Escape from Adobe and Microsoft: A developers book production assembly line revolution

When D.J. Speckhals announced on his personal blog that he was using Git to manage the entire book's publishing process, the comment area exploded. Some people said he was crazy, others said he finally got it right. It wasn't a gimmick-he moved everything from typesetting to collaboration into the command line, used Git's branching model to manage revised versions, stored content in plain text format, and eventually bypassed Adobe InDesign and Microsoft Word to output a professional-level publication. The reason why this matter is worth talking about is not because the technology is so complex, but because it has torn open an industry wound that has existed for decades: Why do people who write books use poster software? Why does the production process of a technical book rely on a proprietary format that was born in 1987?

This article is not a tutorial, but an in-depth disassembly. I will start from the technical principles, to the impact of the industry, to the practical feasibility, to those pitfalls that cannot be avoided, and finally give you a clear judgment: whether this method is worth your time.

1. Technical background: Old scars in the publishing industry

The degree of digitalization in the publishing industry has long been in a state of distortion. The author writes in Word, the editor uses Track Changes to revise, the typesetter uses InDesign to re-type, and the design company uses PDF to deliver clear samples. In this process, every link is converting formats, and information may be lost every conversion. Word's.docx format is essentially a ZIP archive stuffed with XML files;InDesign's.indd format is another closed ecosystem. There is no real interoperability between the two.

Even worse is version control. Word has a "version history" feature, but it is designed for single users, single files. When a three-person editorial team works on multiple chapters of a book simultaneously, version confusion is almost inevitable. Some people use Google Docs to collaborate, but the typesetting capabilities of Google Docs are far from enough for official publication. Some people use LaTeX, which is common in academic circles, but its learning curve can directly dissuade authors with non-technical backgrounds.

D.J. Speckhals 'plan started from this pain point. In a May 2026 blog post, he detailed how he used Git to manage the complete production process of a technical book: content is stored in plain text (Markdown or similar format), Git is responsible for version tracking and collaborative management, Pandoc handles format conversions, GitHub Actions automates the build process, and the output can be PDF, EPUB or HTML. The core logic of this solution is "plain text first"-all content is a human-readable text file, format information is embedded through a lightweight markup language, and version control is taken over by Git.

This idea is not new. The concept of "Docs as Code" has long been popular in the field of technical documentation. GitBook, MkDocs, and Docusaurus are all products of this idea. But Speckhals used it in commercial book publishing and bypassed the giants of Adobe and Microsoft, which is interesting.

2. Analysis of core technology principles

To understand this solution, we must first understand three core components: plain text storage, Git version control, and automated build process. Together, they form a complete book production line.

The core of plain text storage is Markdown or other lightweight markup language. Needless to say, the benefits of Markdown are simple: the syntax is simple, the threshold is extremely low, and it can be opened in any text editor. But Markdown itself is too crude for complex publishing needs. So in practice, some extended versions are often used, such as MDX (which supports embedding JSX components), some extensions of CommonMark, or directly using AsciiDoc-the latter's table support, cross-reference, and index generation capabilities are much better than standard Markdown. According to public sources, many technical publishers (including some of O'Reilly's projects) have begun to try using AsciiDoc as an intermediate format.

Git version control solves collaboration and historical tracking issues. In the traditional process,"which version is this" is a vague concept. Version numbers are maintained manually, and versions are distinguished by dates and file names. Git completely changed this issue. Each commit is an accurate timestamp snapshot, each branch can represent a different version line (first draft, revision, final version), and a merge can clearly show who changed what at what time. The Pull Request mechanism is naturally suited to the editorial review process: authors submit changes, editors pass PR review, and merge after confirmation. This mechanism has been used in software development for twenty years, and it can be borrowed from publishing.

The automated build process is the engine that keeps the entire assembly line running. In Speckhals 'plan, GitHub Actions assumed this role. When an author pushes code to the warehouse, the CI/CD process automatically triggers: Pandoc reads the Markdown source file, applies a style sheet, and generates a PDF or EPUB. Build results can be automatically published to a preview server for online viewing by editors and reviewers. The essence of this process lies in the opposite of "what you see is what you get"-"the source document is the only truth." The format is not hidden in proprietary software, but exists in the form of code that anyone can view, modify, and copy.

The key technical points are as follows:

• Pandoc: A document conversion tool written by Haskell that supports mutual conversion between dozens of formats such as Markdown, LaTeX, HTML, Word, and PDF (according to Pandoc's official documents)
• Git LFS (Large File Storage): Handle picture resources in books and store large files on the Git LFS server. Only pointers are retained in the warehouse (according to GitHub official documentation)
• LaTeX template: Used to generate high-quality PDF output, supporting complex layouts, catalogs, indexes, and cross-references (according to TUG Association data)
• GitHub Actions: Automate CI/CD and support automatic triggering of the build process every Push or PR (according to GitHub official documents)

The entire technology stack is open source, self-hosting, and does not rely on any single vendor. This point will appear repeatedly in subsequent comparisons.

3. Why is this important?

The importance of this matter lies not in the technological breakthrough, but in the fact that it challenges a deep-rooted industry assumption that book production must rely on a proprietary chain of tools.

Adobe InDesign is an industry standard that has monopolized the commercial publishing field for many years. The fee for an InDesign subscription is US$54.99 per month (approximately RMB 400), and the annual fee is US$35.99 per month (according to Adobe's official website pricing). This doesn't include font licensing, plug-ins, and training costs. For individual authors or small teams, this is not a small expense. More importantly, InDesign's learning curve is extremely steep. According to industry estimates, it takes at least 40-60 hours of systematic study for a person without a typesetting background to reach the level of "being able to complete the typesetting of a book independently."

Microsoft Word is doing a little better, but not much better. Word's "revision" feature is designed for office documents, not long text in books. When a book has 20 chapters, 300 pages, and 50 pictures, Word's response speed drops significantly, and format compatibility issues increase exponentially. Many people have the experience of sending a Word document to a typesetting company, but after the other party opens it, the format is completely chaotic and requires a lot of manual repairs.

Speckhals 'plan completely solves this problem. Git is free, Pandoc is free, GitHub's private repository is free for individual users, and Actions has a free quota of 2000 minutes per month. For most book projects, these free resources are sufficient. The cost has changed from a few hundred dollars per year to zero, and the threshold for learning has changed from "professional typesetting software" to "command-line basic operations."

But the more far-reaching impact lies in changes in collaboration models. The traditional publishing process is a linear waterfall: the author writes and hands it to the editor, the editor and proofreading it to the typesetting, and the typesetting is done to the proofreading. Every link has a waiting time, and every link has information loss due to format conversion. Git workflows make it all parallel, asynchronous, and traceable. Editors can directly point out the punctuation problem in page 47, paragraph 3, and the second sentence in PR. The author can immediately modify and push it, and the entire discussion history is saved in Git records. This is not an efficiency improvement, but a fundamental change in the collaboration paradigm.

4. Industry impact and data support

The speed of technological change in the publishing industry has long lagged behind that of the software industry by at least ten years. But there are several data worthy of attention that suggest that the gap is narrowing.

The e-book market share continues to grow. According to the 2025 annual report of the American Publishers Association (AAP), e-book revenue accounts for approximately 25% of the overall book market revenue, an increase of approximately 5 percentage points from five years ago. E-books do not require printing or logistics, but they require digital format. This has created a natural market demand for the "pure digital first" publishing process.

The scale of self-publishing of technical books has expanded. According to industry estimates, the annual growth rate of the number of self-published technical books on platforms such as Lepub and Amazon KDP exceeds 15%. The user profiles of these platforms overlap highly with the developer community-they are familiar with Git, command line, and open source tools. For this group, writing books with InDesign is counter-intuitive.

The ecological maturity of open source document tools has increased. According to the GitHub Octoverse report (2025), the multi-year growth rate of star indicators for open source document tools such as Docusaurus, GitBook, MkDocs, and VuePress remains above 20%. The technical stacks of these tools highly overlap with Speckhals 'solution: Markdown source files + Git version control + automated build. It is a natural extension of the user community from technical documentation to book publishing.

Proprietary software subscription costs continue to rise. Adobe Creative Cloud family barrel subscription prices have increased by approximately 30% in the past five years (according to historical pricing records on Adobe's official website). Microsoft 365 Personal Edition subscriptions will increase from $69 per year in 2019 to $139 per year in 2025 (according to Microsoft's official website). This trend of price increases has pushed users 'psychological thresholds for looking for alternatives to lower and lower.

Together, these data point to one conclusion: the window for technological transformation in the publishing industry is opening. It is not because a revolutionary technology has emerged, but because three factors: the maturity of open source tools, changes in user habits, and cost pressures have finally come together.

5. Actual implementation cases

It is not enough to just talk about the principles. We need to see how real people use this method.

Case 1: D.J. Speckhals 'own technical book project

Speckhals blog detailed how he used this method to complete the production of a technical book. He chose AsciiDoc as the source file format (more suitable for complex publishing needs than standard Markdown), Git manages versioning and collaboration, GitHub Actions handles automated builds, and LaTeX templates generate PDF output.

The specific operating process is as follows: He uses VS Code and the AsciiDoc plug-in to edit the source file, and uses Git to manage the progress of each chapter during the writing process. Every time he completes the first draft of a chapter, he creates a PR, and the editor reviews and comments it in the PR. After confirmation, merge it into the main branch. CI automatically triggers the build, generates a preview PDF and uploads it to the preview site hosted by Netlify. Throughout the process, Speckhals does not need to open any Adobe or Microsoft software, install any proprietary fonts, and all tool chains are open source.

What about the final output quality? He showed off the generated PDF samples with professional layout, complete catalog, click-to-jump index, and normal code highlighting. According to his blog, the output quality is "at a level that can be submitted directly to the printing house."

Case 2: Personal practice of a technical blogger

In the domestic technology community, some developers are also trying to use similar methods to manage their own e-book projects. An author who has long maintained a technical blog on GitHub (anonymous here for privacy reasons) revealed in an interview in 2024 that he used a combination of Git + Markdown + Hugo (a static website generator) to complete the content management and online publishing of three technical e-books.

His workflow goes like this: Each article is stored in a Git repository in Markdown format, and metadata (title, date, label) is managed using Hugo's front matter feature. GitHub Actions is configured with automated build, which automatically generates a static HTML website after each push and synchronizes it to GitHub Pages. His readers can read it directly on the website, or they can download PDF or EPUB (configured through Hugo's output format).

"What surprised me the most was version control. "He said in an interview," I used to write a series of articles and revise a certain concept and found that there were inconsistencies in the previous quotes. I had to manually search the full text. Now I directly git diff and locate all relevant locations in two seconds. "

Although his e-book has not been officially published (it does not involve ISBN and printing), the online version has received a total of more than 500,000 views. According to industry estimates, if technical e-books of similar size go through traditional publishing channels, the royalty income will be approximately RMB 50,000 - 150,000.

What these two cases have in common is that the users have technical backgrounds and basic Git operation capabilities, and do not need to study typography specifically. They used tools they were familiar with to complete publishing level content production.

6. Comparison with competing products/alternatives

The mainstream book production solutions on the market can be roughly divided into four categories: traditional proprietary software, online collaboration platforms, open source document tools, and pure code solutions. They each have their own advantages and disadvantages and are applicable in different scenarios.

It can be seen from the table that these four types of solutions cover the spectrum from "zero technical threshold" to "completely technically controllable". Traditional proprietary software still has advantages in output quality and design freedom, but cost and learning threshold are disadvantages. Online collaboration platforms are suitable for rapid prototyping and team collaboration, but not suitable for formal publication. Tools such as GitBook are a compromise solution that is suitable for technical documents but not for books that require fine typography.

Speckhals 'plan belongs to the far right column. Its core competitiveness is "completely controllable" and "zero cost," at the price of technical threshold. If you can accept this price, it provides flexibility and efficiency unmatched by other options.

My judgment is that for individual authors and small teams with technical backgrounds, the Git solution is the optimal solution in terms of cost and efficiency. For authors with non-technical backgrounds, GitBook or Notion is a more realistic choice from a practical perspective. If you work for a large publishing house and have full-time typesetting staff, you can continue to use InDesign-after all, tools are for the purpose.

7. Technical challenges and limitations

Having said so many advantages, we must talk about the limitations of this plan. There is no point in over-praising. Only by facing problems honestly can we make correct judgments.

Technical thresholds are the biggest obstacle. Git command line, Markdown syntax, YAML configuration, CI/CD process-these things are basic knowledge for developers, but new areas for most authors. According to industry estimates, it takes at least 20-30 hours of systematic learning for a person with no technical background to master this workflow. This is several orders of magnitude higher than the threshold for opening Word to write.

Complex typography requirements are difficult to achieve. Markdown and AsciiDoc both have limitations when dealing with complex layouts. For example, magazine-style multi-column layouts, complex mixed graphics and text, and artistic typography-these things can be done in a few minutes in InDesign, but in plain text solutions, they require a lot of CSS or LaTeX customization. Pandoc's document conversion is not a panacea. Style loss or misplacement may occur during some format conversion.

The collaboration experience is not as good as dedicated tools. Git's collaboration model is optimized for code design and is not perfect for text editing. Editors are accustomed to using Word's "annotation" function to mark directly next to the original text, but Git's PR comments are displayed independently of the original text and need to be switched back and forth. Although VS Code plug-ins like GitLens can improve the experience, there is still a gap between the overall collaboration experience and professional editing software.

Image and resource management requires additional configuration. Pictures in books require manual path management, naming, and compression optimization. In InDesign, pictures are directly embedded into projects and can be dragged and dropped. In the Git solution, you need to consider the configuration of Git LFS, the selection of image formats, and the adaptation of responsive displays-these are all additional cognitive burdens.

Chinese typesetting support still needs to be improved. Pandoc and LaTeX's support for Chinese has greatly improved in the past few years, but there is still a gap compared with some specialized tools. Scenes such as Chinese punctuation compression, vertical text, and minority text support may require more manual debugging.

These challenges are not insurmountable, but they require clear expectations before investing. This plan is suitable for authors who are willing to spend time learning and accept a certain technical cost. If you want to decide to write a book today, start it tomorrow, and hand it in the day after tomorrow, this plan is not suitable for you.

8. Who should pay attention to this matter

People of different identities can get different things from this incident.

Independent technical authors are the largest beneficiary group. If you are writing or planning to write a technical book, this solution directly reduces the cost of your production tools. Adobe subscriptions cost 400 yuan a month, which cost several thousand yuan over the years, while Git solutions cost almost zero. More importantly, version control and automated building can significantly improve your writing efficiency.

Technology bloggers and content creators can also benefit. If you are maintaining a technical blog, Git workflows allow you to manage both blog content and derivative e-book content. One writing, two outputs-this is a very practical efficiency improvement.

Small technical publishers and publishing studios should carefully evaluate this plan. According to public information, some publishing practitioners have tried the "document is code" workflow internally, using Git to manage manuscripts and automated tools to process typesetting. If you are dealing with technical book publishing in a traditional way, assess the feasibility of Git solutions and you may find significant room for cost and efficiency optimization.

Ordinary authors and writers can wait and see, but do not have to rush to follow up. Your main need is to "write good content," and tool selection should serve this goal. If Word already meets your needs, you don't have to switch workflows to keep up. But if you encounter pain points in collaboration, version management, and format conversion, you can consider starting with simple Markdown+Git and migrating gradually.

Software developers may think this is obvious common sense. Git workflows are daily for you, and using it to write books is just a scenario expansion. If your team is working on technical documentation or an internal knowledge base, this solution can be used directly.

9. Prediction of future trends

Based on current observations, I have several clear judgments, which are not nonsense like "maybe","maybe" and "maybe".

text-only publishing solutions will gradually enter the mainstream vision. Not because everyone is turning to it, but because it will become an option that will be seriously considered. The popularity of GitHub has lowered the cognitive threshold of Git, the maturity of tools such as Pandoc has improved usability, and the practices of more pioneers will form reference cases. It's not a question of whether it will happen, but a question of how quickly it will happen.

AI-assisted writing will be deeply integrated with Git workflows. Big language models can already assist writing, polishing, and proofreading. When these capabilities are combined with Git's version control and PR review processes, a new production paradigm will be formed. Speckhals 'plan is purely manual, and an AI-assisted version is likely to appear in the future-the author writes the first draft, the AI generates revision suggestions, and the editor reviews and confirms them in PR. The efficiency improvement of this combination will be orders of magnitude.

there will be more consolidation in the publishing tools market. The existing publishing tool chain is fragmented: writing tools, version control, format conversion, and hosting platforms are independent. In the future, a one-stop platform may emerge to integrate these links and shield users from the underlying complexity. GitBook is already doing similar things, but its customization capabilities are limited. A "document-as-publish" platform like Notion may emerge, allowing users to focus on content and leave technical details to the platform.

Fourth, paper book publishing will not be completely replaced. Pure digital publishing solutions solve the problem of screen reading, but the reading experience, collectible value, and offline readability of paper books are irreplaceable. Git workflows can output PDFs for printing, but the final printing process still requires professional services. The future trend is not "digital replacing paper", but "multiple output formats for digital workflow services."

X. Summary and action recommendations

The core message is simple: Book production can be done without Adobe and Microsoft, and Git workflow is a real and viable alternative. It is not everything. It has technical thresholds and application boundaries, but its advantages in cost control, version management, and collaboration efficiency are real.

If you decide to try, my advice is to start with the smallest feasible step: pick a project you are writing (even if it's just a long article), migrate the content from Word to Markdown, open it with VS Code, initialize the repository with git init, and make a commit. Then evaluate whether the experience meets your needs and decide whether it is in-depth.

Don't start a full set of automated CI/CD at the beginning, and don't configure complex LaTeX templates at the beginning. First experience the experience of Git managing text, and then gradually add complexity. Tools serve people, not the other way around.

For developers already using Git, this solution is almost a zero-cost migration. For authors with non-technical backgrounds, they can start with local editors such as Typora and Obsidian that support Markdown, accumulate familiarity with Markdown, and then gradually introduce Git version control.

The digital transformation of the publishing industry is taking place, just ten years later than the software industry. The purpose of this article is to let you know that this is happening and what it means. The next choice is your own.