Oh no, you!

  • 0 Posts
  • 11 Comments
Joined 2 days ago
cake
Cake day: November 3rd, 2024

help-circle





  • BuT tHeY wErE WrOnG iN 2016!1

    Yes, and no. They estimated a slightly higher chance for a Hillary win over a Donald win, but they were well within the margin of polling error, and they have been for every election. Plus people have a tendency of over-valuing a “51% chance to win”.

    While this is good news, it could mean nothing.

    EDIT: 538 explained it better than I ever could:
    "Statistically, too, there is no meaningful difference between a 50-in-100 chance and a 49-in-100 chance. Small changes in the available polling data or settings of our model could easily change a 50-in-100 edge to 51-in-100 or 49-in-100. That’s all to say that our overall characterization of the race is more important than the precise probability — or which candidate is technically ahead.”





  • Price: Don’t remember. I fired off a message to the guys in manufacturing, I’ll let you know if they have a price. We go through quite a few of these annually, so we have a bulk discount when buying them.

    How: Magic, I guess. Also, I found a link to the ones we use: https://www.microchip.com/en-us/products/clock-and-timing/components/atomic-clocks/embedded-atomic-oscillators/csac

    When we prepare for deployment, they’re connected to a docking unit which provides a clock pulse derived from a GPS signal for high precision. Then the battery pack is attached, the subsea unit is assembled, and everything is deployed, usually via ROV. It’s important that it’s kept powered the entire time upon retrieval docking, as we can then calculate a linear drift value and correct for this in the recorded data.

    UPDATE: We bought “Maaaaaaany thousands” of them in 2018, and we paid 1850 USD per chip. Come to think of it, I remember hearing claims back then that we’d bought 75% of the worlds supply.


  • Some work related trivia I’d like to share with you: My job involves deploying sensors on the seabed to record data. These sensors are battery powered, and rely on highly precise timekeeping to be correct. These devices are synchronized before deployment, and upon retrieval ~2 months later we want none or as little clock drift as possible.

    Tumbleweed sounds

    I’m glad you asked: we achieve this by using CSACs - Chip Scale Atomic Clocks. They are pretty much what the name Implies, and after synchronization it is able to keep the time much better than anything else. Normal clock drift upon retrieval is usually less than a millisecond, and that drift is due to (uninteresting factors not directly related to the CSAC itself)

    CSACs cost a small fortune, but they’re the size of a matchbox and make it all possible. It’s amazing how small atomic clocks can be.


  • Each to their own I’d say, but I’m all for rule bending.

    I’ve always been up front about modding based on what I consider reasonable interpretation of the spirit of the rules combined with common sense. This means that some things that technically should or shouldn’t get deleted gets a treatment it wouldn’t get by strict interpretation of the rules in their literal meaning.

    That being said, my interpretation of the rules may not necessarily conform with those of other mods, so I rarely complain if someone else decides to overrule me.