3 ms·

If we wanted to digitize all ancient texts, what would it take? How many are there? Starting at the beginning of written history ~5,200 years ago to ~2,000 year

by mmooss 2y ago

If we wanted to digitize all ancient texts, what would it take? How many are there? Starting at the beginning of written history ~5,200 years ago to ~2,000 years ago, how many texts do we have from each millenia or century? Are they accessible to be digitized? What about earlier symbols and similar potential communication? For example, caves sometimes have marks that seems to have some symbolic meaning, such as slashes that seem like counts. How many are there to digitize? I'm curious about the scale of it.

jdthedisciple 2y ago
...and more importantly, could we then leverage LLMs to fill in the blanks of ancient history?
- mmooss 2y ago
  What blanks, and how could LLMs accurately fill them?
- empath75 2y ago
  You could use LLMs to give you rough passes at a translation of the texts probably, but they're pretty bad at catching the interesting/surprising details.
observationist 2y ago
More than $5 million, probably less than $100 million, probably - would need a team of qualified experts to know what to photograph and record, a team trained to handle delicate and valuable artifacts, IT people, machine learning experts, gophers, grunts, and finally a team of people to work with institutions. It's not a super complex endeavor, logistically speaking. I don't think it'll ever happen. Some institutions would resist being part of any third party scaled up attempt to record things, others would demand editorial control, and the most asinine petty politics would gum things up.
wongarsu 2y ago
Specifically for cuneiform, there are multiple efforts, covering roughly the time period you mentioned. There is this post, [1], and [2]. There are probably around 1-2 million cuneiform tablets that have been found so far. Many of them complete, but even more of them as fragments. Those fragments mostly just sit around in store rooms. It's a giant puzzle few people even have enough knowledge to attempt. 1: https://cdli.mpiwg-berlin.mpg.de/ https://cdli.mpiwg-berlin.mpg.de/ 2: https://oracc.museum.upenn.edu/ https://oracc.museum.upenn.edu/
- mmooss 2y ago
  Great, thanks. I've run across ORACC before, but I didn't know widespread digitization was part of their aim (i.e., I had the impression it was a database of artifacts digitized elsewhere). > There are probably around 1-2 million cuneiform tablets that have been found so far. Many of them complete, but even more of them as fragments. That's a pretty amazing number - that so many have survived. It seems like it would take too much luck, unless some library has endured this long.
  - thaumasiotes 2y ago
    Those tablets are indestructible once you expose them to fire. The records come from sacked cities. There's no shortage.
    - wongarsu 2y ago
      If you wanted to design a writing medium that enables archeology, it would be difficult to do better than clay tablets. Fired clay is great for records meant to last, being very fire and water resistant, while unfired clay was reusable if you just made it a bit wet again, great as reusable writing surfaces. So many paper, parchment and papyrus records that were meant to be kept have been lost to fire and water over the millennia. Yet with clay tablets a fire not only struggles to harm the records that were meant to last, it turns all the temporary records, scratch pads and notes left around into long-lasting fired clay. Burn a Sumerian city to the ground and you create a snapshot of all their writing for archeologists to find millennia later

by mmooss 2y ago

jdthedisciple 2y ago

...and more importantly, could we then leverage LLMs to fill in the blanks of ancient history?

mmooss 2y ago
What blanks, and how could LLMs accurately fill them?
empath75 2y ago
You could use LLMs to give you rough passes at a translation of the texts probably, but they're pretty bad at catching the interesting/surprising details.

observationist 2y ago

More than $5 million, probably less than $100 million, probably - would need a team of qualified experts to know what to photograph and record, a team trained to handle delicate and valuable artifacts, IT people, machine learning experts, gophers, grunts, and finally a team of people to work with institutions. It's not a super complex endeavor, logistically speaking. I don't think it'll ever happen. Some institutions would resist being part of any third party scaled up attempt to record things, others would demand editorial control, and the most asinine petty politics would gum things up.

wongarsu 2y ago

Specifically for cuneiform, there are multiple efforts, covering roughly the time period you mentioned. There is this post, [1], and [2]. There are probably around 1-2 million cuneiform tablets that have been found so far. Many of them complete, but even more of them as fragments. Those fragments mostly just sit around in store rooms. It's a giant puzzle few people even have enough knowledge to attempt. 1: https://cdli.mpiwg-berlin.mpg.de/ https://cdli.mpiwg-berlin.mpg.de/ 2: https://oracc.museum.upenn.edu/ https://oracc.museum.upenn.edu/

mmooss 2y ago
Great, thanks. I've run across ORACC before, but I didn't know widespread digitization was part of their aim (i.e., I had the impression it was a database of artifacts digitized elsewhere). > There are probably around 1-2 million cuneiform tablets that have been found so far. Many of them complete, but even more of them as fragments. That's a pretty amazing number - that so many have survived. It seems like it would take too much luck, unless some library has endured this long.
- thaumasiotes 2y ago
  Those tablets are indestructible once you expose them to fire. The records come from sacked cities. There's no shortage.
  - wongarsu 2y ago
    If you wanted to design a writing medium that enables archeology, it would be difficult to do better than clay tablets. Fired clay is great for records meant to last, being very fire and water resistant, while unfired clay was reusable if you just made it a bit wet again, great as reusable writing surfaces. So many paper, parchment and papyrus records that were meant to be kept have been lost to fire and water over the millennia. Yet with clay tablets a fire not only struggles to harm the records that were meant to last, it turns all the temporary records, scratch pads and notes left around into long-lasting fired clay. Burn a Sumerian city to the ground and you create a snapshot of all their writing for archeologists to find millennia later