I thought I had a feedback problem.
I had written a manuscript, shared it with a few thoughtful readers, and asked them to tell me what needed to change. The book was still moving. One day I could rewrite a paragraph and move on the next. That was the point of asking for help early.
The first explanations were practical. Readers needed an easier way to leave comments. I needed a cleaner way to collect them. Maybe the answer was a better note-taking habit, a more consistent reference format, or a shared place for responses.
The longer I worked around the problem, the less convincing that explanation became. The issue was not that feedback had too few features. Feedback kept losing what made it useful: the text it was responding to.
The delivery had worked
The file itself had done what I asked it to do.
It reached the reader. It opened on a familiar device. It could be saved, moved, read later, and returned to without requiring a new system. If I wanted to give someone a finished artifact, those would be excellent properties. Portability, offline access, a clear copy, and a simple handoff are not minor conveniences. They are much of what makes a digital book useful.
I was not discovering that files were obsolete. I was discovering that I had asked one to support a different kind of work.
An EPUB or PDF can be the right form for a book someone wants to carry, archive, lend, or read without a connection. A downloadable copy can be exactly what a reader should receive when the work is finished, and the relationship to it is primarily one of access. The format is not the problem simply because it is a file.
My problem began after the file arrived.
The delivery had succeeded. What I wanted next was harder to describe: I wanted a reader to notice something, hold it next to the passage that prompted it, and send the thought back while the manuscript could still respond. The handoff was complete, but the work was not.
That difference was easy to miss because the file made the first part feel complete. Once the attachment was sent, the book looked as if it had reached its destination. In reality, the feedback loop had just started.
A message is not a margin
The most revealing moments were ordinary.
A reader would reach a section and send a message with a useful recommendation. Sometimes the message included a copied fragment. Sometimes it referred to a chapter or a general location. I could usually understand the point. The extra work was figuring out exactly where the point belonged.
After one recommendation, I revised a paragraph and sent a new version of the manuscript. Later, a note came back with a few words from the earlier copy. I had to work out which version the reader had seen, search for the fragment in both files, and read around it before I knew whether the note was about the old wording, the new wording, or the paragraph next to it.
Nothing had gone dramatically wrong. The reader had made a good observation, and I had made a reasonable revision. The conversation had simply become a small investigation before it could return to the idea itself.
On a phone, the separation started even earlier. A thought appeared while the reader was in the book, but the response had to be made somewhere else. The reader had to leave the passage, carry enough of it into a message to make the note intelligible, and trust that I could carry it back.
That was the part I could not fix with better instructions. A page number could identify a place in one copy. A quotation could identify words across copies. Neither one carried the whole moment in which the reader had understood, questioned, or resisted the passage.
The copy problem
In ordinary speech, I would say I sent you the book. What I actually sent was a copy of the book at a particular point in its life.
The book was still a work in progress. The file was a snapshot that let another person receive it. Once the snapshot left, the reader could make notes, I could revise the manuscript, a message could preserve a fragment, and a second file could replace the first. The book remained one project in my head while the actual work spread across copies and message threads.
For a finished artifact, that separation may be a virtue. A stable copy is easier to preserve than a changing conversation. You can store it without also storing every decision, response, and revision that led to it. The file’s boundedness is part of its usefulness.
For an unfinished manuscript, the same boundary created a different cost. The reader was not only receiving content; the reader was helping determine what it would become. A response needed a passage, and the passage needed the version being discussed. The copy was necessary, but it was not the whole situation.
I started noticing how often I used the word “book” for several different things: the manuscript I was editing, the file I had sent, the passage a reader had in front of them, and the conversation we were trying to have. They overlapped enough to feel like one object and separated at exactly the moments when precision mattered.
The file preserved the text. It did not naturally preserve the relationship around the text.
What I was asking the file to do
I had expected one attachment to do too much. It had to deliver the manuscript, give the reader a place to read, and somehow keep us oriented when the wording changed. When that failed, I added more instructions around the file: cite a chapter, copy a phrase, label a version, move the response elsewhere.
Those steps made the reader carry context out of the book and made me put it back. A careful reader could provide an excellent note and still leave me with a reconstruction task.
The issue did not come from individual care. The readers were doing exactly what I had asked. The structure of the request was doing the damage. I had separated the text from the response, then hoped people would keep reconnecting them by hand.
The distinction I had begun to call possession and participation helped me name the cost. Possession means a reader can receive a copy, keep it, and open it again. Participation begins when the reader is trying to change or discuss the work: the passage, the writer, and the response have to stay intelligible together. A file handles the first job beautifully. It was not enough for the second.
A file can contain content very well. A changing conversation is harder to carry because it depends on who read which wording and what happened next. I had been treating those connections as attachments around the file.
The requirement changed
I had begun by asking how to make comments travel better. The diagnosis led somewhere else.
The requirement was not a more elaborate message or a stricter reference format. It was a shared point of return. When a reader had a thought, the passage should stay close enough that the thought didn’t first have to become a detached object. When the manuscript changed, the relationship should be understandable rather than reconstructed from fragments.
Finished books should keep their portability, and a file should remain part of the publishing process. The reading environment also needed to acknowledge what was happening around the book, especially when the work was still open to change.
This was the diagnosis underneath the first idea that would eventually become ReaderPub, but the diagnosis mattered more than the name. A book file was very good at preserving a copy of a book. It was much less suited to being the place where the life around that book continued.
Once I stopped asking how to make comments travel better, the next question became unavoidable: what kind of reader would let the text itself remain the place everyone could point back to?
ReaderPub has continued to grow from the experiment described here. The catalog now includes its first commercial books alongside the public-domain collection, so the question of what a digital book becomes after delivery is no longer only theoretical.