Who produced it, how to reproduce every result, and what we're claiming versus what we aren't.
This site presents original independent research produced in May 2026 using publicly available academic software and peer-reviewed datasets. Transparency about methodology, limitations, and the difference between data and interpretation is central to what we're doing. This page covers all of that.
I am Malcolm, an independent researcher based in Hull, UK. I have no academic background in archaeology, geology, astronomy, or any related field. I run a cleaning business. What I have is a persistent curiosity about the ancient world, a willingness to learn technical tools, and the determination to test ideas with numbers rather than just argue about them narratively. This research was produced in one intensive session in May 2026 using Python, publicly available geological software, and a great deal of patience with error messages.
Academic researchers face publication incentives, institutional pressures, and career consequences for pursuing unconventional hypotheses. An independent researcher faces none of these. The tradeoff is no peer review, no institutional credibility, and no guarantee of methodological correctness. We've tried to compensate for the latter by making everything reproducible and being explicit about limitations. The former is the reader's call to make.
This is not a conspiracy theory site. It is not claiming governments are hiding aliens. It is not claiming the pyramids were built by extraterrestrials as an established fact. It is presenting quantitative evidence — real numbers from real software — that the ancient sites are non-randomly placed, and offering a hypothesis that is consistent with that evidence. The hypothesis is labelled as hypothesis throughout.
Every statistical result on this site can be reproduced by any researcher with access to a computer. The tools are free. The data is public. The scripts are available on request. You do not need to take our word for any number on this site — you can run the same analysis yourself and check the output.
pyGPlates is the Python interface to GPlates, an open-source plate tectonic reconstruction software developed by the EarthByte Group at the University of Sydney. It is freely available from gplates.org. Install using pip:
Python 3.12 was used for this analysis. Earlier versions should work but have not been tested.
The rotation model is the Müller et al. 2022 SE 1Ga Optimised Plate Motion Model v1.2. It is freely available from the EarthByte data portal at earthbyte.org. The file you need is the .rot rotation file. Download it and note the path.
Email research@ancientaliens.co.uk and we will send the full Python script set including: the site coordinate file with plate IDs; the alignment detection script; the Monte Carlo significance test; the control group test; the multi-period temporal analysis; the star alignment quality calculator; and the Antarctic convergence test.
Running the analysis will reproduce:
Minor variations in output (±5 alignments) may occur due to floating-point rounding differences between Python versions and operating systems. The core findings — Z-score, p-value, 20Ma peak, 10,500 BC convergence — are robust to these variations. If you reproduce the analysis and get significantly different results, please email us — it means we have an error that needs correcting.
The ancient sites are non-randomly placed. That is a mathematical fact supported by 10,000 simulations. Something positioned them according to a global geometric system with understanding of the Earth's tectonic history that we have not previously credited to any ancient intelligence. What that something was — human, non-human, or something else entirely — is not answerable from the data alone. We offer the alien intelligence hypothesis as the most coherent framework consistent with all the evidence. We offer it clearly as a hypothesis. The data is real. The interpretation is ours. The two should not be conflated.
The 68 sites were selected by the researcher and may introduce selection bias toward sites known to be significant. A randomly selected set of 68 ancient structures would be a stronger test. We partially addressed this with the modern landmarks control group, but the selection question remains legitimate.
Some sites near plate boundaries have approximate plate ID assignments. The Middle East, Southeast Asia and India-Eurasia collision zone are particularly uncertain. Incorrect plate IDs would affect the temporal analysis results for those specific sites. The core Monte Carlo result uses present-day positions only and is not affected.
The 2-degree threshold for alignment detection is a design choice. Tighter thresholds produce fewer alignments and higher specificity. Looser thresholds produce more alignments and lower specificity. We chose 2 degrees as a balance — but the result would be different with different thresholds. The Z-score remains significant across a range of thresholds, but the exact numbers change.
These limitations are noted in the full paper (Section 10) and are offered here in the spirit of transparency. We believe the core findings are robust to these concerns — but we could be wrong, and we say so explicitly.
For script requests, data queries, methodological questions, or collaboration proposals:
The full paper has been submitted to the Graham Hancock Foundation for consideration. If you are a researcher connected to that organisation:
If you are a journalist, documentary maker, or broadcaster interested in this research, please contact us via the research email address. We are happy to discuss the methodology and findings in detail.