- This is where the term came from: https://x.com/karpathy/status/1886192184808149383?lang=en
- English to Japanese loses plurals, Japanese to English loses most honoriffics and might need to invent a subject (adding information that shouldn't be there and might be wrong). Different languages also just plain have more words than others with their own nuances, and a round trip translation wouldn't be able to tell which word to choose for the original without a additional context.
Translation is lossy. Good translation minimizes it without sounding awkward, but that doesn't mean some detail wasn't lost.
- There is no such thing as not signaling. By not using the turn signal, you are lying to anyone around that you might not see, signaling that you are going straight forward when you aren't.
- Not completely, I'm not logged in on my work laptop and it was only working some of the time (and not like some pages were cached and some weren't, I was refreshing the same page and sometimes it worked and sometimes not).
- > And anything done to make that profit will annoy the community.
I don't keep close track of this, but as far as I remember they haven't tried donations that go only to Firefox/Thunderbird/etc of the person's choice, instead of Mozilla as a whole. That's what people always claim they want in these threads. I doubt donations would be enough, but I think doing it like that would at least be a step in a direction people like instead of are annoyed by, as long as they don't go nagging like Wikipedia.
- An experiment on that from a year ago: https://www.hackerneue.com/item?id=42584400
- "Pants" (clothing) is a good one, means something different depending on location.
- Emptying into the dock instead of having to empty the robot's dustbin weekly and almost everything involving mopping in combined units is within that time range. Lidar mapping was also pretty rare a decade ago, Neato was the first and it took a while before others did it too, then there was apps for controlling no-go zones using those maps instead of variations of virtual walls, if they had anything like that at all.
Roomba was living off of name recognition for most of that period and was far behind in adopting any of it.
- > It’s really only suitable for lurking
If you're not just lurking, log in and reddit doesn't block you.
- When unions gain too much power and the company can no longer respond effectively to market forces (particularly with hiring/firing), leading to the collective ruin they were talking about.
- > I provided a range of leader:worker income/power ratios from 1:0 to 1:1 with no commentary whatsoever on outcomes
For a slight perspective change, the thing that leads to what they mentioned: They're saying you forgot about the range 1:1 to 0:1.
- This is far from the best way to do it, but this is a much easier to understand example of how it could be done without having to read about math:
There's a type of token called a JWT that's really common nowadays, which is composed of 3 parts: Metadata describing encryption for the third part, the actual base64-encoded data, and the encrypted signature. The second part would include "is over 18" and "expiration date" to limit reuse/abuse, and is trivially decoded by anyone to confirm there's no personal information in there.
You'd get this token from your government site and copy/paste it into the site needing verification. The government site would provide a standard public key that can be used with the third part of the JWT to confirm it hasn't been tampered with (verification is built-in to JWT libraries). There would only be one public key that rarely changes, allowing the site to cache it, preventing the government site from correlating users based on timestamps - they never see the JWT from the other site (verification is done locally), and the other site would only need to pull the public key once for however many thousands of people use it.
...that said technical issues aside, I kinda feel like this would be the most acceptable version simply because it doesn't require the average user to trust the math - they could go to a JWT-decoding website and look at it themselves.
- > and how hard it is in English to distinguish between "I have zero want for x" and "I have negative want for x"
"I do not want to X"
"I want to not X"
These are both pretty natural English constructions, though the second is usually used as a retort for clarification after saying the first but meaning the second.
- SwiftKey has this one where you can erase the wrong word and try to correct it, and it instead adds two words: the one you erased and the second attempt after it.
- When I told a co-worker about https://pypi.org/project/voluptuous/ he immediately searched for the name alone, got really wide-eyed and closed the tab, then told us not to do the same.
- "Mac OS through email" was what popped into my mind. No idea how that would work.
- > For example, I've seen people write code which relies heavily on design patterns but then that same code uses an O(n^2) nested loop to find items that are common between two arrays. There is a simple 'pattern' you can use to store the items of the first array in a Set or HashMap and then finding common items is O(n) because Set and HashMap lookups are O(1)... Very useful pattern but I don't believe it has a name. I use it literally ALL the time. The idea of storing intermediate state in some kind of HashMap is a game-changer IMO but there's no name for that pattern of coding.
This is a "hash join" in relational databases. You can see it in the query planner output of at least postgres.
- > The display property should be called display-type.
More importantly to me, "display" has been overloaded with two meanings: Display of the element this rule applies to/how it interacts with surrounding elements (none, block, inline, inline-block) and display of the contents of this element (flex, grid).
Which is why we now also have inline-flex and inline-grid.
Edit: Apparently we can now arbitrarily combine inline/block and flex/grid as two values to "display", no idea when this happened: https://developer.mozilla.org/en-US/docs/Web/CSS/Reference/P...
- > Absolutely-positioned replaced elements should stretch when opposite offset properties (e.g. left+right) are set, instead of being start-aligned.
They do with left+right, and have done this for a very long time. You only get the anchoring if you additionally set width.
I'm also wondering how much of what they come up with could be implemented as an addon instead of a core part of the browser.