Where storytelling, access, and power collide

Access to a book has never been simple. What you can read, how you can share it, who gets to own it: these have always been questions of power as much as technology.

The rise of digital publishing promised to change that. A single device could hold thousands of titles. A story written in Jakarta could reach a reader in Lagos in seconds. The barriers of print, including cost, geography, and physical availability, seemed suddenly surmountable.

But as this research demonstrates, digital books have not made access equal. They have shifted the terrain of inequality, moved the barriers, and in some cases made them harder to see. Corporate monopolies, restrictive legal frameworks, and a persistent digital divide mean that for many readers in Indonesia and across the Global South, the digital bookshelf remains out of reach.

"The technology itself could easily make reading more democratic, but in practice it has been shaped by corporate interests, showing that access is driven more by power and profit than by innovation."

Research synthesis, ARTS3000 2025

This report synthesises research conducted by four Arts students from Macquarie University through the PACE program. Working directly with Poddium across one semester, Peter Turner, Avani Singh, Shun Cheng, and Kagnchanna Meas examined the forces shaping digital book access, the evolution of ebook formats, and the ethical landscape for AI in publishing. Their findings carry direct implications for how platforms like Poddium and the storytellers who depend on them should think and act.

The forces that decide who gets to read

Access to digital books is not primarily a technical problem. It is a sociopolitical one. Three interconnected forces, operating simultaneously and often invisibly, determine who can read, under what conditions, and on whose terms.

01
Corporate monopolies
Amazon's Kindle captured over 50% of print and 75% of ebook sales in the US by 2019. Its proprietary format, DRM controls, and willingness to revoke purchased content reveal a system where corporate priorities override reader rights.
02
Copyright and DRM
Most ebooks are not owned by the reader. They are licensed under terms set by publishers and platforms. Digital Rights Management technology enforces these restrictions technically, locking books to devices and enabling remote revocation.
03
The digital divide
Reliable internet, compatible devices, and digital literacy remain unequally distributed. COVID-19 exposed this sharply: rural Indonesian households without consistent connectivity could not participate in remote learning on equal terms.

The ethics of access vs ownership

Peter Turner's analysis identifies six ethical dimensions at the heart of the access versus ownership tension. Power differences arise because ownership confers control while access confers only permission. Exclusivity versus equity emerges from the same tension: ownership rights grant legal authority to exclude, and closed commercial platforms often replicate or deepen existing inequalities rather than reducing them.

Privacy concerns arise from the data collection embedded in most digital reading systems. Authentication, behaviour tracking, and personalisation all generate reader data that can be exploited without meaningful consent. Protection concerns relate to the risk of cultural exploitation and misattribution, which are particularly acute for Indonesian literary heritage. Independence is undermined when readers depend on platform owners for continued access to content they believe they have purchased. And rights in digital settings, the least settled of the six, concern the extent to which moral and creative rights persist after a work enters a digital platform.

"What appears to be a thriving digital marketplace is a tightly controlled system in which corporate priorities take precedence over user rights and cultural access."

Avani Singh, sociopolitical analysis

What Amazon's dominance means for Indonesian readers

Amazon's market position is not merely a concern for Western publishers. For Indonesian readers and writers, its dominance shapes what is available, at what price, in what format, and under what terms. The infrastructure of Indonesia's growing digital reading market has largely been built by foreign platforms whose primary accountability is to shareholders, not to Indonesian cultural values or legal norms.

The Amazon-Hachette dispute of 2014, in which Amazon manipulated book availability and shipping times to extract better commercial terms from a major publisher, illustrates what platform dominance looks like in practice. Similar dynamics play out across smaller markets where local publishers have even less leverage. Poddium has a structural opportunity to offer something different: a digital reading environment accountable to Indonesian readers, writers, and cultural institutions.

The quiet politics of PDF and EPUB

The format a book lives in shapes who can read it, on what device, and with what rights. Format choices are infrastructure choices, and infrastructure choices are political choices about who gets included and on what terms.

PDF was released by Adobe in 1993, over thirty years ago. EPUB became the declared industry standard in 2007. Both are old technology by any measure. Yet they remain dominant across virtually every reading platform in the world. Understanding why reveals something important about how institutional inertia and power dynamics operate in publishing.

PDF vs EPUB: a comparison

PDF
The fixed-layout format that refuses to die
PDF

Released by Adobe in 1993, PDF was designed to transform text and documents into a digital form that preserved the visual layout of a printed page. Over thirty years later, it remains one of the most widely used digital formats in the world. Its dominance is partly technical, partly institutional: Google Books alone had scanned over 15 million books as PDFs by 2012, and major repositories including ARSIP in Indonesia, Project Gutenberg, and the Digital Public Library of America all rely on it. Migrating that volume of content to any new format would be prohibitively expensive.

The deeper reason PDF persists is psychological. Readers still think in pages. The rectangular shape of a physical book page remains the dominant mental model for what a book looks like, and PDF preserves that shape on screen. This familiarity has proven more durable than any purely technical argument for newer formats.

Device support

Near-universal, including native Kindle support. Works on virtually any device with a screen.

Accessibility

Limited. Screen reader support is inconsistent. Reflowable text is not supported; fixed layout performs poorly on small screens.

Licensing

Open standard. Free to implement. No royalties or permissions required to create or read PDF files.

Best for

Print-ready documents, reference works with precise layout, archival and institutional collections.

Open standard Near-universal support Poor on small screens Limited accessibility
EPUB
The open standard built for screens and readers
EPUB

EPUB was developed by the International Digital Publishing Forum and declared the industry standard in 2007. Unlike PDF, it was designed specifically for digital reading: its text reflows automatically to fit any screen size, it supports semantic structure for screen readers, and its open standard licensing means anyone can implement it without cost. The latest version, EPUB3, supports multimedia, audio, and enhanced accessibility metadata.

The main exception to EPUB's universality is significant: Amazon's Kindle ecosystem does not support EPUB natively. Kindle users must convert EPUB files to Amazon's proprietary AZW format, introducing friction and dependency. This is not a technical limitation but a deliberate business choice, one that illustrates how proprietary ecosystems can undermine open standards even when those standards are technically superior.

Device support

Near-universal, with the notable exception of native Kindle. Requires conversion for Amazon devices.

Accessibility

Strong. Native support for WCAG guidelines, screen readers, and semantic structure. EPUB3 includes built-in accessibility metadata.

Licensing

Open standard. Free to implement. Actively maintained and updated by the W3C since the IDPF merger.

Best for

Long-form reading, mobile devices, accessibility-first design, and any content that needs to adapt to different screen sizes.

Reflowable text Strong accessibility Open standard Kindle incompatibility

"A just and inclusive reading ecosystem should prioritise open standards, accessibility, cultural representation, and ethical AI integration to ensure equitable participation and empower diverse voices in digital storytelling."

Kagnchanna Meas, format and equity analysis

The fairness argument for open formats

Both PDF and EPUB are open standards, freely licensed, and not owned by any single company. Anyone can implement software to read or create these formats without paying royalties or seeking permission. This is the foundation of the interoperability that makes these formats equitable: a reader in rural Java using a low-cost Android device can access the same EPUB file as a researcher at a well-funded university in Sydney.

Closed formats work differently. Amazon's AZW format ties readers to Amazon's device ecosystem. A book purchased from the Kindle store cannot be transferred to a non-Amazon device without conversion. The reader does not own the file; they hold a license to access it through Amazon's infrastructure. This creates a structural dependency on a foreign corporation for access to Indonesian cultural content, which is precisely the kind of dependency Poddium should be designed to avoid.

AI in publishing: opportunity, risk, and the law

Artificial intelligence is already reshaping publishing at scale. For a platform like Poddium, AI is not a future question. It is a present one, with legal and ethical dimensions that require active decisions now.

2023
Indonesia's AI Ethics Law
Law No. 9 of 2023 requires AI to align with IP protection principles
0
Dedicated AI laws in Indonesia
No AI-specific legislation as of 2025; three existing statutes govern by extension
2045
National AI strategy horizon
Indonesia's 25-year roadmap targets global ethical AI leadership by 2045
3
Statutes currently governing AI
Laws No. 27/2022 (data), No. 8/2011 (electronic info), No. 9/2023 (AI ethics)

Four dimensions

Authorship and copyright
Who owns a story co-written with AI?

Indonesian law, like most legal frameworks, does not yet have a clear answer to AI authorship. Copyright requires human creative contribution as its foundation. But what constitutes sufficient human contribution in a workflow where AI generates substantial portions of a text is genuinely contested, and different jurisdictions are arriving at different answers.

A tiered framework is emerging internationally. Works produced entirely by AI attract no copyright protection under most current frameworks, including Indonesia's Law No. 28 of 2014. AI-assisted works may be protected if sufficient human creative input is demonstrable at each stage. The challenge lies in the middle: how much human input is enough, and how should it be documented? For Poddium, the practical implication is architectural. The platform should be designed so that human creative decisions are logged, timestamped, and attributable at every stage of an AI-assisted writing process.

Current status

No Indonesian case law on AI co-authorship. Law No. 28/2014 centres on human creative contribution without defining thresholds.

International direction

US Copyright Office (2023) denied copyright to purely AI-generated works. EU AI Act establishes a risk-tiered framework for creative outputs.

Risk for Poddium

User-generated works on the platform may lack copyright protection if human creative contribution is not documented at each stage.

Recommended action

Build backend logging for every user input: prompts, edits, structural decisions, rewrites. Create a timestamped chain of human authorship.

No Indonesian case law yet International precedent forming Log human contributions
Privacy and data collection
What is collected, and who controls it?

AI systems in publishing routinely collect user data including reading behaviour, session duration, annotation patterns, inputs, and preferences. This data is often used to train or fine-tune models, sometimes without explicit user consent. Indonesia's Law No. 27 of 2022 on personal data protection establishes a legal baseline requiring transparency, informed consent, and data minimisation. Enforcement remains inconsistent, and many users have limited awareness of what data is being collected and how it is used.

The training data question is separate but related. AI models trained on copyrighted or personal material without consent create legal and ethical liability for the platforms that deploy them. The principle is clear even when the law is not: do not train on content you do not have the right to use. Building deletion workflows into data architecture from the beginning is far easier than retrofitting them under regulatory pressure later.

Governing law

Law No. 27/2022 on personal data protection. Requires transparency, informed consent, data minimisation, and right to erasure.

Training data risk

Models trained on copyrighted or personal material without consent create liability. Default to public domain and openly licensed sources only.

User rights

Right to be forgotten requires deletion of personal data when no longer necessary. Build deletion workflows into architecture from the start.

Recommended action

Implement encryption, anonymisation, and explicit consent mechanisms during development. Publish a clear data policy in plain language before launch.

High liability if ignored Enforcement inconsistent Build privacy-first from day one
Liability and accountability
When AI causes harm, who is responsible?

When AI-generated content causes harm, including defamation, cultural misrepresentation, or copyright infringement, questions of legal responsibility become complex. Current frameworks generally place primary liability on platform operators and deployers rather than on the AI systems themselves, which have no legal personhood. Indonesia's Law No. 9 of 2023 explicitly requires AI applications to align with intellectual property protection principles under Indonesian copyright law.

A joint liability approach is recommended: developers build in safeguards, deployers establish clear user terms, and users are informed of their responsibilities. Deepfakes, disinformation, and non-consensual imagery are areas where Indonesian law is still developing. Poddium's community standards and content moderation policies should address these risks proactively, even where the law has not yet caught up.

Who is liable?

Primarily platform operators and deployers. AI systems have no legal personhood. Criminal liability requires demonstrating human intent or negligence.

Indonesian law

Law No. 9/2023 requires AI to respect IP. No dedicated AI liability law yet. Electronic Information Law (No. 8/2011) applies to harmful content.

Emerging risks

Deepfakes, culturally misrepresentative AI outputs, and AI-generated content that infringes on Indonesian oral traditions or unpublished works.

Recommended action

Publish clear community standards before launch. Document training data provenance. Build human review into the publication workflow for AI-assisted content.

Law developing in real time Joint liability framework Early policy = lower risk
AI for access and inclusion
The highest-leverage application for Poddium

The same AI capabilities that raise risks also offer genuine, significant opportunities for equity in digital reading. Text-to-speech, automated translation, alt-text generation, and real-time captioning can make digital reading accessible to readers with disabilities, non-native readers, and communities whose languages are underrepresented in mainstream publishing, in ways that would be economically impossible to achieve through manual processes alone.

For Indonesia specifically, AI-assisted translation between Indonesian and regional languages including Javanese, Sundanese, Batak, and hundreds of others represents a transformative opportunity. Much of Indonesia's richest literary heritage exists in languages with limited digital presence. AI tools that bridge these linguistic gaps do not merely improve a product; they preserve and democratise cultural access in direct alignment with Poddium's mission. This is Poddium's highest-leverage application of AI: not as a content generator, but as an access infrastructure that extends the reach of human creativity.

Accessibility tools

Text-to-speech, screen reader optimisation, adjustable font sizes, high-contrast modes, and alt-text generation for embedded images.

Language equity

AI translation between Indonesian and regional languages can bridge linguistic gaps that have historically excluded minority-language speakers from digital publishing.

Cultural preservation

AI transcription and metadata generation can make oral traditions, manuscript collections, and audio archives searchable and accessible for the first time.

Recommended action

Prioritise AI-powered accessibility features in the first product release. Develop a roadmap for AI translation into at least three major Indonesian regional languages.

High impact, high alignment Technically achievable now Directly serves Poddium's mission

Indonesia's regulatory trajectory

Indonesia currently has no dedicated AI law. The existing framework relies on three statutes: Law No. 27 of 2022 (personal data protection), Law No. 8 of 2011 (electronic information and transactions), and Law No. 9 of 2023 (AI ethics guidelines). These provide a foundation but were not designed with generative AI in mind and leave significant questions unanswered.

Indonesia's national AI strategy, introduced in 2020 and covering a 25-year horizon through 2045, signals the government's intention to become a global leader in ethical AI governance. In 2025, the Indonesian government announced plans to establish a formal legal framework for AI use by companies and platforms, with both civil and criminal penalties for non-compliance. Poddium should build compliance infrastructure now rather than waiting for regulation to arrive.

Five principles for building a fairer digital bookshelf

The research team distilled their comparative analysis into five operational principles. These are not abstract policy proposals. They are concrete commitments that a platform like Poddium can and should act on now, before scale makes them harder to implement.

1
Default to open formats
Prioritise EPUB and PDF over proprietary formats at every level of the platform. Open formats resist lock-in, support established accessibility standards, and are compatible with the widest possible range of devices, including low-cost Android phones more likely to be used by Indonesian readers outside major cities. Closed formats like AZW create structural dependency on corporate ecosystems that can be withdrawn, repriced, or restricted at any time without recourse. The format choice is not a technical detail; it is a statement about whose interests the platform serves.
Practical step: Build all Poddium content output in EPUB by default. Offer PDF as a print-ready alternative. Commit in platform policy never to build proprietary formats that require a Poddium app to read.
2
Build accessibility into the core product architecture
Screen reader compatibility, adjustable font sizes, high-contrast display modes, text-to-speech, and offline reading capability are not edge-case features. For a significant portion of Indonesian readers, including those with low literacy, visual impairments, limited data connectivity, or interrupted educational backgrounds, these features are the difference between meaningful access and exclusion. Universal design is not charity; it is architecture. A product built accessibly from the beginning costs far less than one retrofitted for accessibility under pressure.
Practical step: Audit every user-facing interface against WCAG 2.1 AA standards before public launch. Use EPUB3's native accessibility metadata structures. Test with real users who have different accessibility needs, not just automated tools.
3
Treat AI as an instrument, not an author
For AI-assisted content to attract copyright protection under Indonesian law, human creative contribution must be both real and demonstrable. Design the platform so that users make substantive creative choices at every stage of the writing process, not simply pressing a generate button and accepting the output. Every prompt written, every draft edited, every narrative decision made is an act of authorship. The platform should make those acts visible, logged, and attributable, both to protect users' creative ownership and to protect the platform's legal standing.
Practical step: Build backend logging for every user input during AI-assisted creation: prompts, edits, selections, structural decisions, rewrites. Timestamp and store this record as a demonstrable chain of human authorship.
4
Establish a clear licensing and attribution policy before users need it
If Poddium allows users to publish AI-assisted works, clear, user-facing licensing infrastructure must be in place before the first work is published. The India model is instructive: institutions like Shodhganga default to CC BY-NC 4.0, which allows non-commercial use and AI processing while retaining attribution protections for creators. A similar default, combined with upgrade options for users who need commercial rights, creates a workable foundation. The absence of a clear policy does not create a neutral situation; it creates legal ambiguity that ultimately favours platforms over creators.
Practical step: Draft and publish a licensing policy before public launch. Write it in plain language first. Default to CC BY for standard outputs. Review and update as Indonesian AI law develops.
5
Design for Indonesia's cultural complexity, not against it
Indonesia has hundreds of distinct literary traditions, oral histories, musical forms, and artistic practices. Much of the most significant material from these traditions is not housed in Indonesia at all: the National Library of Australia holds over 200,000 Indonesian monographs, Ohio University maintains 193 hours of historical Indonesian audio recordings, and colonial-era archives across Europe contain materials with no digital Indonesian equivalent. A platform that claims to serve Indonesian storytellers must actively work to surface, preserve, and celebrate this heritage, rather than importing Western publishing frameworks and applying them uncritically.
Practical step: Initiate conversations with ARSIP and the National Library of Indonesia about identifying public domain Indonesian works for digitisation and inclusion in Poddium's content ecosystem. Prioritise works in regional languages. Develop a roadmap for AI-assisted translation into at least three major regional languages.

Four students, one semester, one shared purpose

This research was produced by four undergraduate students in the ARTS3000 program at Macquarie University, a unit designed to connect Arts students with partner organisations and give them the opportunity to contribute meaningfully to real-world research, outreach, and advocacy. All four students were involved in the project's conception, analysis, and writing. Each took primary responsibility for one strand of the research.

PT
Peter Turner
Law & Arts, Social Justice Major
Led the ethical analysis of access versus ownership, developing a six-dimension framework covering power differences, exclusivity, privacy, protection, independence, and digital rights as they apply to Poddium's operating context.
AS
Avani Singh
Sociopolitical Analysis
Examined the corporate, legal, and structural forces shaping access to digital books globally, with a focus on Amazon's market dominance, the implications of DRM for libraries, and the digital divide's effects on the Global South.
SC
Shun Cheng
Format and Standards Research
Traced the full evolution of digital book formats from the early 1990s to the present, analysing the technical, economic, and institutional forces that have kept PDF and EPUB dominant and what this means for equitable digital reading.
KM
Kagnchanna Meas
Fairness and Equity Analysis
Assessed open versus closed ebook formats through the lens of fairness, equity, and control, and developed a detailed framework for what a just and inclusive reading ecosystem should look like in practice.

The team extends its sincere thanks to Amin Rois Sinung Nugroho and Octa Nurhasanah at Poddium for their partnership, guidance, and for the genuinely fascinating brief that made this research both intellectually rich and practically grounded.