How awesome do you feel right now? This is HUUUGE! To think that a scroll was unreadable for so, so long, until we invented machines that let us read it slice by slice. It's such an unfathomable achievement - we made machines that let us read 2000+ year olds fragile scrolls without ever opening them - and you helped do just that.
In March I went to Beam Line 18 at the European Synchrotron Radiation Facility. I had to swap out the scrolls on the xray pedestal. Scrolls that were presented as a diplomatic gift to Napoleon and Josephine by King Ferdinand. France has 2 of the 6 that they were given still in tact. I had to handle both of them. I have never felt more stressed in my life and have never and will probably never again handle such a priceless artifact.
I feel the opposite of that feeling and am immensely proud of everything that the core challenge team has accomplished
I am floored at these achievements. Such amazing work.
If I may ask, when you started thinking about achieving this, what were the first attempts, ideas on how to go about it? What were some of the obstacles that had to be overcome to achieve this ?
The process of trying to read the scrolls has been going on for about 275 years or so, now. Doing it nondestructively via CT scanning and virtual unrolling and reading has been in the works for 25 years or so, so it's a lot of building on previous work.
Virtual unrolling and reading are not terribly hard to do manually, they are just not feasable on a large scale. Like years and years of human time spent tediously clicking on papyrus and labelling ink in renders, so a large amount of automation is required.
A lot of difficulty has come from the first step: xraying the scrolls. It's hard and expensive and difficult to get right. The efforts since this all began with CT scanning 25 years ago has been kneecapped by the data simply not being good enough. We xray on what is AFAIK literally the most powerful xray beamline in the world and we would still like for it to be more powerful and faster. Not to mention the massive amounts of data. For Pherc Paris 3, our largest scroll, the raw reconstructed data is 260 terabytes. That's a lot of data to have to deal with.
Where can we read about the xray setup? e.g the type of sensor, if/how the target and/or beam is scanned, any fancy gratings/etc, what kind of CT algorithms are used
Do you know what kinds of features the model is picking up on to distinguish ink from papyrus? And did you have any labeled data (images where a human expert has identified ink or perhaps a scan of a burnt scroll with known content) to help train it?
Certainly my Mark 1 eyeballs would not obviously perform better than random guessing at this task. Although my eyeballs are, if nothing else, nerfed by only being able to see a 2D slice of the data.
Yes. Most of the ink we have come across is carbon based. This leaves a certain texture on the scrolls that is recoverable and viewable with fairly basic physically based rendering, though how much ink is recoverable varies greatly from one character to the next. I don't have links handy but we just published updates to our data viewer page on our website. Pherc.Paris.4 I believe has the best overlay of ink.
A lot of labeled data is available on our ftp server which has public access
When you say "physically based rendering" do you mean that one could build a PBR model based on the (unrolled?) xray data, render that model, and be able to see the ink?
Some images on that page, specifically the "alpha composite" and "combined alpha" images, are a pretty simple PBR (if it's even that complex; it's just a composite rendering over a 3d array to a 2d image) rendering with no ML based ink detection in the input.
Outstanding work! I've participated in the challenge, but didn't get far. One of the questions I had at the time was - if I'm going to use ML to detect ink, could it invent hallucinated letters, or even parts of text, and how to prevent that?
Yes, it's quite possible for ML to hallucinate ink, though it is on a much more local scale, like predicting a slightly longer stroke, filling in more of a character than is actually in the data, etc. Perhaps enough to change a reading of a character or show where ink isnt. It is difficult for ink detection to hallucinate grammatical and idiomatic greek and latin.
What is the input to the ML algorithm? Does it know the surrounding context so that it has a chance to deduce "if this stroke is slightly longer then the end result will be idiomatic greek and latin"?
Just as with redacted documents (consistently blocked terms) or bad OCR jobs (wrong or missing characters), even if only a certain percentage comes out unmangled it is more readable than having no data at all.
A stable base corpus and some dynamic programming will allow you to clean up the remainder[0].
The core challenge team is focused on the technology side to provide the images of ink to our team of papyrologists and they do the transcription, translation, reading, and scholarship.
This announcement was part of a larger conference being put on by Frederica Nicolardi, our lead papyrologist. The livestream of each day are available at: https://www.youtube.com/@cispemgigante/streams .
We've been trying to automate since the beginning. A lot of it is automated but it's mostly the easier and less damaged parts of the scrolls. Scanning takes a few days for the biggest scrolls but the amount of human refinement is still a multi month process.
Random shower thought: I wonder if it would be better in the long term to stop digging out archeological findings. The more we excavate, the more damage we do for future archaeologists who will have the superpower of reading these texts without even needing to dig the scrolls free and open them.
Archaeologists think about this a lot. Many digs leave portions intact specifically so that future scientists, with access to techniques and technologies beyond what's available now, can research them.
Modern archaeologists are painfully aware that theirs is a destructive science, and do their best to mitigate that. The most extreme example is probably the tomb of the First Emperor, Qin Shi Huang, where official policy on excavation can be boiled down to "not yet".
We stand on the shoulders of those that came before us. People have been trying to unroll and read the scrolls for 250 some odd years now. Had they not laid the groundwork for all that time we wouldn't be making the progress we are now.
IIRC 99% of all of the existing scrolls are still in Italy's possession. I think the breakdown is something like ~350 are mostly in tact, another ~1000 are damaged but still "scroll like", and the remaining hundreds are shattered fragments.
We are a core team of about 10 researchers and developers working full time on work that applies to all of the scrolls. We also ahve 4 full time annotators that tend to work on one scroll at a time. The amount of time spent on any given scroll varies with how difficult and large it is.
There is an extremely large overlap between a lot of the work we do with medical imaging, CT scanning, XRay technology, and such. A lot of the ML models and frameworks we have used and adapted for our purposes originated in the medical field for things like cancer detection or segmenting different body parts.
I am researching for a talk on the philosophy of code, the similarities of engineering and art, and why we enjoy reading old code. This amazing work you folks have done may be an interesting tangent.
The biggest question I have for you is why you imagine we are so interested in reading these old scrolls. Surely some of it is to see whether or not, technically, we can. Surely some of it is to get a glimpse into the human expression inscribed on them. Are we looking to learn anything, or just to connect with our ancestors? I'd like to hear your take on it, both for why you think it's important and, if you know, why your colleagues feel similarly.
I wrote this as an answer to a different question but I think it applies to what you're asking as well
> Though I have an interest in Old Norse and I spend a lot of time reading Scandinavian runestones. > 90% of them are grave markers for a dead father, mother, brother, sister, cousin, etc. If I've learned anything from that, it's that people across time and space all lead lives as real and complex as anyone else's. Their joys were as high as mine have been and their sorrows as low as mine have been.
A VSauce video I watched a long time ago described that realization as "chronosonder". I think trying to understand those that came before us and why they made the decisions that they did given the circumstances they were in can help better inform us of the things we choose to do given our own circumstances.
Otherwise, I think that a lot of things are worth doing just to see if it's possible. I like to lift weights and I'm training to lift the Dinnie Stones one day; a pair of stones that are a combined ~730 pounds. The physical and mental benefits of exercise and training are well documented and great but at the end of the day I just _really_ wanna pick up 2 stones. There's nothing more to it than that, and that's ok with me.
One of the things we said a lot in 2023 was "We just wanna read the scrolls" but that slogan has unfortunately fallen a bit by the wayside as the goal and path got longer and initial hype started to fade, but I think it perfectly encapsulates why: The scrolls are there. They can be read. Why not read them?
1. Why is that a realization, are there really people who say "Scandinavians are just mechanical" or "9th century people were made out of wood"? Why would their lives be assumed not to be "real", what even is that mindset?
2. "Real and complex lives" doesn't mean "just the same as ours", mind you.
> 1 Are there really people ... Why would their lives be assumed not to be "real", what even is that mindset?
Yes, there are a very great many!
The philosopher David Gray says that most modern thinking sees our way of life and liberalism and "progress" as meaning growth and change. It implies it is inevitable, a kind of always changing improvement.
Change that has occurred is for the good and its impossible to go back. I like the ${current_year} meme where someone says "it's 2026 things have changed, sweety". The joke is funny because that's what people actually say and that they say this every year but they don't notice that they say that every year.
So the modern way of life has many people who view people in the past as not real, as figuratively made of wood, who are primitive, who didn't lead complex lives.
David Gray concludes by saying that Liberalism therefore needs to be constantly fought for, that you cannot rest on your laurels and think that humanity is naturally and inexorably progressing.
These scrolls and History as a whole challenges a fundamental psychological investment in modern liberalism.
To think of the world as always improving and evolving for the better directly opposes a kind of empathy about how people 2500 years ago are the same human beings as we are. The scrolls should humble us.
Given this.
> 2. "Real and complex lives" doesn't mean "just the same as ours", mind you.
They are more like ours than we like to imagine. We prefer to think of ourselves as improved.
I am not a papyrologist or a classicist, rather I'm a computer scientist, so my expertise is unfortunately not in _what_ the scrolls say, rather how we get there. That being said I think and hope that there will be a trove of things that has no known provenance at all, completely lost works that elude the public memory.
Other members that were on the team before me had already proved it out before I came along so I knew it was possible. The cool thing for me though was specifically doing some physicically based rendering techniques. How well these work varies greatly, but on a few segments in one scroll they work extremely well. I whipped up some simple code to composite layers, did up a render, and without any ML at all was looking at multiple rows of text that no one had read for 2000 years. That was neat.
There's also the Telegony. Odysseus has a son through Circe who winds up killing him and marrying Penelope. Odysseus son through Penelope, Telemachus, marries Circe. There's some wild stuff that doesn't survive.
Looking through these it’s crazy to find out that The Iliad is only 1 of like 5 original texts on the Trojan war. We’re reading book 2 of a 5 book series
That's what was thought, but maybe not -- only one of the three so far looks Epicurean, which is not what was expected. Maybe it's a fluke, but historians are buzzing a bit about whether it might be broader than expected.
The Epicureans were particularly hostile to the Jews and Christians, because Epicureans deny Providence or the active intervention of the divine in human affairs. See Horace Sermones 1.5.
It's more like the Christians and the Jews were particularly hostile to Epicureans and Stoics, because those mocked the claims about the existence of an all-powerful God that requires prayers.
The Epicureans and Stoics did not care much about Christians and Jews, but after the Christians obtained the power in the Roman Empire they made great efforts to persecute and discredit the Epicureans and the Stoics, as the most dangerous kinds of non-believers. (Unlike the rational Epicureans and Stoics, the traditional polytheists could be much easier converted to Christianity, by inventing a set of Christian saints to which the former polytheists could redirect the prayers and the holidays to which they were habituated.)
The Christian propaganda has created a false image of the Epicureans, which has persisted until today.
The Epicureans were not atheists, but they had a very different conception about what Gods are. They thought that in nature there are a lot of entities that have a god-like power, i.e. humans are too small and weak to influence them in any way, but the life of the humans is strongly dependent on the actions of those entities, so they can rightly be considered as gods. Examples of such entities are the Sun, the Moon, storms, volcanos etc.
Unlike in the traditional Greek and Roman religions, where it was believed that for each such natural phenomenon there exists some sentient god, who can be convinced to change the events to a more favorable outcome by prayers and sacrifices, the Epicureans believed that the gods, even supposing that they were sentient, in any case they do not care about humans more than humans care about ants, so there is absolutely no point in praying to them or bringing sacrifices to them.
Therefore humans should conduct their life according to ethic principles, but without worrying about what gods may think about their actions.
Many modern humans would probably agree with the Epicurean philosophy, which was completely different from what the Christian propaganda claimed, e.g. that Epicureans were some kind of sinners addicted to pleasures.
Indeed, the 3 beliefs attributed to Epicureans there, i.e.:
a) one who denies the existence of prophecy and maintains that there is no knowledge communicated from God to the hearts of men;
b) one who disputes the prophecy of Moses, our teacher;
c) one who maintains that the Creator is not aware of the deeds of men.
are actually accurate enough renderings of what an Epicurean might have said in a discussion with a Jew, because as I have mentioned, Epicureans believed that there are gods, but those do not pay attention to humans and do not attempt to communicate with humans, because humans are insignificant for them.
This is quite different from how Epicureans were portrayed in Christian literature, where calumnies against them were preferred for avoiding any direct controversy.
in the paper it says "The recovered text is a philosophical treatise on ethics, and the evidence points to a Stoic work: it turns on human nature, impulse, and the moral progress of human beings, and its final preserved column names Aristocreon — nephew and disciple of the great Stoic Chrysippus — which, together with the language and themes of the text, places it in a Stoic context and dates it to the 2nd century BC."
I'm interested to know about the approaches that you tried with the ML, and then decided to not use. In practice, the options are so many. How did you come up with the final approach - and was there a systematic way to decide which options to go for?
I am not on the research team, rather on the production side of things, so my knowledge on that is pretty limited. I think one of the main takeaways from a lot of the research, though, on both the segmentation side and the ink detection side, is that it's a lot less about what models and techniques and such you use, but how good your training data is. Gathering ground truth is hard, and if you don't have a lot of good ground truth, it doesn't matter if your code is perfect, you'll never get results.
That is a general truth of most ML; many models _can_ find the information in the data, if the data is good enough. If it is not, then likely no model can.
You brought up what I'm most curious about: Where does the ground truth come from for this work since you can't just to unwrap a scroll to tell if the model got it right or, presumably, make a facsimile scroll and wrap it up.
The ground truth comes from manual work. The scrolls can be unwrapped virtually, manually, through extensive pointing and clicking by a human on the boundaries of the scroll. This, in and of itself, is not particularly hard in sections of the scroll that are preserved well, but is extremely tedious and slow and error prone. We have a team of annotators who do manual annotation and refinement through custom software we've written, mostly improving on automatically generated segmentations and unwrappings.
Once you have some unwrapped papyrus, you can render it to an image and look for ink. Ink leaves a certain texture that can be identified by the naked eye and labeled. Between these two processes you get the segmentation and ink detection ground truth. Segments can be flattened virtually through existing software and algorithms.
I'm sure that process is described somewhere on the project's site and, being a lazy human (and unwilling to ask LLMs to summarize it for me), I leaned on you for a human answer. I really appreciate you taking the time to answer. Thank you.
I can see why you'd be attracted to this project from a "let's solve problems computationally" perspective (never mind the historical side). It sounds like there are some cool problems in there.
The eye toward automating the process that the project seems to be targeting is particularly cool, too. This kind of stuff that makes me have real enthusiasm for ML.
Imagine a worst case scenario: the Herculaneum scrolls turn out to be just the works of this one mediocre pet philosopher. What would we still expect to learn from them, and what would the next step be?
Though I have an interest in Old Norse and I spend a lot of time reading Scandinavian runestones. > 90% of them are grave markers for a dead father, mother, brother, sister, cousin, etc. If I've learned anything from that, it's that people across time and space all lead lives as real and complex as anyone else's. Their joys were as high as mine have been and their sorrows as low as mine have been.
How do get to do that? As in what did you study to get the prerequisite knowledge, and how did you find this particular job? When I see interesting jobs I'm anyways curious what path lead there
I am a computer scientist. I studied CS in university, worked in the semiconductor industry for a while, got started as a participant in the challenge aspect of the Vesuivus Challenge. They were hiring, I sent in an application, interviewed, and was offered the job.
30 scrolls, maybe? Something like that. I scanned Pherc Paris 4 and Pherc Paris 3 at Beam line 18 at ESRF back in March.
The team did "the campfire scroll" experiment a few years ago to replicate carbonization, unrolling, and ink detection. That is the only case I am aware of. It proved the method could work but it's not a source of say training data; it varies too much from the real scrolls.
The main limitation is time and cost. We have to scan on what is AFAIK the most powerful x-ray beam line in the world. It is not cheap
You had to pay? I understand the machine cost many hundreds of millions of dollars, but I would have thought for academic researchers doing open science, the beamtime is free (funded by the govt / science trusts).
The beam time is unfortunately not free. I scanned Pherc Paris 4 and Pherc Paris 3 in March and had the final shift on the beam. As I was removing the scroll from the scanning pedestal the next team of scientists were already in the lab getting their samples ready. It's a well oiled machine and they've got customers.
The way these things normally work is that the project starts with some sort of a grant. Then that grant pays for all of the costs of the project: peoples' salary, materials used, time on equipment, plus money for the buildings and administration (overhead).
In this case the time on the equipment would need to be included, both a portion of the cost of building/maintaining it, and probably the energy needed to run it. Even where the government is providing the grant (likely here), it still needs to be accounted for.
That varies greatly on the state of preservation of the scroll. For some of the scrolls we can recover entire columns of text. But this is a best case. Plenty of scrolls, or portions of scrolls, are extremely damaged and warped to where our current methods cannot unroll them through any combination of automated and human driven unrolling. Both of these still have massive headroom for improvement, but achieving that headroom is hard as the preservation gets worse.
To give numbers, for ideal portions of scrolls, we can read 100% of the characters. In nonideal portions of scrolls, we can read 0% of the characters. It's not really possible to quantify how much we could theoretically recover of that 0% through better methods, and how much is truly destroyed.
BS in CS from a big state school in the USA. I have a hobby interest in history. I learned about the challenge on YouTube. Got involved contributing because I needed money. Then they put out a job posting. I applied, interviewed, and was hired.
It has gotten harder, unfortunately. One of the barriers to entry is simply the massive amounts of data; not everyone can set aside $100s worth of HDD or SSD space to play around. That said I have done a lot of work to dramatically reduce the amount of storage and bandwidth needed.
We unfortunately get a lot of slop submissions, which is unfortunate. I think a _really_ good place to start is simply joining the discord and looking at the data we've published and trying to replicate something or anything really. We understand that not everyone is a researcher that can jump in making awesome immediately applicate submissions.
Granted, that's pretty specifically for people that want to submit for prizes and prize money. Everyone on the team absolutely loves to talk shop and interact with real people with real interest, so if you show it in the discord we are all more than happy to help, engage, fix bugs, gvmive advice, etc.
I would personally love to see more open source and contributed papyrology and translation, musing on difficult readings etc.
For the more technically inclined, testing software, pointing out bugs, and actually running and trying to fix things is a huge positive that we like. We get a lot of slop submissions that are just someone pasting an issue on our GitHub into codex or Claude. We don't want to encourage that. We can do that ourselves.
That's a tough one to give a strong estimate of. Some scrolls are easier or harder to unwrap and read for a multitude of different reasons, mostly due to how damaged the scroll was in the eruption, and how easy or not the ink is to read. IIRC from what we've scanned of the herculaneum collection, none of the ink is easily visible via spectrum alone, so we have to use a lot of ML and physically based rendering techniques to be able to find ink. That also requires unwrapping and segmentation _before_ any ink detection.
For iron gall ink with high enough iron concentration, the ink stands out in the xray volume through simply masking off low values, such as was shown in our campfire scroll experiment a few years ago. No herculaneum scrolls show similar ink.
Most of the evidence so far points towards carbon based ink. I am not sure if any of the scrolls we have scanned show strong evidence of iron gall based ink. I know that there are different types and preparation methods for different carbon based inks, but I do not know if it is possible to determine which kind(s) were used solely from inspecting the xrays.
I am, though, not a papyrologist, so historical ink making, preparation, and usage are not my field.
That have been dug up? I think 600 or so still exist. Perhaps about 2000 or so have ever been excavated. We have scanned about 30 of them. Still underground? I've seen various counts. Maybe more than 10000?
Shame there's a modern city over most of Herculaneum. I'd love to excavate the remainder. Now that we can read what we find, there's a good scientific reason to do so now instead of waiting.
I am unaware of those fragments in particular. Though we have scanned a dozen or so fragments, mostly to help guide ink detection, since the ink in them is often more visible in visible and/or near IR light, but can be hard to impossible to detect in the xray spectrum.
Did anyone on the team come from a non-science, non-math, non-academia background? Did anyone working on this just teach themselves and start contributing?
Yes. Sean, who was a co-winner of the 2024 prize, IIRC has no formal background in ML, computer science, AI, etc. He is one of our core researchers and the most productive team member.
I've been on the Discord for a couple of years now, and poking around with submissions as well. Sean and the entire team deserve so much praise for all of this work.
It's easy to just read about the breakthrough and see it as one neat, linear line to get there, and hard to comprehend the hours, months and years that so many spent to get there. Big congrats to you, Sean, Nat and the entire team!
You have a potential to rewrite the history of European Antiquity quite substantially. The Herculaneum set of scrolls is enormous and must contain a lot of hitherto unknown.
That comes with a set of peculiar risks. Once your work starts producing something that contradicts previous work of Very Important People, they will lobby to stop you. Be prepared for that.
Science should be neutral and always value new evidence. Scientists as humans are unfortunately subject to all sorts of passions.
We have very little written material surviving from Rome, at least from the period before a codex (book) was invented, which was more durable that a scroll. Often, we only know of one source describing important events, and when it comes to political struggles and civil wars, the perspective of the defeated party often did not survive. The punishment of damnatio memoriae was practised and even among the early emperors, Caligula and Nero were subject to a form thereof. (This library in Herculaneum was buried 11 years after Nero's death.) I would be surprised if everything in the scrolls perfectly aligned with the record that survived for 2000 years and that was filtered by both random chance and political/religious censorship. Even Christians later destroyed some pagan texts.
BTW personally, I would love for some textbook of Etruscan to emerge from there. This was once again a language whose teaching was banned in Rome.
