3 ms·

Here is the quotes database of one of the older projects this is (potentially) based on: https://github.com/JohannesNE/literature-clock/blob/master/litclock_an

• gwern 2y ago
  13,000 seems surprisingly small, all things considered. Seems like the sort of task where you could automate it with a small LLM instance and process millions of books easily. Then extract all of the key strings and have a set of tens of thousands of ways to describe time which can be searched for as fixed-string matches, and might be of interest in its own right. (How do mentioned times change over history? Do they get more precise? Do books take place at wider ranges of the day?)
  • bigiain 2y ago
    There's only 1440 minutes in a day. So there's almost 10 quotes for each possible minute to display (assuming an even enough distribution). Even if you check the time a lot - I guess I probably check the time 3 or 4 times every day between 9:20 and my regular 9:30am meeting - it'll probably be ~15 days before I hot a duplicate quote. That feels like "enough" to me?
    • gwern 2y ago
      I'm not sure your math there on the birthday paradox is right (you only check the time during a single 10 minute range every day, no matter what? and you wouldn't look at it out of curiosity or for the pleasure of seeing a new quote?), but 15 days to the first collision doesn't seem like that long to me, for a piece of furniture that one would expect to be good for years to come and which costs $200+. And note that the number will go up rapidly from whenever the first expected collision is: only 10 quotes per minute means you're going to see a lot of duplicates after a few hundred days.