How the Great Pyramid of Giza Was Built: A New Computational Theory Says the Ramp Was Hidden Inside It All Along
For more than two thousand years, anyone trying to explain how the Great Pyramid of Giza got built has run into the same brick wall. Or rather, the same 2.3 million limestone blocks. The math is brutal. To finish Khufu's tomb inside the pharaoh's roughly 27-year reign, builders would have needed to set a stone every two to three minutes during working hours, for two decades, with copper chisels and rope. No iron, no pulleys, no wheels carrying serious weight. Herodotus wrote about ramps. Modern Egyptologists have argued about ramps. None of the proposed ramps quite work.
A new paper in the journal npj Heritage Science takes a different swing at it. The author is Vicente Luis Rosell Roig, a Spanish computer engineer with a PhD and no formal Egyptology training, who started sketching the idea in 2020 after watching a documentary and finding the standard explanations unsatisfying. His proposal: the ramp was never visible because it was always part of the pyramid. Builders left certain perimeter stones out as they worked upward, creating a helical path along the edges of each face. As construction progressed, those gaps were backfilled from the top down. By the time the capstone went on, the ramp had quietly become the building.
He calls the model the Integrated Edge-Ramp, or IER. And unlike most pyramid theories, which tend to live and die in conference rooms, this one comes with code, simulations, and a list of physical predictions an archaeologist could go test tomorrow.
Start with what we actually know. The Great Pyramid is 230 meters on each side at the base and originally stood 147 meters tall. The blocks average around 2.5 tons. Some of the granite beams above the King's Chamber weigh between 50 and 80 tons. The total construction window, accounting for quarrying, river transport, seasonal flooding pauses, and the actual masonry work, has long been estimated at 20 to 27 years. Pierre Tallet's discovery of Merer's papyrus diary at Wadi al-Jarf in 2013 gave us, for the first time, a logbook from an actual workman during Khufu's reign. He was hauling Tura limestone casing blocks down the Nile. The construction was real, organized, and bureaucratic.
The problem is what happens at the pyramid itself. A straight external ramp tall enough to reach the top would require almost as much material as the pyramid. Spiraling external ramps wrapped around the structure run into surveying nightmares because you can no longer sight the corners to keep the geometry true. Internal corkscrew ramps, the kind Jean-Pierre Houdin proposed in the mid 2000s, have appeal but constrain the workflow to a single moving column of stone. At the pace required, that column becomes a bottleneck.
Rosell Roig built a 3D parametric model that places stones one at a time using a discrete-event logistics simulator. Then he ran a finite element analysis on each stage of construction to check whether the partially built pyramid, with chunks of its perimeter temporarily missing, would actually stand up. The headline result: an adaptive multi-ramp version of the IER can sustain dispatches every four to six minutes, gives a median on-site duration between 13.8 and 20.6 years at 95 percent confidence, and pushes the total project, with quarrying and barge runs and seasonal stops, to 20 to 27 years. That tracks with every previous estimate, but for the first time the number falls out of a bottom-up simulation rather than a back-of-envelope calculation.
How the ramp hides itself
Think of each face of the pyramid as gaining one stone course at a time. In the IER model, the outermost ring of stones on a given course is left incomplete in a few specific places. Those gaps trace a sloped path along the edge of the pyramid, climbing upward like a ribbon wrapped around the structure. Workers drag sleds full of blocks along this path, deliver them to where they're needed in the interior, and move on. When the next course goes down, the gaps shift to maintain the helix. Critically, when a section of ramp is no longer needed, it gets backfilled with the stones that were always supposed to go there. The ramp does not get torn down. It gets completed.
Run that strategy on all four faces simultaneously, and a single ramp becomes four parallel ramps. The wide bottom courses, where horizontal distribution dominates the workload, can support all four at full throughput. As the pyramid narrows toward the top and there is less perimeter to work with, the system reduces to fewer active corridors without disrupting the flow.
That parallelism is the part previous models missed. Houdin's internal ramp was elegant but serial. The IER is a pipelined, distributed haul system, which is exactly what you would design if you handed the problem to a logistics engineer.
Below is the basic geometry of how the temporary ramp sits inside the courses and what backfilling looks like as the pyramid rises.
What the muons saw
This is where the IER stops being just clever and starts being interesting. Since 2016, a multinational project called ScanPyramids has been mapping the inside of Khufu's monument using cosmic-ray muon radiography. Muons are charged particles produced when cosmic rays slam into the upper atmosphere. They pass through stone but lose energy at predictable rates, which means detectors placed inside or beside a structure can build up density maps of what is solid and what is empty.
ScanPyramids found two large unexplained voids. The smaller one, behind the chevron stones on the north face, was confirmed in 2023 to be a corridor roughly nine meters long and two meters by two meters in cross section, with a vaulted ceiling. They got a tiny endoscope through a six-millimeter crack in the masonry and took the first photographs of a space no human had seen in 4,500 years. The larger anomaly, the so-called Big Void above the Grand Gallery, is around 30 meters long and has not yet been visually inspected.
These cavities have launched a thousand speculative articles. Hidden burial chamber? Treasure room? Stress-relief gap?
Rosell Roig's model offers a more boring possibility, which is usually a sign you are onto something. The geometry of the IER's predicted ramp paths, including their slopes and the corner turns where the helix wraps around the pyramid's edges, lines up with the position and orientation of several of the muon-detected anomalies, including the North Face Corridor. Under this reading, the voids are not religious or symbolic. They are leftover construction infrastructure. Bits of the haul corridor that, for whatever reason, did not get fully backfilled when the project closed out.
The southeast corner of the pyramid, in particular, shows distinctive wear patterns that have puzzled archaeologists. Rosell Roig argues these could mark entry points where the heaviest stone flow occurred during construction. The blocks coming up from the quarries on the south side of Giza would have funneled through that corner before being dispatched along the helical path. If you were running the simulation and asking which corner should look most beaten up, that is the one the model picks. The match is not proof, but it is the kind of unforced agreement that makes you raise an eyebrow.
That is a falsifiable claim. If the next round of robotic endoscopy or deeper muon scans reveals the predicted edge-fill signatures, intermediate corridors at the geometries the model expects, or compaction wear on the southeast corner where the model says the heaviest traffic flowed, the IER gets serious empirical support. If those features are absent, the model is in trouble. Either way, you learn something.
One thing I find honest about Rosell Roig's paper is how strict it is about Old Kingdom technology. No wheeled heavy transport. No iron tools. No compound pulleys. The toolkit is copper chisels, water-lubricated wooden sledges, ropes, levers, earthen embankments, and Nile barges. That last one is not a small detail. The Tura limestone for the casing came from quarries on the east bank of the Nile, and the granite for the inner chambers came all the way from Aswan, 800 kilometers south. Merer's diary describes barge runs delivering Tura blocks during the final years of construction, when the casing was being put in place.
The Hatnub quarry, about 250 kilometers south of Giza, gave us our best physical evidence of how Old Kingdom workers handled steep grades. Excavations there in 2018 found a ramp from Khufu's reign with a 20 percent slope, flanked on both sides by staircases studded with postholes. The going theory is that workers tied ropes around wooden posts and hauled sleds up the incline using the posts as friction points to multiply pulling force. Not pulleys in the modern sense, but the same idea. The IER model assumes ramp slopes consistent with what was demonstrably possible at Hatnub.
And the Sinki pyramid, an unfinished step pyramid south of Abydos, preserves four small construction ramps, one on each face, going up perpendicular to the structure. That is direct, on-site precedent for the multi-face parallel ramp strategy the IER requires. It is not Khufu, but it is the same culture, broadly the same era, and the same architectural problem at a smaller scale.
The labor side has also become less mysterious in the past few decades. Excavations of the workers' settlement south of the pyramid by Mark Lehner and Zahi Hawass have shown a permanent town with bakeries, breweries, and a rotating workforce that probably numbered around 20,000 to 25,000. Skeletal evidence shows the workers were well fed, sometimes well treated, and definitely not slaves. They were divided into named gangs, and graffiti scratched into stones inside the relieving chambers above the King's Chamber actually preserves some of those gang names, including phrases like "Friends of Khufu" and "Drunkards of Menkaure." It was a national project run with bureaucratic precision, which is precisely the kind of organization that could sustain a tightly choreographed multi-ramp haul system without it falling apart.
What sets this paper apart from previous pyramid theories, including some I find more aesthetically pleasing, is the way it was constructed. Rosell Roig published the entire computational pipeline on Zenodo as open code and data. Other researchers can change the input parameters, slope, ramp width, friction coefficient, block size, work crew capacity, and rerun the simulation against the Khafre or Menkaure pyramids, the Bent Pyramid, the Red Pyramid, or anything. It is a framework for testing construction hypotheses, not just a one-shot proposal.
That is a different kind of argument than has typically been made about the pyramids. Most theories end at the conceptual sketch. This one ends at a Git repository.
There are still serious gaps. Finite element analysis is only as good as its inputs, and Old Kingdom limestone is not a standardized material with well-characterized failure modes. The model's stress and settlement results stay within plausible bounds, but plausible is not proven. The granite beams above the King's Chamber, some of which weigh 60 tons or more, are not really addressed by an edge-ramp strategy designed for typical 2.5-ton blocks. Houdin's separate proposal of an internal grand gallery used as a counterweight track for those heavy beams might still be needed as a complement.
And the IER is, as Rosell Roig himself notes, a hypothesis. The presence of the Big Void above the Grand Gallery is consistent with his model. It is also consistent with a couple of other readings. We will not know until someone gets a camera in there.
I have read a lot of pyramid construction theories. Most are someone's hobbyhorse dressed up in pseudo-engineering. The good ones, like Mark Lehner's external spiral ramp or Houdin's internal corkscrew, are plausible but unfalsifiable in any practical sense. You cannot really test them without dismantling the pyramid, which nobody is going to do.
The IER is different in a specific way. It makes predictions about features that should still be present inside the existing structure. Wear patterns at corners. Filled-in notches at specific elevations. Density anomalies at specific positions. Muon scanning is getting better every year. The ScIDEP collaboration is now scanning Khafre's pyramid with new scintillator-based detectors. The technology to test Rosell Roig's model exists and is being improved.
The pyramid is sitting there. Whatever ramp built it, if any trace remains, is sitting inside it. Sooner or later somebody is going to look.