4 ms·

Am I the only one who read "Type I and II decisions" and thought "System I and II decisions" (from Thinking, Fast and Slow?). I like the idea of distinguishing

Am I the only one who read "Type I and II decisions" and thought "System I and II decisions" (from Thinking, Fast and Slow?). I like the idea of distinguishing reversible from irreversible decisions, I just wish he'd come up with better names...

SilasX 11y ago
1 vs 2 is a bad, non-mnemonic naming convention to use, for either of those cases. Sadly, it also appears in "type I vs type II" errors. Even "false positive vs false negative" isn't helpful, since the decision of what counts as "positive" is arbitrary. Better to refer to the category directly: eg, "false ham vs false spam", "false acceptance vs false rejection". While we're on the topic Amazon and bad naming, what's with AWS EC2 instances? They have "spot" vs "on demand" instances, which have very different properties, ideal for different use cases ... and yet, in the context of economics and markets, the terms are basically synonyms!
- bonzini 11y ago
  I'm not sure about "false acceptance vs false rejection" being better than "false positive vs false negative". False positive pretty much always means you give incorrect bad news; false negative means you give incorrect good news. "Acceptance" may mean either "ham" or "spam" depending on the context.
  - rwjwjuwjudf 11y ago
    Positive means you found the special thing you are looking for; negative means you just found an ordinary thing. False positive means you thought you found the special thing but you didn't, it was just ordinary; false negative means you thought you found an ordinary thing but actually it was special. False positives mean you throw away ham as spam (terrible), false negatives mean you let some spam into your mailbox (not a big deal). But false positives also mean you diagnose benign as malignant (annoying), and false negatives also mean you diagnose malignant as benign (fatal). So depending on what you are looking for, it varies whether false negatives or false positives are worse. Or maybe they can be equally bad. I don't think you need to rename positive and negative as long as positive is understood as unusual and negative is understood as normal.
    - SilasX 11y ago
      But spam is more "usual" than ham, so ...
      - rwjwjuwjudf 11y ago
        Hmmm, good point. Ok, if you count individual messages, there is more spam, but if you look at the kinds of spam messages, they are way less diverse. So spam goes into one of ~10 buckets (419, pharmacy, fake conferences, view my profile, etc.), but the rest goes into 1000 different buckets. But you still have a good point.
srt146 11y ago
I wouldn't read into it too much. Amazon has adopted every stupid management fad and religion under the sun from six sigma to kaizen to lean to 5s to agile, etc. The idea being that if you are an expert on something totally made up you must deserve that big fat raise and promotion. I feel sorry for all the people who are going to have to go through Bezos method decision making training (black belt omega critical OT level VIII version).
- huac 11y ago
  Amazon has a history of hiring MBA's - they hired a ton from HBS before the crash - when other tech companies didn't. Maybe it's Bezo's Wall Street background.

SilasX 11y ago

1 vs 2 is a bad, non-mnemonic naming convention to use, for either of those cases. Sadly, it also appears in "type I vs type II" errors. Even "false positive vs false negative" isn't helpful, since the decision of what counts as "positive" is arbitrary. Better to refer to the category directly: eg, "false ham vs false spam", "false acceptance vs false rejection". While we're on the topic Amazon and bad naming, what's with AWS EC2 instances? They have "spot" vs "on demand" instances, which have very different properties, ideal for different use cases ... and yet, in the context of economics and markets, the terms are basically synonyms!

bonzini 11y ago
I'm not sure about "false acceptance vs false rejection" being better than "false positive vs false negative". False positive pretty much always means you give incorrect bad news; false negative means you give incorrect good news. "Acceptance" may mean either "ham" or "spam" depending on the context.
- rwjwjuwjudf 11y ago
  Positive means you found the special thing you are looking for; negative means you just found an ordinary thing. False positive means you thought you found the special thing but you didn't, it was just ordinary; false negative means you thought you found an ordinary thing but actually it was special. False positives mean you throw away ham as spam (terrible), false negatives mean you let some spam into your mailbox (not a big deal). But false positives also mean you diagnose benign as malignant (annoying), and false negatives also mean you diagnose malignant as benign (fatal). So depending on what you are looking for, it varies whether false negatives or false positives are worse. Or maybe they can be equally bad. I don't think you need to rename positive and negative as long as positive is understood as unusual and negative is understood as normal.
  - SilasX 11y ago
    But spam is more "usual" than ham, so ...
    - rwjwjuwjudf 11y ago
      Hmmm, good point. Ok, if you count individual messages, there is more spam, but if you look at the kinds of spam messages, they are way less diverse. So spam goes into one of ~10 buckets (419, pharmacy, fake conferences, view my profile, etc.), but the rest goes into 1000 different buckets. But you still have a good point.

srt146 11y ago

I wouldn't read into it too much. Amazon has adopted every stupid management fad and religion under the sun from six sigma to kaizen to lean to 5s to agile, etc. The idea being that if you are an expert on something totally made up you must deserve that big fat raise and promotion. I feel sorry for all the people who are going to have to go through Bezos method decision making training (black belt omega critical OT level VIII version).

huac 11y ago
Amazon has a history of hiring MBA's - they hired a ton from HBS before the crash - when other tech companies didn't. Maybe it's Bezo's Wall Street background.