It sounds useful to me because you can use tone information to help with the translation, which text-to-text translation can't do. But I'm not sure if that's how this model actually works.
Why am I unable to find a single good video on the internet demonstrating the mpemba effect occuring? If people were really heating their water before freezing it in Aristotle's time, then this shouldn't require any precise technology to reproduce. The article addresses the fact that this seems very easy to test, but then doesn't explain why no one's reproduced it on camera.
Because it’s bullshit. People who observed this supposed effect don’t even bother to place both containers in the same fridge at the same time (so it’s not affected by the thermostat’s timing) nor do they weigh the ice afterwards.
What happens if you do that with cold water? Also, there's no indication of the water freezing. How would you tell if the water is freezing or not once it's landed in the snow?
Yes, the only videos demonstrations involve throwing the water into the air, but nothing in this article or the Wikipedia page about the effect suggest that throwing the water is necessary for the effect to occur. I don't even know where these people got the idea to do the experiment that way. Their result is way less interesting because it can be explained by surface area.
What an interesting trick. A starting point to intuit a resolution: yes you have a 50% chance of doubling and a 50% of halving. But the doubling only happens if you have x dollars, and the halving only happens if you have 2x. So you can see you either gain x or lose x, so your expected return is 0.
When the amount you multiply your money by depends on the amount of money you currently have, you have to factor in your initial money to each case to compute expected return.
Oh, that actually really helped me getting a grasp on the problem. I feel like other comments and sections of the article were trying to get at that idea, but your phrasing of it made it finally click for me. Thanks.
Every time there's a post giving advice on making money or entrepeurship, the majority HN response is "but success is mostly luck" + "you need not want this kind of success". But when there's a post on, say, how to win at poker, people don't feel this need to bring up these points, even though they still apply at least as much. Somehow we're unable to accept the concept of picking a game and playing it when the game is getting rich or making an impact. I suppose it's because this game is so culture dominating that it affects those who don't play it too, as it makes everyone feel like they're supposed to be playing it, and resisting requires extra effort.
I guess this is fine and inevitable. But it makes me wonder where to go to find focused quality discourse about entrepreneurship / getting rich if not HN.
Haha, even less effective than I expected but enjoyed the read nonetheless.
I think if you have the luxury of assuming every token is a dictionary word, you can do much better by simply encoding each word as its index in the dictionary. Using the pidgeonhole principle, you can probably prove that this is at least as good as the autocorrect approach no matter how sophisticated the autocorrect is.
There once was a group of old friends who would get together on a regular basis for social interaction, telling stories and making jokes. As these folks aged, they ended up telling pretty much the same old stories and jokes again and again until they reached the point where they just assigned a number to each.
Decades later, they would gather and sit in silence for long periods until one or the other of them would just say something like "47" and all the other would laugh.
I’m not so sure it’s “at least as good” as sophisticated autocorrect. If you have a sufficiently advanced prediction model, it seems to me that you could make better guesses on context, which might, just by chance, outperform this approach. If you narrowed it down to only words that matches using this approach, but still used some kind of prediction based on context to pick which, I’d expect it to likely perform much better.
I don’t expect that would work. It would compress long words, but most short words would be encoded with numbers longer than the word themselves, and those typically make up most of the words.
The trick is to encode the numbers in binary, not plaintext. But still you'd probably want to use a Huffman coding rather than plain indexes so that the common words are shorter.
> I think if you have the luxury of assuming every token is a dictionary word, you can do much better by simply encoding each word as its index in the dictionary.
It's the standard pool/index tradeoff : instead of storing an array of possibly-duplicate objects, you simply store all unique objects once in a pool and store the array as an array of references into that pool.
You win if the original array of duplicate objects was so long or so duplicated that replacing it with references into the pool is a worthwhile reduction. The objectSize/indexSize ratio also plays some role.
The vast majority of text objects anybody generates won't exceed 4MB though. 4MB is 4 million ascii/utf8 characters, if we assume a typical word in English is 4 characters for simplicity, that's 1 million words without spaces. A quick Google search for "Novel that is 1 million words" yields the fact that this is twice the word count of Lord of the Ring and the word count of the first 3 A Song of Ice and Fire (Game of Thrones) books. Accounting for whitespace, longer common words, and inefficient encoding schemes would bring that overestimate down to, what, 150K words? a 300-page or 600-page book (depending on the spacing) according to Google, still massive.
I see it only working where there's massive pooling, like you say. An OS or a tool provides a known dictionary service and you call into it with a text string and get back an array of indices that you can then decode with another call. That amortizes the dictionary cost among all possible uses of the service, which is a lot if it's a standard well-known service. Another scenario is perhaps in a database\cloud storage\social media, any single text object might be small but the total text they store overall is massive.
Except that the corpus doesn't need to be stored as part of the compressed message, and can be considered part of the compression algorithm. It increases the size of the decoder by ~4MB, but doesn't increase the size of each message.
The autocorrect system might be able to do better if it had an understanding of grammar and proper english like GPT3, and could actually fill in missing words, replace words, etc
It'd be more accurate and illuminating to say that a px is an angle.
The reference pixel is the visual angle of one pixel on a device with a pixel density of 96dpi and a distance from the reader of an arm’s length. For a nominal arm’s length of 28 inches, the visual angle is therefore about 0.0213 degrees. For reading at arm’s length, 1px thus corresponds to about 0.26 mm (1/96 inch).
- https://www.w3.org/TR/css-values-4/#absolute-lengths
This explains why 1px is small on phones and large on televisions; the physical size varies by expected reader distance. The angle tries to stay constant, though in practice it gets rounded to the nearest multiple of the physical pixel size.
I think the main problem is that they're too attention grabbing. Especially in a UI in HN where everything else is plain text, an emoji would have so much visual weight your eyes would have trouble focusing on anything else on the page. Forbidding such visual grenades helps keep the focus on the content.
Ever since I learned that the frequency ratio between adjacent keys on a piano is always a constant twelfth root of 2 regardless of whether you're going black-white or white-white, I've been sort of frustrated with the way pianos are designed. Transposing a song to a different key could have been effortless if keys were structured differently, such as with uniformly alternating white and black keys, rather than the 2 black 3 black groups we have now. By optimizing pianos for C major / A minor, we've made everything else so much more difficult. And no other instrument does this afaik. String instruments don't obscure the constant ratios at all, so it's easier to transpose songs on them.
I suspect someone may reply to this with "it'd be harder to keep track of what notes you're on if keys were uniform like that", but I don't think we actually rely on feeling the keys to know where we are. Would be willing to be proven wrong on that. But I suspect simply coloring/shading each fifth and octave differently should do the job, maybe also texturing them differently for the blind.
Pianist here. I see where you're coming from, but I definitely do also rely on the different shapes of keys to know where I am without looking. I've been learning to play the button end of an accordion and personally I find the uniformity makes it harder. But maybe that's just what I'm used to.
As an aside, I find c major a horrible key to play in, again I think the uniformity makes it harder and less ergonomic, but the same thing makes it easier to teach to beginners hence why everyone starts there.
Another data point: I play the organ, which involves playing a large keyboard with your feet (https://en.m.wikipedia.org/wiki/Pedal_keyboard). The structure of black notes is crucial to being able to play without looking (which is important because you’re simultaneously doing various other things). Concretely, I couldn’t imagine playing the organ without being able to slam my foot into the right hand side of a B flat in order to know that my foot is on top of the B. If you watch an experienced organist’s feet you’ll see them do this all the time.
This is pretty eye opening to me, I would have never thought anyone found C major more difficult to play in than other keys. To me, difficulty is directly proportional to the amount of stuff in the key signature, but that may be because I've never had proper education (~5 years of lessons as a child from a very casual teacher).
The way we happen to notate music doesn't help to endear people to f sharp major, certainly! If you keep practising all the scales and try to learn to do more by ear/improvise, I'm pretty sure you'll end up enjoying other keys more.
It is so annoying that standard guitar tuning isn't all perfect fourths and we've decided to throw in a major third between the G-B strings. Bar chords be damned, I want to be able to visualize it symmetrically and have chord shapes that work the same across the fretboard.
As a longtime guitarist myself, I found it quite easy to learn.
You can think of each row as a guitar string. In the default setup, each row is a perfect fourth above the row below. The green lights are the white keys on a piano; the unlighted keys are the black ones. The blue lights are C natural.
Skip a square diagonally up and to the right for an octave.
Chord shapes are uniform across the whole surface.
Keys are sensitive to pressure, left-right movement, and top-bottom movement. In the default configuration, rocking left and right yields vibrato and moving top-bottom controls some other MIDI parameter (usually filter cutoff, but it's configurable).
Sliding finger pressure to different keys to the right and left gives portamento. Sliding to the next row up or down gives legato.
Once you get used to these parameters, it becomes super expressive.
It's a MIDI controller, not a complete instrument, so you need a MIDI-controllable source of sounds. It's an MPE instrument, so, if you want the full experience, you want some kind of MPE-capable source of sounds (MPE is "MIDI Polyphonic Expression", and it means that multiple MIDI channels are dedicated to the instrument in order to give full, separate expression to each of several voices). MPE is a relatively new MIDI development that is not supported everywhere yet, so you have to pay attention to which sound devices support it if you want the full experience.
I have one; it's my favorite electronic instrument that I've ever tried. I also have an Eigenharp Pico, and I like it, but I like the Linnstrument better. (It doesn't look as cool as an Eigenharp, though :-).
When I decided to get one, I discovered that they're a bit hard to buy. Not very many places have them, and when I was looking, even Roger Linn didn't have any available. I eventually found one for list price at Detroit Modular (https://www.detroitmodular.com/).
I watched the LinnStrumentalist video per your first link. Amazing stuff. And it does seem to reduce the amount of finger contortion required compared to most keyed or stringed instruments.
It does, though, seem to demand that the player look at the keyboard. For an instrumentalist, this may not be a big deal. For a performer accustomed to connecting to his audience with his face, as do many singers, it seems like a big drawback.
Yes, the Eigenharp is easier to play without looking, but I can say a couple of things about the need to look at the Linnstrument.
I've only had mine for a few weeks, but the need to look at it seems to be slowly disappearing. Now, the surface does feel really uniform, which is not so helpful for that. As you probably expect, there's not much on it that you can use to orient your hands by touch. Still, I do seem to be gaining some ability to know where my fingers go without looking. Partly it seems to be just acquiring a feel for how far apart the keys are (and you can feel the boundaries between keys). Part of it is starting to know which keys I'm touching from the pitches it's producing.
The second thing is that I've seen videos of players who attach a guitar strap and sling it over their shoulders (it comes with the attachments for that), playing it flat against their chests.
Taking those two things together, I'm fairly confident that it's possible to learn to navigate the keyboard by feel.
I haven't tried the over-the-shoulder thing, myself. So far, I play it on a table like a piano, or like a lap steel guitar or dobro.
I find it easy to learn, both comfortable, and expressive, and it's quickly become my favorite MIDI controller. I prefer it over both piano-style keyboards and MIDI guitars.
You also want open chords. There’s nothing stopping you from tuning the way you describe you’ll find it much harder to play harmony that’s musical if you do though.
I have one, in the form of a Chromatone keyboard (https://chromatone.jp/chromatone/index.html, no longer available). The uniformity does make it harder to play by touch, but it otherwise has a lot of advantages. I have been experimenting with different colours and tactile markers (little rubber feet intended to stick on the bottom of things) but I haven't settled on anything yet.
This is really interesting, thank you for sharing! I wonder if it would have been more successful if it had just two rows instead. The duplicate rows add more power, but at the cost of adding ambiguity when sightreading, as you have to impromptu figure out what fingering you want to use. And it also makes it more expensive and intimidating looking. I feel like two rows would provide the core benefits while being more accessible. But I could be totally off, as I've never touched this thing.
The more compact nature of the layout means you pretty much need more than 2 rows, or fingering some things seems to become very difficult. It feels like your fingers get tangled more easily if you stick to two rows (which you do at first, out of habit). I don't think it adds much to the cost (the different places to play the same note are all on the same physical lever), but it certainly makes it look intimidating, and the "playing by touch" issue is unsolved AFAIK.
I think the physical size of pianos is why the first decent design "won"; you don't take your own piano to where you're playing, you play whatever's there. In the electronic age there's a chance for Janko to take off, as that's not universally true any more, but I'm not holding my breath.
Transposing is effortless on most modern keyboard instruments: you press the transpose button. Now you have a choice of what key it sounds in and a choice of what fingering you find convenient.
Isn't that better than your suggestion, where you're locked into one fingering?
I don't think real pianos are tuned precisely like that, and the use of a dominant key may have been about historical trade-offs in tuning between different keys, but that's mostly a guess.
There's also the ergonomic aspects of key size and hand width.
Guitar and stringed instruments are easy to transpose because there are no "blessed" notes (other than the open strings, but even those can be turned nonstandard).
