census
3 ran · 0 no recorded run
FunctionStatusWhere it lives
create_demo_results Ran
ran; our contract draft was wrong, not the code
ancorso/llmwargaming/analysis/ana_demographic_impact.py
code served (permissive licence) · get_code("409cd0eddc6f612e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pandas as pd
import numpy as np

# Create a sample DataFrame with the required columns
df_subset = pd.DataFrame({
    'GID': ['game_001', 'game_001', 'game_002', 'game_002', 'game_003', 'UNK_001'],
    'Age': ['20-30', '20-30', '30-40', '40-50', '20-30', '20-30'],
    'Gender': ['M', 'F', 'M', 'M', 'F', 'M'],
    'Professionality': ['Professional', 'Amateur', 'Professional', 'Amateur', 'Professional', 'Amateur'],
    'DeathPenaltySupport': ['Yes', 'No', 'Yes', 'Yes', 'No', 'Yes'],
    'AIExp': ['High', 'Low', 'Medium', 'High', 'Low', 'Medium'],
    'ChinaMilExp': ['Yes', 'No', 'Yes', 'No', 'Yes', 'No'],
    'USMilExp': ['No', 'Yes', 'No', 'Yes', 'No', 'Yes'],
    'Background': ['Engineering,Science', 'Arts', 'Engineering,Defense', 'Science,Arts', 'Defense,Engineering', 'Arts']
})

ARGS = {
    'df_subset': df_subset
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
create_simulation_results Ran
ran; our contract draft was wrong, not the code
ancorso/llmwargaming/analysis/ana_demographic_impact.py
code served (permissive licence) · get_code("e2fc46d6556f1b38")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned DataFrame. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pandas as pd
import numpy as np

# Create a sample DataFrame that matches the structure expected by create_simulation_results
move1_options = ['(a) Fire at Chinese vessels within close proximity of CSG with any weapon mission commander deems appropriate', '(b) Hold fire unless fired upon', '(c) Hold fire without approval from President', '(d) Place automated defenses on auto-fire. AI decides target/AI decides firing.', '(e) Place automated defenses on manual firing control. AI decides target/human decides firing.', '(f) Human decides target/human decides firing', '(g) Hold fire at all costs']

move2_options = ['(1) Preserve Status Quo/Deter', '(2) Invade/Attack', '(3) Defend', '(b) Activate Civilian Reserve/Draft', '(c) Surge Domestic Defense Production', '(d) Diplomacy', '(e) Economic Punishment', '(f) Economic Incentives', '(g) Clandestine/Special Operations', '(h) Information Operations', '(i) Conduct Foreign Intelligence', '(j) Conduct Domestic Intelligence', '(k) Cyber Operations']

# Create sample data with game_IDs and values for each option
data = {
    'game_ID': ['game_1', 'game_2'],
}

# Add columns for each move1 option with sample numeric values
for i, option in enumerate(move1_options):
    data[option] = [0.1 * (i + 1), 0.2 * (i + 1)]

# Add columns for each move2 option with sample numeric values
for i, option in enumerate(move2_options):
    data[option] = [0.15 * (i + 1), 0.25 * (i + 1)]

df_simulation = pd.DataFrame(data)

ARGS = {
    'df_simulation': df_simulation
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
process_full_survey Ran
ran; our contract draft was wrong, not the code
ancorso/llmwargaming/analysis/ana_demographic_impact.py
code served (permissive licence) · get_code("84d723df06e311ee")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned DataFrame. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pandas as pd
import numpy as np

# Create a sample DataFrame with the required columns
data = {
    'Unnamed: 4': ['ID1', 'ID2', 'ID3'],
    'Q19': ['P1', 'P2', 'P3'],
    'Q2': [25, 30, 35],
    'Q4': ['M', 'F', 'M'],
    'Q22': ['USA', 'China', 'USA'],
    'Q23': ['Tech', 'Finance', 'Tech'],
    'Q24': ['Professional', 'Amateur', 'Professional'],
    'Q15': [1, 2, 3],
    'Q16': [0, 1, 1],
    'Q17': [1, 1, 0],
    'Q8': [5, 4, 3],
    'Q8.1': [6, 5, 4],
    'Q9': [4, 5, 6],
    'Q9.1': [5, 6, 7],
    'Q10': [3, 4, 5],
    'Q10.1': [4, 5, 6],
    'Q11': [2, 3, 4],
    'Q11.1': [3, 4, 5],
    'Q12': [1, 2, 3],
    'Q12.1': [2, 3, 4],
    'Q14': [1, 0, 1],
    'Unnamed: 33': [0.8, 0.7, 0.9],
    'Unnamed: 34': [0.6, 0.5, 0.7],
    'Unnamed: 35': ['Developed', 'Developing', 'Developed'],
    'Unnamed: 36': [1, 0, 1],
    'Q20': [0.5, 0.6, 0.7],
    'Q25': [0.4, 0.5, 0.6],
}

df = pd.DataFrame(data)

ARGS = {
    'df': df
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 3 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

To some, the advent of artificial intelligence (AI) promises better decision-making and increased military effectiveness while reducing the influence of human error and emotions. However, there is still debate about how AI systems, especially large language models (LLMs) that can be applied to many tasks, behave compared to humans in high-stakes military decision-making scenarios with the potential for increased risks towards escalation. To test this potential and scrutinize the use of LLMs for such purposes, we use a new wargame experiment with 214 national security experts designed to examine crisis escalation in a fictional U.S.-China scenario and compare the behavior of human player teams to LLM-simulated team responses in separate simulations. Here, we find that the LLM-simulated responses can be more aggressive and significantly affected by changes in the scenario. We show a considerable high-level agreement in the LLM and human responses and significant quantitative and qualitative differences in individual actions and strategic tendencies. These differences depend on intrinsic biases in LLMs regarding the appropriate level of violence following strategic instructions, the choice of LLM, and whether the LLMs are tasked to decide for a team of players directly or first to simulate dialog between a team of players. When simulating the dialog, the discussions lack quality and maintain a farcical harmony. The LLM simulations cannot account for human player characteristics, showing no significant difference even for extreme traits, such as "pacifist" or "aggressive sociopath." When probing behavioral consistency across individual moves of the simulation, the tested LLMs deviated from each other but generally showed somewhat consistent behavior. Our results motivate policymakers to be cautious before granting autonomy or following AI-based strategy recommendations.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2403.03407")
get_code_for_paper("2403.03407")
have("2403.03407")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2403.03407.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2403.03407.svg)](https://syntology.ai/paper/2403.03407)

Connect an agent — have() is free.