10 ms·

Turns out "technical debt" also applies to national archives.

by mpfect 3mo ago

jfengel 3mo ago
More than you can possibly imagine. There are warehouses full of unread papers. Any one of which could contain a reference to somebody or something important. There was a recently discovered letter, possibly to Shakespeare's wife, which would completely change our understanding of their marriage, and even the way his plays depict women. The only way to find such things is by hordes of grad students trudging their way through fragile paper and messy handwriting.
- mmooss 3mo ago
  I hate to say it, but might LLMs transform archival work? Not by replacing researchers, but by inputting everything (or orders of magnitude more than we could previously) and outputting to the researcher a prioritized list of documents / etc to examine?
  - computerdork 3mo ago
    Oh, wow, that is actually an interesting application of ai
  - jfengel 3mo ago
    If you could automate transcription, it would be an enormous boon to researchers. Reading the handwriting would be really hard, and it would be a massive effort to move all that paper. Just handling it is hard; it's not like flipping through mass-manufactured books. But I suspect that you could spend a few million dollars to revolutionize the field.
    - order-matters 3mo ago
      >automate transcription this also means trusting the LLM to decide what things mean. but there is very likely a great middle ground of having LLMs take their best guesses and then verifying the output on significant finds. the risk is in LLM understating something important, false negatives, leading to putting stuff at the bottom of the pile that appears mundane but isnt
      - mmooss 3mo ago
        That's why I suggest the output would be a prioritized list of documents for the researchers to review; the LLM doesn't get the final say, it just makes recommendations. Yes, things would be missed, but the resesarchers might in theory find much more value than their current search method.
    - kevin_thibedeau 3mo ago
      This is already the case with genealogical sites that have ML OCR creating searchable indices of handwritten documents.
  - Digory 3mo ago
    I had ChatGPT translate some old, handwritten French legal documents for family history purposes. It was far more accurate than I expected. At scale, with better models, we might have a way to clear out the old archives. Not only could you translate, you could ask it to triage the discoveries. "Would the average person find this noteworthy?"
    - contingencies 3mo ago
      I have a ton of handwritten German stuff from the 19th century. My grandmother could make a fair stab at it, but nobody left can read it. I've shown modern Germans and they are at a loss. Thanks for your idea, I will give it a look. Any tips on model/method/training?
      - CamperBob2 3mo ago
        Try both Gemini Pro and ChatGPT. They are both outstanding at reading almost-unreadable documents. Use the highest thinking level your account supports. (If you want to post a sample or two here, I'll try it. I like to collect difficult out-of-distribution test materials.)
      - Digory 3mo ago
        This was about a year ago, and it was pretty rudimentary. I'd just drop in jpegs into ChatGPT and ask for a translation of small sections. If I were doing it today, I'd probably do it in codex or code, have it develop a plan, look for better tools, etc. Report back if you have success!
  - garlic_enjoyer 3mo ago
    Assuming they have been transcribed, yes. The key idea that makes LLMs special is the attention mechanism. Maintaining attention over volumes of data is boring for most humans. Also, to be pedantic, just taking about LLMs in this context is a tad reductive. There are many deep learning models involved in archival work that aren't language models. I encourage you to read into this post for more context on what I mean: https://news.ycombinator.com/item?id=48675179 https://news.ycombinator.com/item?id=48675179
  - lazyasciiart 3mo ago
    The bottleneck is physical work, as I understand it. And primarily delicate physical work that does not destroy the already disintegrating materials that are piled up in boxes for miles. https://www.aaa.si.edu/documentation/digitizing-entire-collections-chapter-4-for-scanning-technicians https://www.aaa.si.edu/documentation/digitizing-entire-colle...
contingencies 3mo ago
I found a copy of the oldest film ever shot in China in the School of Oriental and African Studies (SOAS) library in London. The camera had been personally loaned to the French administrator in question by the Lumiere Brothers. The film had been entered in to the catalogue but nobody had looked at it in decades and they didn't have equipment to do so. The university wound up digitizing it with funds donated by the alumni and I was invited on my return from the US to address the alumni association on my research.

Turns out "technical debt" also applies to national archives.

by mpfect 3mo ago

Turns out "technical debt" also applies to national archives.

More than you can possibly imagine. There are warehouses full of unread papers. Any one of which could contain a reference to somebody or something important. There was a recently discovered letter, possibly to Shakespeare's wife, which would completely change our understanding of their marriage, and even the way his plays depict women. The only way to find such things is by hordes of grad students trudging their way through fragile paper and messy handwriting.

mmooss 3mo ago
I hate to say it, but might LLMs transform archival work? Not by replacing researchers, but by inputting everything (or orders of magnitude more than we could previously) and outputting to the researcher a prioritized list of documents / etc to examine?
- computerdork 3mo ago
  Oh, wow, that is actually an interesting application of ai
- jfengel 3mo ago
  If you could automate transcription, it would be an enormous boon to researchers. Reading the handwriting would be really hard, and it would be a massive effort to move all that paper. Just handling it is hard; it's not like flipping through mass-manufactured books. But I suspect that you could spend a few million dollars to revolutionize the field.
  - order-matters 3mo ago
    >automate transcription this also means trusting the LLM to decide what things mean. but there is very likely a great middle ground of having LLMs take their best guesses and then verifying the output on significant finds. the risk is in LLM understating something important, false negatives, leading to putting stuff at the bottom of the pile that appears mundane but isnt
    - mmooss 3mo ago
      That's why I suggest the output would be a prioritized list of documents for the researchers to review; the LLM doesn't get the final say, it just makes recommendations. Yes, things would be missed, but the resesarchers might in theory find much more value than their current search method.
  - kevin_thibedeau 3mo ago
    This is already the case with genealogical sites that have ML OCR creating searchable indices of handwritten documents.
- Digory 3mo ago
  I had ChatGPT translate some old, handwritten French legal documents for family history purposes. It was far more accurate than I expected. At scale, with better models, we might have a way to clear out the old archives. Not only could you translate, you could ask it to triage the discoveries. "Would the average person find this noteworthy?"
  - contingencies 3mo ago
    I have a ton of handwritten German stuff from the 19th century. My grandmother could make a fair stab at it, but nobody left can read it. I've shown modern Germans and they are at a loss. Thanks for your idea, I will give it a look. Any tips on model/method/training?
    - CamperBob2 3mo ago
      Try both Gemini Pro and ChatGPT. They are both outstanding at reading almost-unreadable documents. Use the highest thinking level your account supports. (If you want to post a sample or two here, I'll try it. I like to collect difficult out-of-distribution test materials.)
    - Digory 3mo ago
      This was about a year ago, and it was pretty rudimentary. I'd just drop in jpegs into ChatGPT and ask for a translation of small sections. If I were doing it today, I'd probably do it in codex or code, have it develop a plan, look for better tools, etc. Report back if you have success!
- garlic_enjoyer 3mo ago
  Assuming they have been transcribed, yes. The key idea that makes LLMs special is the attention mechanism. Maintaining attention over volumes of data is boring for most humans. Also, to be pedantic, just taking about LLMs in this context is a tad reductive. There are many deep learning models involved in archival work that aren't language models. I encourage you to read into this post for more context on what I mean: https://news.ycombinator.com/item?id=48675179 https://news.ycombinator.com/item?id=48675179
- lazyasciiart 3mo ago
  The bottleneck is physical work, as I understand it. And primarily delicate physical work that does not destroy the already disintegrating materials that are piled up in boxes for miles. https://www.aaa.si.edu/documentation/digitizing-entire-collections-chapter-4-for-scanning-technicians https://www.aaa.si.edu/documentation/digitizing-entire-colle...

contingencies 3mo ago

I found a copy of the oldest film ever shot in China in the School of Oriental and African Studies (SOAS) library in London. The camera had been personally loaned to the French administrator in question by the Lumiere Brothers. The film had been entered in to the catalogue but nobody had looked at it in decades and they didn't have equipment to do so. The university wound up digitizing it with funds donated by the alumni and I was invited on my return from the US to address the alumni association on my research.