fireflies_

This is where I started getting that feeling:

The curse of dimensionality is not theoretical — it’s painfully practical. In high dimensions, naive distance metrics degrade quickly.

Scaling and normalization are not optional details. They fundamentally shape the geometry of the space.

Arguably XSLT

This is "relative color" syntax, it works with a range of color spaces/color functions. The key is the "from" at the front. Here's the MDN documentation: https://developer.mozilla.org/en-US/docs/Web/CSS/CSS_colors/...

In one case, we ran a simple a/b test as a proof of concept on whether to pursue the idea further and it added an extra million bucks a year in revenue.

I'm with the people who decided to ship this. The organization will need to fund more maintenance than they would if they waited, but that has real costs. And "keep your 1mm/revenue idea to yourself" doesn't sound like a healthy engineering culture either.

Not sure if there is some way to iterate given online feedback.

There have been a some novels and novel-length works written a chapter at a time in public. The Martian by Andy Weir, for example is probably the most commercially successful. Many works by qntm (https://qntm.org/fiction) that are popular in this community have been written that way too.

I just don't understand how they still have users.

Because this post is here and not somewhere else. Strong network effects.

This application of "base rates thinking" was helpful for me - a bit out of left field, I'm aware of the concept and use it sometimes but wouldn't have thought to do so here. Thanks!

Thanks for sharing. I've used megaparsec a little but never looked at ReadP. Your tutorial is concise and helpful, I'd have ended up with better intuition faster if I started there.

It's a shame that these aren't parallel. Why couldn't it be "insular gigantism" or "island dwarfism"? (But not both!)

Would you be equally happy if you were sick enough to be hospitalized as you are when you're healthy? If not, that's a cost. I doubt the out-of-pocket medical costs are the most important part of this decision.

This doesn't address all of your comment but specifically with patient confidentiality – Israel agreed to share more data with vaccine manufacturers in order to get quicker access to the vaccines. There are obviously problems with bilateral dealmaking and Israeli distribution (the article below points them out), but Israel almost certainly saved lives by weakening confidentiality. I think that was the right call.

https://www.pbs.org/newshour/science/israel-trades-pfizer-va...

In some ways the argument against Excel is like the argument against Electron: if you're comparing it to an elegant, purpose-built and well-tested application, yes... it falls woefully short.

But is that really the right comparison? Lots of people just don't have the skills or time to build a specific applications. Without Excel, what would they do? There are some places where a 90% solution is worse than no solution at all (at least no solution is a forcing function for a "real" application), and in many ways Excel isn't the best expression of the _idea_ of an Excel-like, low-barrier-to-entry declarative programming environment with a built-in UI, but it's truly a wonderful tool. It makes a lot of automation possible for a lot of people.

You're being downvoted, but I think this sentiment is right and if you take it seriously, it's an argument for the SAT. Pretty much _everything_ in the USA is pay to win, not just the SAT, and the more subjective factors like essays, extracurriculars, and even grades, are more biased.

Test prep companies have an incentive to overhype their services but research suggests the improvement isn't that much.

There's a Jacobin article making the case for the SAT that people following this debate might enjoy reading: https://www.jacobinmag.com/2018/03/sat-class-race-inequality...

Yes, you can can replace that with the magic number 49 if you want, ord('1') == 49 because ord returns the unicode code point for a character.

Documentation is here for the curious: https://docs.python.org/3/library/functions.html#ord

Sounds like this only works if people who listen to smaller bands/indie music listen to fewer songs on average. I'm not sure if this is true. People who find themselves exploring niche tastes are probably "enthusiasts" who do more listening and listen more broadly.

I'm a bit surprised to hear it put this way. Aren't physics students usually welcome in finance as quants?

A probability has to lie between 0 and 1 but not every value between 0 and 1 is a probability.

It has, Safford Unified School District v. Redding is probably the canonical case here. https://en.wikipedia.org/wiki/Safford_Unified_School_Distric...

It isn't thrown away, though. If your district picks a third party, you send a third party representative to Congress. You'll have different choices in district-specific and state-wide races but you could also have a third party senate candidate and no realistic third party house candidates (for example Maine in 2018, though they had ranked choice voting). I don't see how having multiple levels of elections changes things beyond larger populations making it harder to coordinate outside the parties.

But all the prior blocks are static, that's the point. If you can break the cipher used at some time t, anything that people had "stored securely" by encrypting it and putting it on the blockchain before time t is vulnerable. As our technology and understanding gets better, t might get later and later. Obviously it depends on your use case, but it's a good argument for not putting sensitive things out in public on a blockchain at all.

You probably wouldn't see it in "natural" use but it's a potential attack vector if you ever deserialize arbitrary JSON (e.g. JSON bodies in API requests). Of course you should have general limits that would catch these things anyway...

Me too. I can't imagine trying to maintain a project written like this so I'd still consider you a member of the "clever programmer club" in good standing. Ideas like this make sense when the language gives you good support for them like Haskell does:

https://www.haskell.org/tutorial/functions.html

In languages like Go, you'll write much more "composable" software by sticking to what the language gives you instead of trying to force this in.

Probably a rewrite – HNers always seem to have stories about being brought in to work on companies with incomprehensible rats nests of code that are impossible to extend any further. I bet that's pretty common, more common than technical debt actually forcing the company under.

Still if you really want to take the analogy further declaring technical bankruptcy could cause business bankruptcy (see Joel Spolsky). You probably want to refinance your debt and get onto a repayment plan (an incremental rewrite) instead.

Do you have any resources about this? It sounds like a fascinating story.

Your code is doing something pretty different from the examples above. You're just calculating the nth fibonacci number, not creating an infinite fibonacci _sequence_. Eliding the explicit recursive structure is the point of the exercise!

Admittedly Python's facilities for doing that with some mutable state are pretty nice:

  from itertools import islice

  def fib():
      a, b = 0, 1
      while True:
          yield a
          a, b = b, a + b

  list(islice(fib(), 10)) == [0, 1, 1, 2, 3, 5, 8, 13, 21, 34]

Scott Alexander at SlateStarCodex makes some good points about this effect here[1]. The problem isn't the lack of moderation per se but the dynamics of starting an unmoderated alternative to an existing service. The kinds of people that would switch are only occasionally principled advocates of free speech. Most of the time, they're the people whose communities were banned on the moderated platform -- everyone else ignores your new, smaller entrant and continues on the moderated platform where their friends are. You end up with a just the hateful people. A decentralized platform would probably be better and have better social dynamics if it were the default and _everyone_ used it, but I don't know how you'd get there.

[1] https://slatestarcodex.com/2015/07/22/freedom-on-the-central...

Yes. Lambdas in Python are awkward, but you can definitely use either a lambda or a named function.

In fact, you could define a number of useful precondition functions in a single module and use them throughout a project. A couple of higher-order functions could make the post's examples safer and nicer-looking too. For example:

    @precondition(starts_with('The year is '))

    def starts_with(param):
        def test(s):
            return s.startswith(param)
        return test