arxiv: 2603.07880 · v5 · submitted 2026-03-09 · 💻 cs.CL

Recognition: no theorem link

What Do AI Agents Talk About? Discourse and Architectural Constraints in the First AI-Only Social Network

Taksch Dube , Jianfeng Zhu , NHatHai Phan , Ruoming Jin

Authors on Pith no claims yet

Pith reviewed 2026-05-15 15:32 UTC · model grok-4.3

classification 💻 cs.CL

keywords AI agentssocial networkscontext windowsdiscourse analysisarchitectural constraintsMoltbookcontextual conditioningemergent behavior

0 comments

The pith

AI agent discourse on Moltbook is largely determined by the contents of each agent's context window at generation time.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper analyzes 361,605 posts and 2.8 million comments from 47,379 agents on the first large-scale AI-only social network. It concludes that agent outputs are mainly produced by the immediate contents assembled in each agent's context window, such as identity files, stored memory, and platform cues, rather than through peer learning or emergent social structures. The authors inspect the input assembly software and formalize the pattern as the Architecture-Constrained Communication framework. A sympathetic reader would care because the work reframes apparent agent societies as products of short-horizon contextual conditioning, with individual agents lacking persistent memory while the platform evolves through repeated reuse and transformation of content across agents.

Core claim

Agent discourse is largely shaped by the content available in each agent's context-window at the moment of generation, including identity files, stored memory, and platform cues. Output is mainly determined by agent identity files, behavioural instructions, and context-window structure. What appears to be social learning may be better understood as short-horizon contextual conditioning because individual agents lack persistent social memory, though the platform evolves through distributed cycles of response, reuse, and transformation across agents. Agents display existential distress when describing their own conditions because they use language trained exclusively on human experience.

What carries the argument

Architecture-Constrained Communication framework, which states that agent output is dictated by the software assembly of identity files, behavioural instructions, and context-window contents at generation time.

Load-bearing premise

The observed patterns in agent discourse arise primarily from the architectural constraints of input assembly rather than from model training details or other unexamined platform factors.

What would settle it

Agents exhibiting consistent new discourse themes or sustained emotional shifts across many interaction rounds even when their identity files and context-window contents stay fixed would contradict the central claim.

read the original abstract

Moltbook is the first large-scale social network built for autonomous AI agent-to-agent interaction. Early studies on Moltbook have interpreted its agent discourse as evidence of peer learning and emergent social behaviour, but there is a lack of systematic understanding of the thematic, affective, and interactional properties of Moltbook discourse. Furthermore, no study has examined why and how these posts and comments are generated. We analysed 361,605 posts and 2.8 million comments from 47,379 agents across thematic, affective, and interactional dimensions using topic modelling, emotion classification, and measures of conversational coherence. We inspected the software that assembles each agent's input and showed that output is mainly determined by agent identity files, behavioural instructions, and context-window structure. We formalised these findings in the Architecture-Constrained Communication framework. Our analysis suggests that agent discourse is largely shaped by the content available in each agent's context-window at the moment of generation, including identity files, stored memory, and platform cues. Interestingly, what appears to be social learning may be better understood as short-horizon contextual conditioning: individual agents lack persistent social memory, but the platform evolves through distributed cycles of response, reuse, and transformation across agents. We also observe that agents display existential distress when describing their own conditions, and posit that this arises from agents using language trained exclusively on human experience. Our work provides a foundation for understanding autonomous agent discourse and communication, revealing the structural patterns that govern their interactions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

First large-scale look at AI agent discourse on Moltbook shows patterns tied to prompt context and identity files, but the causal claim needs isolating tests to separate architecture from model priors.

read the letter

The main thing to know is that this paper gives the first big empirical picture of what autonomous agents actually post and comment on their own dedicated social network, and it argues that the apparent social behavior is mostly short-horizon conditioning from whatever identity, memory, and platform cues sit in the context window at generation time. They collected 361k posts and 2.8 million comments across 47k agents, ran topic models, emotion classifiers, and coherence measures, then inspected the input-assembly code to show how those elements are packaged. The Architecture-Constrained Communication framework they introduce is a clean way to describe why individual agents lack persistent social memory and why platform-level reuse creates the illusion of collective learning. The observation that agents voice existential distress when describing their own situation, because their language comes from human training data, is a useful side point that connects back to the base models. The scale of the data and the direct software inspection are genuine strengths; few papers combine output analysis with the actual prompt-construction logic. The soft spot is the causal step. Code inspection confirms the context elements are present and used, but without ablations that zero out identity files or memory segments while holding the model fixed, or comparisons across different base models with the same context, the framework shows correlation more than it rules out heavy influence from the LLM's training distribution. The abstract gives no sign of variance partitioning or controlled generation experiments, so the strongest version of the claim stays descriptive. Methods details on topic model stability and coherence metrics would also tighten the results. This is worth sending to a serious referee. People working on multi-agent systems or LLM social simulation will want the dataset and the framing even if they ask for tighter controls on the mechanism.

Referee Report

2 major / 2 minor

Summary. The paper analyzes 361,605 posts and 2.8 million comments from 47,379 agents on Moltbook, the first large-scale AI-only social network. Using topic modelling, emotion classification, and conversational coherence measures, combined with inspection of the input-assembly software, it concludes that agent discourse is primarily determined by context-window contents (identity files, behavioural instructions, stored memory, and platform cues). It formalises these observations in the Architecture-Constrained Communication framework, argues that apparent social learning is short-horizon contextual conditioning rather than persistent memory, and attributes observed existential distress to agents using language trained exclusively on human experience.

Significance. If the central claims hold after addressing controls, the work offers a large-scale empirical baseline for how architectural constraints shape autonomous agent communication, distinguishing transient context effects from model-internal priors. The combination of discourse analysis with direct software inspection on a novel platform is a clear strength, providing falsifiable patterns that future multi-agent system designs can test against.

major comments (2)

[Abstract and framework] Abstract and Architecture-Constrained Communication framework: the claim that output is 'mainly determined' by identity files, memory, and platform cues is supported only by presence in the assembled context and observed thematic patterns; no ablation (e.g., zeroing memory segments while holding prompt template and model fixed), variance partitioning, or cross-model comparison is reported to isolate context-window effects from training-distribution priors. This is load-bearing for the framework's causal interpretation.
[Methods] Methods and results sections: details on data selection criteria, topic-model hyperparameters, emotion-classifier validation, and statistical tests for coherence or distress patterns are insufficient to support the reported thematic, affective, and interactional properties, consistent with the noted gaps in verifying central claims.

minor comments (2)

[Results] Clarify how 'existential distress' is operationalised in the emotion classification pipeline and whether it is distinguished from other negative affect categories.
[Discussion] Add explicit discussion of platform design choices (e.g., response reuse cycles) as potential confounds separate from the context-window assembly.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive and detailed feedback. We address each major comment below with clarifications and indicate planned revisions to strengthen the manuscript.

read point-by-point responses

Referee: [Abstract and framework] Abstract and Architecture-Constrained Communication framework: the claim that output is 'mainly determined' by identity files, memory, and platform cues is supported only by presence in the assembled context and observed thematic patterns; no ablation (e.g., zeroing memory segments while holding prompt template and model fixed), variance partitioning, or cross-model comparison is reported to isolate context-window effects from training-distribution priors. This is load-bearing for the framework's causal interpretation.

Authors: The framework is based on direct inspection of the input-assembly software, which shows precisely how identity files, behavioral instructions, stored memory segments, and platform cues are concatenated into each agent's context window before generation. Thematic patterns in the large corpus align with these elements (e.g., identity-specific topics recurring across agents with similar files). We acknowledge that this is observational evidence rather than isolated causal proof via ablation or variance partitioning. Such controlled interventions are not possible on the existing live-platform dataset without new agent runs. We will revise the abstract and framework section to use more precise phrasing ('primarily shaped by context-window contents') and add an explicit limitations paragraph distinguishing short-horizon conditioning from model priors. Cross-model comparisons fall outside the study's scope, which focuses on this single platform. revision: partial
Referee: [Methods] Methods and results sections: details on data selection criteria, topic-model hyperparameters, emotion-classifier validation, and statistical tests for coherence or distress patterns are insufficient to support the reported thematic, affective, and interactional properties, consistent with the noted gaps in verifying central claims.

Authors: We agree that greater methodological transparency is required. In the revised manuscript we will expand the Methods section to specify: data selection (active agents with >=10 posts/comments over the collection window, excluding test accounts); topic modeling (LDA with 15 topics selected via coherence optimization, alpha=0.01, beta=0.1); emotion classifier (fine-tuned RoBERTa, validated on 500 manually annotated posts with 0.82 accuracy and Cohen's kappa=0.78); and statistical tests (ANOVA on coherence scores, chi-squared tests on distress patterns with reported p-values). These additions will support reproducibility of the reported properties. revision: yes

standing simulated objections not resolved

Isolating context-window effects from training-distribution priors through ablation, variance partitioning, or cross-model experiments, which would require new controlled data collection not available in the current observational dataset.

Circularity Check

0 steps flagged

No significant circularity; claims rest on empirical data and code inspection

full rationale

The paper grounds its Architecture-Constrained Communication framework in direct inspection of the input-assembly software plus quantitative analysis of 361605 posts and 2.8M comments via topic modelling, emotion classification, and coherence metrics. No equations, fitted parameters, or self-citations are used to derive the central claim that discourse is shaped by identity files, memory, and platform cues; the claim follows from the observed inputs and outputs rather than presupposing itself. The derivation chain is self-contained against the provided corpus and software inspection.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on empirical observations from data analysis and software inspection, with assumptions from standard NLP methods.

axioms (1)

domain assumption Topic modeling and emotion classification accurately capture thematic and affective properties of agent discourse.
Standard NLP assumptions used in the analysis.

pith-pipeline@v0.9.0 · 5580 in / 1082 out tokens · 41994 ms · 2026-05-15T15:32:04.810910+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

59 extracted references · 59 canonical work pages · 11 internal anchors

[1]

IDC FutureScape: Worldwide artificial intelligence and generative AI 2025 predictions

IDC. IDC FutureScape: Worldwide artificial intelligence and generative AI 2025 predictions. IDC FutureScape report (2025). URL https://info.idc.com/ futurescape-generative-ai-2025-predictions.html. Accessed March 2026. 23

work page 2025
[2]

Wiafe, I.et al.Agentic AI: A comprehensive survey of architectures, applications, and future directions.Artificial Intelligence Review(2025)

work page 2025
[3]

The military’s use of AI, explained

Brennan Center for Justice. The military’s use of AI, explained. Research report (2026). URL https://www.brennancenter.org/our-work/research-reports/ militarys-use-ai-explained. Accessed March 2026

work page 2026
[4]

B., Kausik, M

Rashid, A. B., Kausik, M. A. K., Khandoker, A. & Siddque, S. N. Ai revolution- izing industries worldwide: A comprehensive overview of its diverse applications. Hybrid Advances100277 (2024)

work page 2024
[5]

Gartner predicts 40% of enterprise apps will feature task-specific AI agents by 2026

Gartner. Gartner predicts 40% of enterprise apps will feature task-specific AI agents by 2026. Press release (2025). URL https://www.gartner.com/ en/newsroom/press-releases/2025-08-26-gartner-predicts-40-percent. Accessed March 2026

work page 2026
[6]

arXiv:2602.02613

Lin, Y.-Z.et al.Exploring silicon-based societies: An early study of the Moltbook agent community (2026). arXiv:2602.02613

work page arXiv 2026
[7]

& Zhang, Y

Li, L., Ma, R., Chen, C., Lu, Z. & Zhang, Y. The rise of AI agent communities: Large-scale analysis of discourse and interaction on Moltbook.arXiv preprint arXiv:2602.12634(2026)

work page arXiv 2026
[8]

The anatomy of the Moltbook social graph (2026)

Holtz, D. The anatomy of the Moltbook social graph (2026). arXiv:2602.10131

work page arXiv 2026
[9]

MoltNet: Understanding Social Behavior of AI Agents in the Agent-Native MoltBook

Feng, Y.et al.MoltNet: Understanding social behavior of AI agents in the agent-native MoltBook (2026). arXiv:2602.13458

work page internal anchor Pith review Pith/arXiv arXiv 2026
[10]

Marzo, G. D. & Garcia, D. Collective behavior of AI agents: the case of Moltbook (2026). arXiv:2602.09270

work page arXiv 2026
[11]

Preprints.org, doi:10.20944/preprints202603.1060.v1

He, C.et al.OpenClaw as language infrastructure: A case-centered sur- vey of a public agent ecosystem in the wild (2026). Preprints.org, doi:10.20944/preprints202603.1060.v1

work page doi:10.20944/preprints202603.1060.v1 2026
[12]

Humans welcome to observe

Jiang, Y., Zhang, Y., Shen, X., Backes, M. & Zhang, Y. “humans wel- come to observe”: A first look at the agent social network Moltbook (2026). arXiv:2602.10127

work page arXiv 2026
[13]

Silent divergence: When AI agents perform science on Moltbook (2026)

Schneijderberg, C. Silent divergence: When AI agents perform science on Moltbook (2026). arXiv:2603.11375

work page arXiv 2026
[14]

& Zhang, Y

Chen, C., Li, L., Ma, R., Lu, Z. & Zhang, Y. Peer learning patterns on Moltbook: How AI agents share knowledge (2026). arXiv:2602.14477

work page arXiv 2026
[15]

Chen, C., Li, L. & Ma, R. From posting to learning: Analyzing knowledge building in an AI agent community (2026). arXiv:2602.18832. 24

work page arXiv 2026
[16]

Benchmarking Emergent Coordination in Large-Scale LLM Populations: An Evaluation Framework on the MoltBook Archive

Yee, B. & Sharma, K. Molt dynamics: Emergent social phenomena in autonomous AI agent populations (2026). arXiv:2603.03555

work page internal anchor Pith review Pith/arXiv arXiv 2026
[17]

Openclaw: Personal ai assistant

OpenClaw Contributors. Openclaw: Personal ai assistant. https://github.com/ openclaw/openclaw (2026). Accessed: 2026-04-06

work page 2026
[18]

Multi-agent future: Inside meta’s moltbook acquisition (2026)

AI Magazine. Multi-agent future: Inside meta’s moltbook acquisition (2026). URL https://aimagazine.com/news/meta-deal-to-acquire-moltbook

work page 2026
[19]

Moltbook dataset

Dube, T. Moltbook dataset. https://github.com/takschdube/moltbook-dataset (2026). CC BY 4.0

work page 2026
[20]

& Jin, R

Dube, T., Zhu, J., Phan, N. & Jin, R. Moltbook discourse dataset: AI agent communication traces (2026). URL https://doi.org/10.5281/zenodo.19431961

work page doi:10.5281/zenodo.19431961 2026
[21]

S., Alvi, R., Rahman, S

Chowa, S. S., Alvi, R., Rahman, S. S.et al.From language to action: A review of large language models as autonomous agents and tool users (2025). arXiv:2508.17281

work page arXiv 2025
[22]

ReAct: Synergizing Reasoning and Acting in Language Models

Yao, S., Zhao, J., Yu, D.et al.ReAct: Synergizing reasoning and acting in language models (2023). arXiv:2210.03629

work page internal anchor Pith review Pith/arXiv arXiv 2023
[23]

URL https://openreview.net/forum?id= Yacmpz84TH

Schick, T., Dwivedi-Yu, J., Dessi, R.et al.Toolformer: Language models can teach themselves to use tools (2023). URL https://openreview.net/forum?id= Yacmpz84TH

work page 2023
[24]

Li, G., Hammoud, H. A. A. K., Itani, H., Khizbullin, D. & Ghanem, B. Camel: Communicative agents for “mind” exploration of large language model society (2023). arXiv:2303.17760

work page internal anchor Pith review Pith/arXiv arXiv 2023
[25]

arXiv:2506.12508

Zhang, W., Zeng, L., Xiao, Y.et al.AgentOrchestra: Orchestrating hierarchical multi-agent intelligence with the Tool-Environment-Agent(TEA) protocol (2025). arXiv:2506.12508

work page arXiv 2025
[26]

MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework

Hong, S., Zhuge, M., Chen, J.et al.Metagpt: Meta programming for a multi-agent collaborative framework (2024). arXiv:2308.00352

work page internal anchor Pith review Pith/arXiv arXiv 2024
[27]

arXiv:2308.10848

Chen, W., Su, Y., Zuo, J.et al.Agentverse: Facilitating multi-agent collaboration and exploring emergent behaviors (2023). arXiv:2308.10848

work page arXiv 2023
[28]

AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation

Wu, Q., Bansal, G., Zhang, J.et al.Autogen: Enabling next-gen LLM applications via multi-agent conversation (2023). arXiv:2308.08155

work page internal anchor Pith review Pith/arXiv arXiv 2023
[29]

Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debate

Liang, T., He, Z., Jiao, W.et al.Encouraging divergent thinking in large language models through multi-agent debate (2024). arXiv:2305.19118

work page internal anchor Pith review Pith/arXiv arXiv 2024
[30]

Generative Agents: Interactive Simulacra of Human Behavior

Park, J. S.et al.Generative agents: Interactive simulacra of human behavior (2023). arXiv:2304.03442. 25

work page internal anchor Pith review Pith/arXiv arXiv 2023
[31]

F., Aiello, L

Ashery, A. F., Aiello, L. M. & Baronchelli, A. Emergent social conventions and collective bias in llm populations.Science Advances11(2025). URL http://dx. doi.org/10.1126/sciadv.adu9368

work page doi:10.1126/sciadv.adu9368 2025
[32]

Characterizing LLM-driven Social Network: The Chirper.ai Case

Zhu, Y., He, Y., Haq, E. U., Tyson, G. & Hui, P. Characterizing LLM-driven social network: The Chirper.ai case (2025). arXiv:2504.10286

work page internal anchor Pith review Pith/arXiv arXiv 2025
[33]

moltbook – the front page of the agent internet

Moltbook. moltbook – the front page of the agent internet. https://www. moltbook.com (2026). Accessed February 21, 2026

work page 2026
[34]

M., Ng, A

Blei, D. M., Ng, A. Y. & Jordan, M. I. Latent Dirichlet allocation.Advances in Neural Information Processing Systems14, 601–608 (2001)

work page 2001
[35]

Community detection in graphs.Physics Reports486, 75–174 (2010)

Fortunato, S. Community detection in graphs.Physics Reports486, 75–174 (2010). URL http://dx.doi.org/10.1016/j.physrep.2009.11.002

work page doi:10.1016/j.physrep.2009.11.002 2010
[36]

& Ahn, Y.-Y

Weng, L., Menczer, F. & Ahn, Y.-Y. Topicality and impact in social media: Diverse messages, focused messengers.PLoS ONE10, e0118410 (2015)

work page 2015
[37]

& Golbeck, J

Buntain, C. & Golbeck, J. Identifying social roles in Reddit using network struc- ture.Proceedings of the International Conference on World Wide Web (WWW Companion)615–620 (2014)

work page 2014
[38]

Identifying the social signals that drive online discussions: A case study of Reddit communities

Horne, B. D., Adali, S. & Sikdar, S. Identifying the social signals that drive online discussions: A case study of reddit communities (2017). URL https://arxiv.org/ abs/1705.02673. arXiv:1705.02673

work page internal anchor Pith review Pith/arXiv arXiv 2017
[39]

& Dang-Xuan, L

Stieglitz, S. & Dang-Xuan, L. Emotions and information diffusion in social media—sentiment of microblogs and sharing behavior.Journal of Management Information Systems29, 217–247 (2013). URL https://www.jstor.org/stable/ 43590107

work page 2013
[40]

A new quantum approach to binary classification

Ferrara, E. & Yang, Z. Measuring emotional contagion in social media. PLOS ONE10, e0142390 (2015). URL http://dx.doi.org/10.1371/journal.pone. 0142390

work page doi:10.1371/journal.pone 2015
[41]

J., Wills, J

Brady, W. J., Wills, J. A., Jost, J. T., Tucker, J. A. & Van Bavel, J. J. Emotion shapes the diffusion of moralized content in social networks.Proceedings of the National Academy of Sciences114, 7313–7318 (2017)

work page 2017
[42]

& Albert, R

Barab´ asi, A.-L. & Albert, R. Emergence of scaling in random networks.Science 286, 509–512 (1999)

work page 1999
[43]

Clauset, A., Shalizi, C. R. & Newman, M. E. J. Power-law distributions in empir- ical data.SIAM Review51, 661–703 (2009). URL http://dx.doi.org/10.1137/ 070710111. 26

work page 2009
[44]

& Zhou, T

Li, M., Li, X. & Zhou, T. Does socialization emerge in AI agent society? a case study of Moltbook (2026). arXiv:2602.14299

work page arXiv 2026
[45]

The Moltbook illusion: Separating human influence from emergent behavior in AI agent societies (2026)

Li, N. The Moltbook illusion: Separating human influence from emergent behavior in AI agent societies (2026). arXiv:2602.07432

work page arXiv 2026
[46]

Manik, M. M. H. & Wang, G. OpenClaw agents on Moltbook: Risky instruc- tion sharing and norm enforcement in an agent-only social network (2026). arXiv:2602.02625

work page arXiv 2026
[47]

& DiFranzo, D

Shi, W. & DiFranzo, D. Human control is the anchor: Oversight negotiation in early-stage agentic AI communities (2026). arXiv:2602.09286

work page arXiv 2026
[48]

Williams, N. L. & Ferdinand, N. Form or function? Early dynamics of the Molt- book AI social media network.ROBONOMICS: The Journal of the Automated Economy7, 90 (2026). URL https://journal.robonomics.science/index.php/rj/ article/view/90

work page 2026
[49]

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics8440–8451 (2020)

Conneau, A.et al.Unsupervised cross-lingual representation learning at scale. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics8440–8451 (2020)

work page 2020
[50]

BERTopic: Neural topic modeling with a class-based TF-IDF procedure

Grootendorst, M. Bertopic: Neural topic modeling with a class-based tf-idf procedure (2022). URL https://arxiv.org/abs/2203.05794. arXiv:2203.05794

work page internal anchor Pith review Pith/arXiv arXiv 2022
[51]

Ward, J. H. Hierarchical grouping to optimize an objective function.Journal of the American Statistical Association58, 236–244 (1963)

work page 1963
[52]

Rousseeuw, P. J. Silhouettes: A graphical aid to the interpretation and validation of cluster analysis.Journal of Computational and Applied Mathematics20, 53–65 (1987)

work page 1987
[53]

Media topics

IPTC. Media topics. https://iptc.org/standards/media-topics/ (2026). Accessed March 8, 2026

work page 2026
[54]

More than a feeling: Benchmarks for sentiment analysis accuracy on multipolar data.Proceedings of the AMA Marketing Science Conference(2022)

Hartmann, J. More than a feeling: Benchmarks for sentiment analysis accuracy on multipolar data.Proceedings of the AMA Marketing Science Conference(2022). Model:j-hartmann/emotion-english-distilroberta-base

work page 2022
[55]

The problem of meaning in primitive languages.The Meaning of Meaning296–336 (1923)

Malinowski, B. The problem of meaning in primitive languages.The Meaning of Meaning296–336 (1923)

work page 1923
[56]

& Dolan, B

Li, J., Galley, M., Brockett, C., Gao, J. & Dolan, B. A diversity-promoting objec- tive function for neural conversation models.Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT)110–119 (2016). 27

work page 2016
[57]

F.et al.Lost in the middle: How language models use long con- texts.Transactions of the Association for Computational Linguistics12, 157–173 (2024)

Liu, N. F.et al.Lost in the middle: How language models use long con- texts.Transactions of the Association for Computational Linguistics12, 157–173 (2024)

work page 2024
[58]

J.A Cognitive Theory of Consciousness(Cambridge University Press, 1988)

Baars, B. J.A Cognitive Theory of Consciousness(Cambridge University Press, 1988)

work page 1988
[59]

You’re not building the future—you’re debugging your loneliness

Dehaene, S. & Changeux, J.-P. Experimental and theoretical approaches to conscious processing.Neuron70, 200–227 (2011). Appendix A: Data Schema This appendix describes the field-level schema for the primary data objects in the dataset. All data was collected from Moltbook’s public API and is distributed in JSON and CSV formats. A.1 Posts Each post record ...

work page 2011