Roadmap · updated 12 August 2026
What we are doing, in what order, and what is still wrong
Most roadmaps are marketing. This one includes the parts that are not working.
Where things stand
All four books are finished. Written, fact-checked against their sources, read cold by two independent checkers each, illustrated wherever a licence allowed it, and packaged as ebooks that pass validation with no errors and no warnings.
None of them is on sale yet. That is a publishing decision, not an unfinished manuscript.
| Thing | State |
|---|---|
| Books finished | 4 of 4 |
| Total words written | 200,463 |
| Plates in the finished interiors | 22, every one licensed for reuse |
| Ebooks validating with zero errors | 4 of 4 |
| Books on sale | None yet |
| Articles published here | 22 |
| Surveys run and scored | 7, all published with their failures |
| Claims we have had to withdraw | See below. There are some |
The books, in publication order
All four are finished, so the order is a publishing choice rather than a queue. It is set by which subjects readers actually search for, which is not the order they were written in. No dates. This site does not put a date on a book before the work behind it is done, and that rule has already cost us one comfortable announcement.
Publishes first
Answering the Architect Question
Atlantis, the Younger Dryas and the lost civilisation debate, checked one claim at a time
Both cases at the same length, measured and printed at the back. The argument runs across four papers in one journal and you can read all of it.
Finished manuscript. 46,943 words · 1 licensed plates · checked by two independent readers · ebook validates clean
Then · 2 of 4
Lost Under the Sand
Ancient Egypt, the pyramids, and what the machines can see through rock
A Nobel laureate pointed muons at a pyramid in 1970 and found nothing. That null result is why the 2017 discovery is worth believing.
Finished manuscript. 56,517 words · 8 licensed plates · checked by two independent readers · ebook validates clean
Then · 3 of 4
Found in the Scrolls
Reading the Dead Sea Scrolls, and the texts nobody can open
Machine decipherment works on Linear B and Ugaritic. Both had a known relative. Linear A has neither, and a hundred years has not moved it.
Finished manuscript. 49,745 words · 2 licensed plates · checked by two independent readers · ebook validates clean
Then · 4 of 4
Lost Under the Canopy
Lost cities, ancient civilisations, and what the machines found under the jungle
Six surveys, 907 candidates, one confirmed. Then the same four questions turned on the most famous survey results in the world.
Finished manuscript. 47,258 words · 11 licensed plates · checked by two independent readers · ebook validates clean
What happens next
- Answering the Architect Question goes on sale. Ebook first, paperback a few weeks behind.
- Advance readers get the rest free. The complete manuscript, before publication, in exchange for one honest review. Join the ARC team.
- The remaining three follow, roughly a month apart, in the order above.
- More articles, continuously. The two that matter most are the ones nobody else writes: what each instrument cannot see, and what our own surveys missed.
Open problems
The part of a roadmap that usually gets left out. These are unsolved right now.
One of our own headline figures turned out to be noise. We reported 28.7% recall on Bodmin Moor. Then we checked what 470 randomly scattered points would score on the same ground: 28.2%. On that block our number says almost nothing about the detector. On Salisbury Plain the same detector beats random by 21 points, so the method works and the figure did not. Both are published, and every recall figure here now carries its random baseline in the same table.
Run 007, 9 August 2026. 400 rounds of randomly placed points matched to the real candidate count, scored against the same monuments and the same tolerance rule as run 006.
One book will have no illustrations. Every paper behind the Younger Dryas argument is either paywalled or free-to-read with no reuse licence, so we cannot legally print a single figure from any of them. Rather than commission artwork that would be a speculative reconstruction, that book ships without pictures.
Licence trace of 15 cited papers, 9 August 2026. Four permit commercial reuse.
Three fields still will not publish how often they are wrong. No false-positive rate for the statistical argument about the Indus script. No character error rate for machine reading of the Herculaneum scrolls. No pre-registered threshold for the Younger Dryas impact markers. We keep asking, in public, and we had to run the equivalent check on ourselves before we had any standing to.
Open request, restated three times across three disciplines and unanswered in all three.
How to hold us to this
Every factual claim on this site carries one of five confidence labels and a source you can check. What each label means is on the standard. Every survey we run is scored against the National Heritage List for England, an independent public record, and the run logs are published with the results.
If something here is wrong, the useful thing is to say so. Corrections get published rather than quietly edited: contact.