A claim is only as good as the attempts to break it that it survives. So the whole project is arranged to make breaking it easy: the text, the proof, and the code are public, the counting choices are disclosed and ranked, and nothing depends on trusting the people behind it. We would rather be corrected than be comfortable.
What would falsify it
The central claim is a single identity. On a fixed, public text, two independent counts over the twenty-nine marked chapters, the letters and the words, both come to 39,349, and that number equals 19² × P(29). There are two clean ways it could fall.
The first is arithmetic. Run the proof on the same text and show that the totals do not actually reproduce. That would end it on the spot, and it is the easiest thing in the world to attempt, because the file and the code sit on the Verify page.
The second is subtler and more interesting. Show that the counting conventions are cherry-picked: that small, equally defensible changes to the rules break the result, and that there is no principled reason to prefer the project's choices over the ones that break it. The project tries to head this off by disclosing and ranking the choices up front, but disclosure is not the same as being right. The discovery framework and the piece on probability are where to push.
The manuscript frontier
This is the most important open problem, and the project says so plainly. The proof runs on the standard 1924 Cairo text in the Hafs reading, the one almost everyone uses today. The honest next level is to run the same counts against the earliest surviving manuscripts, where the bare consonantal skeleton, the rasm, differs in small ways from the modern printed text.
That is real, careful, slow work, and it is not finished. Anyone wanting to understand the manuscript record should read the scholarship directly. The work of Dr. Marijn van Putten on the early written transmission of the Quran, and the Quranic Arabic Corpus assembled under Professor Eric Atwell, are good places to start. Neither is connected to this project and neither endorses it. They are simply where the serious evidence about the text lives. The project's own piece on this is The Manuscript Question.
How to read the odds honestly
Sooner or later someone asks what the odds are of all this arising by chance. It is a fair question, and also the most abused number in the whole field, because a large improbability is easy to manufacture and easy to misread. The project's estimate sits somewhere around one in a trillion, but the more important thing is knowing what a number like that can and cannot establish. That is laid out in full in What Are the Odds. Treat any improbability, the project's own included, as something to interrogate rather than be impressed by.
Three ways to engage
Run it yourself
The fastest review needs no one's permission. Clone the code, run it, change the inputs, and see what holds and what breaks. Start on the verify page.
Send a written critique
Found a miscount, a questionable choice, or a flaw in the method? Write it up and send it. Serious critiques are answered, and anything that improves or corrects the work is credited by name, with your consent. Use the contact form.
Book a call
If you would rather talk it through, scholars, mathematicians, skeptics, and the simply curious are all welcome to book a nineteen minute conversation. No debate to win, just clarification, in both directions.
How we hold it
The mathematics is fixed: the file either reproduces 39,349 or it does not. The readings and the writing around it are offered as contemplation and open questions, not as rulings on anyone. We ask questions when we want clarity, we assume good faith, and we keep the door open. That posture is not weakness in the argument. It is the argument.