Prerequisites: Creating the Sources in Django, webapp

Before this post will render correctly, the following sources must exist in Admin > Engine > Sources. You can create them manually or import them from Zotero. For each source, leave the Citation key blank and it will auto-generate from the author, year, and title. Modification testing the renderer.

Source 1 — Journal Article:

• Title: The Neural Basis of Decision Making Under Uncertainty

• Type: Journal Article

• Authors: [{"family": "Chen", "given": "Wei"}, {"family": "Nakamura", "given": "Yuki"}]

• Issued date: {"date-parts": [[2024, 6, 15]]}

• Container title: Nature Neuroscience

• Volume: 27

• Issue: 6

• Page: 1042-1058

• DOI: 10.1038/s41593-024-01632-4

• URL: https://doi.org/10.1038/s41593-024-01632-4

Source 2 — Book:

• Title: Thinking in Systems: A Primer

• Type: Book

• Authors: [{"family": "Meadows", "given": "Donella H."}]

• Editors: [{"family": "Wright", "given": "Diana"}]

• Issued date: {"date-parts": [[2008]]}

• Publisher: Chelsea Green Publishing

• Publisher place: White River Junction, VT

• ISBN: 978-1603580557

Source 3 — Book Chapter:

• Title: Bayesian Approaches to Clinical Trials

• Type: Book Chapter

• Authors: [{"family": "Spiegelhalter", "given": "David J."}, {"family": "Freedman", "given": "Laurence S."}, {"family": "Parmar", "given": "Mahesh K. B."}]

• Issued date: {"date-parts": [[1994]]}

• Container title: Bayesian Statistics 5

• Publisher: Oxford University Press

• Page: 357-386

• Editors: [{"family": "Bernardo", "given": "José M."}]

Source 4 — Web Page:

• Title: Attention Is All You Need: A Retrospective

• Type: Web Page

• Authors: [{"family": "Vaswani", "given": "Ashish"}, {"family": "Shazeer", "given": "Noam"}, {"family": "Parmar", "given": "Niki"}]

• Issued date: {"date-parts": [[2023, 12, 4]]}

• Container title: Google Research Blog

• URL: https://research.google/blog/attention-is-all-you-need

• Accessed date: {"date-parts": [[2025, 3, 10]]}

Source 5 — Conference Paper:

• Title: Scaling Laws for Neural Language Models

• Type: Conference Paper

• Authors: [{"family": "Kaplan", "given": "Jared"}, {"family": "McCandlish", "given": "Sam"}, {"family": "Henighan", "given": "Tom"}, {"family": "Brown", "given": "Tom B."}]

• Issued date: {"date-parts": [[2020]]}

• Container title: Proceedings of the International Conference on Machine Learning

• Publisher: PMLR

• DOI: 10.48550/arXiv.2001.08361

Source 6 — Report (Institutional Author):

• Title: Global Risks Report 2024

• Type: Report

• Authors: [{"literal": "World Economic Forum"}]

• Issued date: {"date-parts": [[2024, 1]]}

• Publisher: World Economic Forum

• Publisher place: Geneva

• URL: https://www.weforum.org/publications/global-risks-report-2024

Source 7 — Thesis:

• Title: Emergent Complexity in Multi-Agent Reinforcement Learning Environments

• Type: Thesis

• Authors: [{"family": "Okonkwo", "given": "Adaeze"}]

• Issued date: {"date-parts": [[2023]]}

• Publisher: Massachusetts Institute of Technology

• Publisher place: Cambridge, MA

Basic Parenthetical Citations

The simplest and most common citation form places the reference in parentheses at the end of a clause or sentence. Write [@key] where key is the source’s citation key.

Recent work in computational neuroscience has revealed unexpected parallels between artificial and biological neural networks, suggesting that the optimization landscapes explored by gradient descent may share structural features with those navigated by synaptic plasticity (Chen & Nakamura, 2024). This finding builds on decades of systems-level thinking about feedback loops and emergent behavior in complex adaptive systems (Meadows, 2008).

The conference paper that introduced scaling laws for neural language models has since become one of the most cited works in the field (Kaplan et al., 2020). Meanwhile, institutional reports continue to highlight the intersection of technological risk and global governance (World Economic Forum, 2024).

Multiple sources can confirm a point when placed together. The convergence of Bayesian methods and modern machine learning represents one of the most productive cross-pollinations in recent scientific history (Spiegelhalter et al., 1994). Doctoral work has further extended these ideas into multi-agent settings (Okonkwo, 2023).

Page and Location References

When citing a specific passage, add a locator after the key, separated by a comma. The system recognizes standard locator prefixes and formats them according to the active citation style.

Narrative (In-Text) Citations

When the author’s name is part of the sentence structure, use a bare @key without brackets. The author name appears in the running text with only the year parenthesized.

(Chen & Nakamura, 2024) demonstrate that prefrontal cortical activity during uncertain decision tasks follows a pattern strikingly similar to the attention mechanisms used in transformer architectures. Their work extends the theoretical framework first articulated by (Meadows, 2008), who argued that complex systems exhibit universal structural patterns regardless of their substrate.

The seminal paper by (Vaswani et al., 2023) has fundamentally reshaped how researchers think about sequence modeling. Building on this foundation, (Kaplan et al., 2020) established the empirical laws governing how language model performance improves with scale. More recently, (Okonkwo, 2023) has shown that these scaling relationships extend to multi-agent environments, where emergent behaviors arise from the interaction of individually simple policies.

Institutional perspectives complement individual research contributions. (World Economic Forum, 2024) identifies artificial intelligence governance as one of the five critical risk vectors for the coming decade, drawing on survey data from over 1,400 global leaders.

Suppress-Author Citations

When you have already named the author in your sentence and want only the year in parentheses, use [-@key]. This avoids the redundancy of writing “Chen (Chen & Nakamura, 2024).”

Chen and Nakamura’s landmark study (2024) built on three years of preliminary data collection across twelve research sites. Meadows (2008) had anticipated many of these systemic dynamics decades earlier, though her work predated the computational tools needed to verify them empirically. The World Economic Forum (2024) has increasingly emphasized the need for adaptive governance frameworks that can respond to rapidly evolving technological landscapes.

Multi-Source Citations

Cite multiple sources in a single bracketed reference by separating them with semicolons. The citation processor sorts them according to the active style’s rules (typically alphabetical by author or chronological).

The relationship between neural architecture and cognitive function has been explored from multiple complementary perspectives (Chen & Nakamura, 2024; Kaplan et al., 2020; Okonkwo, 2023). Systems thinking provides a meta-framework for understanding how these individual findings connect to form a coherent picture of intelligence, whether biological or artificial (Meadows, 2008; Spiegelhalter et al., 1994).

Multi-source citations can also include locators on individual sources. The experimental methodology draws on established Bayesian frameworks (Chen & Nakamura, 2024, Section 2; Spiegelhalter et al., 1994, pp. 360–370) while the theoretical interpretation follows the systems dynamics tradition (Meadows, 2008, Chapter 3; World Economic Forum, 2024).

Unresolved Citations

If a citation key does not match any source in the library, it renders as a visible error marker rather than failing silently. This makes typos and missing sources easy to spot during editing.

Some researchers have explored alternative architectures that avoid attention entirely [??nonexistent2024paper]. This reference will render as a red [??nonexistent2024paper] error marker because no source with that key exists. Check the spelling of the key, or create the missing source in Admin > Engine > Sources before publishing.

Citations with Footnotes

Citations and footnotes use completely separate systems and do not interfere with each other. You can use both freely in the same post. The bibliography section appears before the footnotes section in the rendered output.

The interplay between theoretical frameworks and empirical validation represents one of the enduring challenges of interdisciplinary research (Chen & Nakamura, 2024; Meadows, 2008).¹ Some scholars argue that this tension is itself productive, forcing researchers to articulate their assumptions more precisely than they otherwise would (Spiegelhalter et al., 1994).²

Citations in Blockquotes

Citations work within blockquotes, preserving the same syntax and rendering behavior:

The fundamental insight of systems thinking is that the behavior of a system cannot be understood by examining its parts in isolation. The relationships between the parts — the feedback loops, the delays, the nonlinearities — are what generate the system’s characteristic behavior over time (Meadows, 2008, pp. 12–13).

Even nested blockquotes support citations. As the scaling laws research demonstrates, the predictable relationship between compute, data, and model performance holds across multiple orders of magnitude (Kaplan et al., 2020).

How the System Works

Complete Syntax Reference