4 ms·

Theres no technical limitation preventing books online from either having an associated tags system to reproduce locality. Its more that online literature colle

by zanny 7y ago

Theres no technical limitation preventing books online from either having an associated tags system to reproduce locality. Its more that online literature collections are hugely fragmented due to IP policy and are often treading on landmines trying to associate someone elses work in a novel way.

imglorp 7y ago
Seems more akin to a user-adjustable recommendation algorithm like Amazon or Netflix uses. Sometimes you want "what other users looked at after searching for this book," or "books with many similar citations," or "books with a similar wordvec distance."
- majormajor 7y ago
  There serendipity encountered in the stories in this thread is essentially just different on one axis: it's not "personalized" or otherwise user-aware, it's content-driven active organization and presentation. That system can have its own ways of breaking down - consider trying to curate a collection of books on a topic you don't know about - but it gets at the limits of our current algorithmic approaches. Our systems don't know the content itself, just the feedback various others give about the content.
  - cameronbrown 7y ago
    I think a future function of the librarian will be virtual curation more and more (tagging, categorisation, etc..). Similar to what the archive.org people are doing.
mattkrause 7y ago
In principle, sure, but in practice... The bandwidth of a shelf is incredibly high. You can review hundreds of titles per minute. Along with the actual title and author data printed on the spine, the book’s condition gives you some hints about its age and popularity. There’s no friction if you want to pause and examine one title more deeply, and you can trivially pick up where you left off once done. The physical arrangement of books also papers over any deficiencies in the metadata, so that you’ll probably find something relevant, even if its title and keywords don’t match your initial expectations. It’s obviously not perfect: you need to compare books yourself, since reviews or circulation data are rarely available, especially right at the shelf. Nevertheless, I’ve never seen a web interface that was even close to as efficient as a library or bookshop.

by zanny 7y ago

imglorp 7y ago

Seems more akin to a user-adjustable recommendation algorithm like Amazon or Netflix uses. Sometimes you want "what other users looked at after searching for this book," or "books with many similar citations," or "books with a similar wordvec distance."

majormajor 7y ago
There serendipity encountered in the stories in this thread is essentially just different on one axis: it's not "personalized" or otherwise user-aware, it's content-driven active organization and presentation. That system can have its own ways of breaking down - consider trying to curate a collection of books on a topic you don't know about - but it gets at the limits of our current algorithmic approaches. Our systems don't know the content itself, just the feedback various others give about the content.
- cameronbrown 7y ago
  I think a future function of the librarian will be virtual curation more and more (tagging, categorisation, etc..). Similar to what the archive.org people are doing.

mattkrause 7y ago

In principle, sure, but in practice... The bandwidth of a shelf is incredibly high. You can review hundreds of titles per minute. Along with the actual title and author data printed on the spine, the book’s condition gives you some hints about its age and popularity. There’s no friction if you want to pause and examine one title more deeply, and you can trivially pick up where you left off once done. The physical arrangement of books also papers over any deficiencies in the metadata, so that you’ll probably find something relevant, even if its title and keywords don’t match your initial expectations. It’s obviously not perfect: you need to compare books yourself, since reviews or circulation data are rarely available, especially right at the shelf. Nevertheless, I’ve never seen a web interface that was even close to as efficient as a library or bookshop.