This is the story of how this published book, Issues in Contemporary Constitutional Law, came to be. It is a story of an intensive collaboration among students, me, and AI. We start with a conventional law school writing seminar, one in Advanced Constitutional Law. Instead of engaging in the losing battle of prohibiting students from using AI, I embrace the inevitable. I tell my 11 students that they must use AI, at least in topic selection and development, and that they are strongly encouraged to collaborate with it in researching and writing their papers. I guide them to CustomGPTs that I have created to ruthlessly examine proposed topics, use AI to supplement my own human edits of their interim submissions, have them use AI to help develop interim presentations of their work, and use AI in conversion of final AI-assisted drafts from their weapon of habit, Microsoft Word, to LaTeX, a medium favored in the sciences for serious book production. I tell them throughout that if all goes well and they consent, I will convert their efforts into a book of which they can be proud and which they can promote to prospective employers. In short, I turn the conventional often-forgettable law school seminar into something in which the students and I can take permanent pride. And thus there is now a Kindle edition with paperback and hardback edition coming imminently or actually here at the time you read this blog entry.

You will have to read the book to judge the results. But I believe the works in the anthology range from very good to superb. Basically, AI enabled the strongest and most enthusiastic students to produce works which, while still very much their own, were extraordinary. They were better written, better researched and – above all – more robustly reasoned than work I have hitherto encountered in three decades even from top students at a strong but non-elite law school. And it enabled students of perhaps lesser ability, enthusiasm or time to produce work considerably better than papers generally produced in a semester-long seminar. I could thus focus my attention on the strength of argumentation and the creativity of ideas rather than the usual slog through run-on sentences, dangling modifiers and other symptoms of the failure of our educational system to invest in human-generated, high-quality writing. True, there were a couple of papers that were, in all candor, not particularly strong, but they were also those where it appeared that the student had either made less use of AI or, in one case, had detectably delegated far too much to AI. Those students self-selected not to publish, along with two others who feared either that their paper was not good enough (not true, actually) or that they could be retaliated against for expressing views that are unpopular among many in power today. But what you see is not some cherry-picked prize winners from an entire law school's 2L or 3L class. Nor are they papers where the instructor had suddenly, independent of AI, become a master of paper supervision. They are, I believe, the result of diverse and talented people being able to achieve something extraordinary with help from a non-human.

AI had three more roles to play. First, it helped me provide better feedback to the students. Yes, I did think independently about each of the submissions. I used whatever insight and wisdom I have accumulated over the years. But, like many professors, I have blind spots. I don't consider every angle. AI helped me not only focus on issues of organization that the papers displayed but also on substantive matters I might have missed. To be sure, AI, in my opinion, sometimes got it wrong. We worked iteratively and collaboratively. But, just as I do when my electric car goes into "self-driving mode," I kept my hands on the steering wheel and ultimately overrode its judgments on occasion. On balance, however, my own critiques of students' interim work was unquestionably enhanced by using AI along the way.

Second, my contribution, in addition to superintending the seminar, vigorously questioning the students as they presented interim findings and conclusions, and doing the usual guiding and editing, was to write a Foreword to the book. I excerpt it below. Yes, a lot of the ideas in that Foreword are my own. AI was not sufficiently introspective or mercenary to wonder whether this sort of work would receive much credit at merit-raise time. AI did not volunteer cynicism about ABA accreditation standards or faculty incentive structures. But along the way, in this as in other works, AI was my research and writing assistant. It was a wall against which to bounce ideas. It helped me develop concise summaries of the student authors' work. Again, buy the book. You can then judge whether my self-appraisal is accurate. Speaking of "buy the book," there is a reason to do so beyond actually learning about constitutional law or assessing whether I am exaggerating. I am donating the royalties each year to a prize for a student who, the following year, produces the best work in my seminar or in a human-AI collaboration.

AI's third role was in book production. It turns out that I had underestimated the complexity of converting Microsoft Word files into LaTeX that not only compiled in the usual way but was consistent among the various student works and satisfied Amazon's own exactions with respect to the margin prisons, uniqueness of sectional headings, and other matters that came as a rude awakening. I had asked the students to make an effort in this regard. And most did so. But I gave them insufficient instruction to enable them to excel. But here too AI was invaluable. It guided me through rectifying all the compile-time errors the various files generated. It interpreted all the error messages coming from the Kindle Previewer and the Amazon publishing interface. It was the usual AI performance. Certainly it did way better than I would have done on my own. On the other hand, it sometimes got confused, lost the thread, or downright erred. Anyone who thinks that superhuman intelligent AI is around the corner, how very wrong you are. One of the cool things about AI is that it helps you appreciate how amazingly smart we humans actually are in certain ways. (We have incredibly large context windows, for one thing.) Still, we plugged away, endured occasional spasms of verbal abuse that I directed at various large language models, and ultimately made the Amazon Deities happy. It even helped create a cover. A fringe benefit: AI has now written up a guide for future students and me on what we all need to do to make future productions of this nature a less painful experience. I present it below in an appendix. Feel free to make use of it should you decide to further this experiment in legal pedagogy.

I hear it now. "Great, Professor Chandler. Sounds like you did a good thing. Thanks for virtue signalling on your blog. But can this be scaled? Should it be scaled?"

My answer to the first question is a definite yes. Certainly it is not difficult to tell students to use AI to help them write their papers. Surprise – many of them are doing so to various degrees without your encouragement. (What do you think Grammarly is? What do you think Lexis and Westlaw are these days? Heck, what do you think Google is in 2026? Drawing the line between "good AI" and "brain-rotting AI" is not so easy.). And it is not difficult for you to feed student work at various stages to AIs either custom trained to constructive review such work or even to the best modern AI models. Claude Opus, ChatGPT and Gemini 3 Pro are pretty damn smart. And it would likely take most faculty less than an hour to develop a prompt that captures their own values in student work. Maybe it is diversity of sources, maybe it is consideration of opposing arguments, maybe it is inclusion of a particular perspective or a particular collection of legal materials. Without much difficulty you can draft something that reflects your views. And, with classic meta-prompting, if you can't quite articulate what you want, you can ask AI for help. Ask it for three options. Pick the best and keep iterating. Moreover, you don't have to have a one-size-fits-all prompt. You can read a draft and tell the AI to write 300 words explaining to the student how their proposal sounds an awful lot like the majority's approach to the due process clause in Lochner v. New York. You can ask the AI to explain to the student why they needed to consider whether there were meaningful distinctions or why, perhaps, there was something to be said for a revival of Lochnerism. Yes, you could probably write that yourself. But might such an effort be more of a cryptic marginal scrawl, "Lochner?" or "Isn't this rebadged Lochnerism" rather than the more thorough and likely more pedagogic approach taken by AI. By the way, you should tell all of this to your students. Many will play amateur professor before you ever see the draft. The results of AI-assisted self-assessment are likely to be very positive on balance.

So, yes, this can be scaled. Law faculty can do this. Law students can do this. And, with the exception of the Word to LaTeX matter, which might not have been necessary, it just isn't that hard. But that leaves the question of whether it should it be scaled. Did I do a good thing or a bad thing by not only letting the students use AI but also by forcing them to do so. Maybe I promoted my students (and perhaps myself) by letting AI infiltrate the core of a writing seminar and then producing a self-published book. But what happens when they can't use AI? Did I just undermine the very purpose of the writing requirement that forms a tradition at most law schools and, in one form or another, is required by the American Bar Association for law school accreditation? I discuss this more in the Foreword, but let me sketch some points here. (The short answer is "Yes, but cautiously.")

What we should scale is an “augmented seminar” model in which AI is treated as a tool for stress-testing ideas early, widening the research funnel, and generating structured critique—while the student remains responsible for verification, synthesis, and defense of every claim worth keeping. That last clause is not window dressing. At least until those distant days of the 2030s when we all have AI implants, if a student cannot explain, orally and under questioning, why a proposition is true, why the leading counterarguments fail, and what a cited case actually holds, the seminar has failed—no matter how polished the prose looks. If, however, the course design forces students to interrogate the model’s output, purge slop, verify primary sources, and repeatedly restate their argument in their own voice under pressure, the presence of AI can push them toward higher-order legal thinking rather than away from it. I am pleased that almost all of my students did quite well in this regard.

Let me conclude by quoting from the Foreword:

I did worry about making it too easy for students to delegate their scholarly efforts to AI. In part for the edification of the class but mostly to force students to internalize their own papers, I compelled numerous oral presentations of their work. At present, AI can't readily fake that. And I could see that either because they were responsible students or because they feared detection, the students did internalize their own work. They had read and thought about what they had written. What I saw was prior wrestling with AI resulting in drafts and final papers that had a higher degree of sophistication and fewer needless writing distractions than those in prior seminars produced by students admitted and trained in roughly equivalent ways,

The papers in this volume are evidence of the coming shift in legal scholarship. They are ambitious. They tackle complex, intersectional areas of constitutional law. They do not shy away from the hardest questions. Moreover, these students learned a skill that will be indispensable in their future practice: verification. In a world where text is cheap and easily generated, the premium on truth has never been higher. My students learned that if they cited a case suggested by the AI, they had to read the case. If the AI claimed a statute existed, they had to find the code section. This habit of radical skepticism is perhaps the most valuable lesson of the semester.

We are living through a Gutenberg moment. The cost of generating intelligent-sounding text has dropped to near zero. This changes the economics of legal practice, and it must change the pedagogy of legal education. We can no longer test students solely on their ability to summarize cases; we must test them on their ability to solve problems, to construct novel arguments, and to manage the vast intelligence amplification that these tools provide.

Guide: Producing a Multi-Author LaTeX Book for Kindle

This guide documents the key LaTeX configuration choices needed to produce a multi-author essay collection (such as student papers) for Kindle publishing. It is based on a working preamble and chapter structure that successfully passed Kindle Previewer validation and produced a clean KPF file.

Document Class Configuration

The openany option is critical for e-readers. It prevents LaTeX from inserting blank pages before chapters (which it normally does to ensure chapters start on right-hand pages in print). On Kindle, these blank pages create jarring interruptions. The 12pt base font size ensures readable text across various screen sizes and device settings.

\documentclass[openany, 12pt]{book}

Geometry

Use uniform margins rather than asymmetric print layouts with gutter space. Kindle reflows content dynamically based on user font size preferences, so elaborate margin schemes provide no benefit and can cause unexpected behavior.

\usepackage[margin=1in]{geometry}

Heading Customization

A distinctive feature of this setup is the avoidance of bold headings throughout. All heading levels use italics instead, which can improve rendering consistency on e-ink displays.

\usepackage{titlesec}

% Chapter (Roman numerals)
\titleformat{\chapter}[display]
  {\normalfont\huge\itshape}
  {\chaptertitlename\ \thechapter}{20pt}{\Huge}

% Section (A, B, C)
\titleformat{\section}
  {\normalfont\Large\itshape}
  {\thesection.}
  {1em}
  {}

% Subsection (1, 2, 3)
\titleformat{\subsection}
  {\normalfont\large\itshape}
  {\thesubsection.}
  {1em}
  {}

% Subsubsection (a, b, c)
\titleformat{\subsubsection}
  {\normalfont\normalsize\itshape}
  {\thesubsubsection.}
  {1em}
  {}

% Paragraph (i, ii, iii)
\titleformat{\paragraph}[display]
  {\normalfont\normalsize\itshape}
  {\theparagraph.}
  {0.5em}
  {}
\titlespacing*{\paragraph}{0pt}{3.25ex plus 1ex minus .2ex}{1em}

The \normalfont directive ensures no bold creeps in from default settings. Each level steps down in size while maintaining the italic style for visual hierarchy without weight variations.

Legal/Academic Numbering Hierarchy

Legal writing typically uses the I, A, 1, a, i outline format rather than standard academic numbering. This requires explicit redefinition of counters at each level:

\setcounter{secnumdepth}{5}

\renewcommand{\thechapter}{\Roman{chapter}}      % I, II, III
\renewcommand{\thesection}{\Alph{section}}       % A, B, C
\renewcommand{\thesubsection}{\arabic{subsection}} % 1, 2, 3
\renewcommand{\thesubsubsection}{\alph{subsubsection}} % a, b, c
\renewcommand{\theparagraph}{\roman{paragraph}}  % i, ii, iii

Setting secnumdepth to 5 ensures that even the paragraph level receives numbering, which is necessary for deep legal outlines.

Table of Contents Adjustments

Roman numerals take more horizontal space than Arabic numerals. Without adjustment, chapter numbers like "VIII" will collide with chapter titles in the table of contents.

\usepackage{tocloft}
\setlength{\cftchapnumwidth}{4em}
\setlength{\cftsecnumwidth}{3em}

These width settings provide adequate space for wide Roman numerals and lettered sections.

Chapter Author Attribution

For multi-author collections, each chapter needs author attribution. This custom command places the author name in italics beneath the chapter title and adds it to the table of contents as an indented entry without a section number:

\newcommand{\chapterauthor}[1]{%
  \vspace*{-10pt}
  {\large\itshape #1}
  \par\vspace{20pt}
  \addcontentsline{toc}{section}{\protect\numberline{} \itshape #1}%
}

Usage in chapter files:

\chapter[Short Title]{Full Chapter Title}
\chapterauthor{By Student Name}

The empty \numberline{} creates the indentation without printing a number, so the author name appears subordinate to the chapter title in the TOC.

Hyperlink Configuration

Kindle requires functional hyperlinks for navigation, but default blue link colors look garish on e-ink displays. Using near-black links maintains functionality while appearing professional:

\definecolor{almostblack}{RGB}{25,25,25}
\usepackage[colorlinks=true, linkcolor=almostblack]{hyperref}

The colorlinks=true option is essential. Without it, hyperref draws boxes around links, which render poorly on Kindle devices. Colored text handles the indication of clickability instead.

Paragraph Spacing

Adding inter-paragraph spacing improves readability on screens where traditional indentation may be less visually distinct:

\setlength{\parskip}{0.5\baselineskip plus 2pt}

The plus 2pt provides flexibility for page breaking.

URL Handling

Legal citations frequently contain long URLs that must break across lines. The xurl package handles this automatically:

\usepackage{xurl}

This prevents URLs from overrunning margins or creating overfull hbox warnings.

Typography Enhancement

\usepackage{microtype}

The microtype package improves character spacing and hyphenation, reducing awkward line breaks that become more noticeable on narrow e-reader screens.

Document Structure

The three-part structure using frontmatter, mainmatter, and backmatter divisions ensures proper page numbering and chapter numbering behavior:

\begin{document}
\frontmatter
\maketitle
\tableofcontents
\input{Foreword}

\mainmatter
\input{chapters/Author1}
\input{chapters/Author2}
\input{chapters/Author3}

\backmatter
\end{document}

The \frontmatter command produces roman numeral page numbers and suppresses chapter numbering (appropriate for prefaces and tables of contents). The \mainmatter command switches to arabic page numbers and enables chapter numbering. The \backmatter command suppresses numbering for any closing material.

Chapter File Conventions

Cross-Reference Anchors

Use both \hypertarget and \label to ensure compatibility with hyperref navigation:

\hypertarget{AuthorName-intro}{}
\section{Introduction}\label{AuthorName-intro}

Prefixing labels with the author name prevents collisions when multiple chapters have sections with the same name (such as "Introduction" or "Conclusion").

Footnotes

Standard \footnote{} commands work well. Kindle converts these to popup notes or hyperlinked endnotes depending on the device. Keep footnote content self-contained since readers may view them out of context.

Block Quotations

The standard LaTeX quote environment works reliably:

\begin{quote}
Extended quotation text here.
\end{quote}

Custom List Numbering

When legal outline style requires starting enumeration at a specific point or using non-default labels:

\begin{enumerate}
\def\labelenumi{\Alph{enumi}.}
\setcounter{enumi}{2}
\item First item (will be labeled C.)
\end{enumerate}

Case Citations

Use \emph{} for case names (renders as italics) and \textsc{} for journal abbreviations in citations:

\emph{Williams v. Rhodes}, 393 U.S. 23 (1968)
97 \textsc{B.U. L. Rev.} 1309

Complete Preamble Template

\documentclass[openany, 12pt]{book}
\usepackage[margin=1in]{geometry}
\usepackage[normalem]{ulem}
\usepackage{booktabs}
\usepackage{authblk}
\usepackage{quoting}
\usepackage{soul}
\usepackage{xcolor}
\usepackage{amsmath}
\usepackage{makecell}
\usepackage{tabularx}
\usepackage{xurl}
\usepackage{float}
\usepackage{graphicx}
\usepackage{microtype}
\usepackage{titlesec}

% Numbering hierarchy: I, A, 1, a, i
\setcounter{secnumdepth}{5}
\renewcommand{\thechapter}{\Roman{chapter}}
\renewcommand{\thesection}{\Alph{section}}
\renewcommand{\thesubsection}{\arabic{subsection}}
\renewcommand{\thesubsubsection}{\alph{subsubsection}}
\renewcommand{\theparagraph}{\roman{paragraph}}

% Heading formats (italicized, not bold)
\titleformat{\chapter}[display]
  {\normalfont\huge\itshape}
  {\chaptertitlename\ \thechapter}{20pt}{\Huge}

\titleformat{\section}
  {\normalfont\Large\itshape}
  {\thesection.}
  {1em}
  {}

\titleformat{\subsection}
  {\normalfont\large\itshape}
  {\thesubsection.}
  {1em}
  {}

\titleformat{\subsubsection}
  {\normalfont\normalsize\itshape}
  {\thesubsubsection.}
  {1em}
  {}

\titleformat{\paragraph}[display]
  {\normalfont\normalsize\itshape}
  {\theparagraph.}
  {0.5em}
  {}
\titlespacing*{\paragraph}{0pt}{3.25ex plus 1ex minus .2ex}{1em}

% TOC width adjustments for Roman numerals
\usepackage{tocloft}
\setlength{\cftchapnumwidth}{4em}
\setlength{\cftsecnumwidth}{3em}

% Chapter author command
\newcommand{\chapterauthor}[1]{%
  \vspace*{-10pt}
  {\large\itshape #1}
  \par\vspace{20pt}
  \addcontentsline{toc}{section}{\protect\numberline{} \itshape #1}%
}

% Hyperlink configuration
\definecolor{almostblack}{RGB}{25,25,25}
\usepackage[colorlinks=true, linkcolor=almostblack]{hyperref}

% Paragraph spacing
\setlength{\parskip}{0.5\baselineskip plus 2pt}

\usepackage{cleveref}

\title{Your Book Title}
\author{Editor Name (Ed.)}
\date{\today}

Kindle-Specific Considerations

Avoid Complex Tables

Tables with elaborate layouts reflow poorly on Kindle. The sample chapter uses block quotes and flowing prose rather than complex tabular structures. When tables are necessary, keep them simple with minimal columns.

No Problematic Floats

While the float package is loaded for flexibility, avoid placing figures and tables in locations where they might float unpredictably. On reflowable e-books, floats can end up far from their referencing text.

Test in Kindle Previewer

Before publishing, run the compiled PDF through Kindle Previewer to check for rendering errors, verify TOC navigation, test footnote popups, and confirm hyperlinks function correctly.

Pre-Publishing Checklist

• Compile without errors or overfull hbox warnings
• Run through Kindle Previewer
• Verify TOC displays all chapters and authors correctly
• Test TOC navigation links
• Confirm footnote popups work on simulated devices
• Check that all internal cross-references resolve
• Verify long URLs break appropriately
• Review on both e-ink and tablet simulations in Previewer

Adapting the LaTeX Preamble for Print (KDP Paperback or Hardback)

This guide covers the key differences between the Kindle (e-book) preamble and the print preamble. It assumes familiarity with the Kindle setup documented previously.

Font Size Reduction

% Kindle
\documentclass[openany, 12pt]{book}

% Print
\documentclass[openany, 11pt]{book}

Print books typically use smaller base fonts than e-books. The 11pt size works well for a 7x10 trim and keeps page count reasonable while remaining readable.

Custom Page Geometry with Gutter

\usepackage[
  paperwidth=7in,
  paperheight=10in,
  inner=1.30in,
  outer=0.95in,
  top=1.00in,
  bottom=1.00in,
  includeheadfoot
]{geometry}

This is the most significant change. Key considerations:

• Paper dimensions: Must match your KDP trim size exactly (here 7x10 inches)
• Asymmetric margins: The inner margin (gutter side, toward the spine) is larger than the outer margin to account for binding. Text near the spine becomes harder to read without this extra space.
• includeheadfoot: Ensures headers and footers fit within the specified margins rather than extending beyond them

For other common KDP trim sizes, adjust accordingly:

% 6x9 example
paperwidth=6in,
paperheight=9in,
inner=1.00in,
outer=0.75in,

Hyperlink Visibility

% Kindle
\usepackage[colorlinks=true, linkcolor=almostblack]{hyperref}

% Print
\usepackage[hidelinks]{hyperref}

Print books should not show colored or boxed links. The hidelinks option removes all visual indication of hyperlinks since they serve no function on paper. The custom almostblack color definition can be removed entirely from the print preamble.

URL Spacing Control

\Urlmuskip=0mu plus 1mu

This line appears only in the print preamble. It fine-tunes the spacing around URL break points, reducing excessive stretching that can look awkward in justified print text.

Table of Contents Depth

\setcounter{tocdepth}{2}

The print version explicitly limits TOC depth to 2 levels (chapters and sections). This keeps the TOC compact for physical page constraints. The Kindle version omits this setting, allowing deeper nesting since e-book navigation benefits from more granular links.

TOC Number Widths

% Kindle
\setlength{\cftchapnumwidth}{4em}
\setlength{\cftsecnumwidth}{3em}

% Print
\setlength{\cftchapnumwidth}{3em}
\setlength{\cftsecnumwidth}{2.5em}
\setlength{\cftsubsecnumwidth}{3.2em}

Print uses slightly tighter TOC spacing (3em vs 4em for chapters) since the fixed page width is known and controlled. A subsection width is also specified for the print version.

Chapter Author Command Simplified

% Kindle
\newcommand{\chapterauthor}[1]{%
  \vspace*{-10pt}
  {\large\itshape #1}
  \par\vspace{20pt}
  \addcontentsline{toc}{section}{\protect\numberline{} \itshape #1}%
}

% Print
\newcommand{\chapterauthor}[1]{%
  \vspace*{-10pt}
  {\large\itshape #1}
  \par\vspace{20pt}
}

The print version removes the \addcontentsline call. Author names do not appear in the print TOC, keeping it more compact. In the Kindle version, the indented author listing aids navigation; in print, the physical chapter opening page serves that purpose.

Summary of Print-Specific Considerations

1. Match KDP trim size exactly in geometry settings
2. Add gutter margin (inner > outer) for binding
3. Use hidelinks since hyperlinks are non-functional on paper
4. Reduce base font to manage page count
5. Limit TOC depth for physical space constraints
6. Remove author names from TOC to keep it compact
7. Fine-tune URL spacing with \Urlmuskip for justified text

Workflow Note

Maintain two separate main files (e.g., main-kindle.tex and main-print.tex) with different preambles but identical \input{} chapter references. This allows single-source chapter management while producing both formats.