That makes him about as fast as a 300 baud modem. And probably around 1.8x the fastest keyboard typists.
Edit: Curious if this as "optimized" as it can be. Is there a theoretically faster way to get words from your hands to something digital?
I'm also curious how this works for court cases where a language other than English comes into play, alongside English. Canada might be a good example, where French and English dialogue might often co-exist.
> I'm also curious how this works for court cases where a language other than English comes into play, alongside English. Canada might be a good example, where French and English dialogue might often co-exist.
The related Penti-Chorded keyboard apparently works fairly well for German [0], and Japanese Kanji [1]. I'd imagine that using the arpeggio to shift the characters would work similarly with the steno.
> Is there a theoretically faster way to get words from your hands to something digital?
For the sake of, err, discussion, I can envision some sort of wearable glove like input device. I know I can play notes on a guitar far faster than he is inputting characters because of the technique involved. I feel like it could be possible to come up with a faster method playing around with that idea.
But the ultimate would be hands free, no? Makes one wonder if this is yet another skill/job about to be blown away by automation.
Offline speech recognition with perfect accuracy will probably come standard with every phone in a couple of years.
Speech to text is interesting, though the 360 words / minute is already quite a lot faster than typical speech. Though probably slower than the text to speech blind users use (450 words/minute?).
I wonder if there's a blind stenographer that can really fly, but just isn't well known.
I bet there are some unknown talents in the closed captioning space.
Unrelated, I've seen some interesting typos there, especially with strong English accents like Scottish or Irish. And you can see them improving as episodes progress. The "RedRock" Irish soap is a good example. They get much better over time.
_humans_ are in the 5% word error rate ballpark for large vocabulary recognition. I think generalized perfect accuracy is very very far away unfortunately.
> I can envision some sort of wearable glove like input device. I know I can play notes on a guitar far faster than he is inputting characters because of the technique involved.
Can confirm. My uncle is a stenographer, I learned all about it.
If I remember correctly, just to get his license he had to hit 270w/m. He said he can get up to about 300w/m.
The raw transcripts are composed of shorthand codes, and letters that represent syllables and other components of speech. I saw that many words were just represented by one or a few letters. So, by "words", we are often talking about codes of only a few letters. They are printed on thermal receipt width paper, and floppy disk (this was about 15 years ago).
> I'm also curious how this works for court cases where a language other than English comes into play, alongside English. Canada might be a good example, where French and English dialogue might often co-exist.
It shouldn't matter what language it is, because the stenographer has their own custom shorthand dictionary. The raw transcript that they generate is for the stenographers internal use. Only after the trail do they transcribe it into human readable form. Lawyers then have to pay the stenographer a fee for their transcripts. My uncle told me that it is common for multiple lawyers to each pay him a fee that he charges by the page.
He, like most court reporters, outsources their transcription to other people via Email. It requires human transcription, because the raw transcripts need to be interpreted. It is not a something that you can just run through a filter. His scopists[1] will email him back and fourth for clarification on parts of the transcript they don't understand. This is important, because their transcripts are the official legal record of what happened in a trial.
Actually, it's more like the court reporter is the official witness. They are attesting to what happened in the courtroom. The actual instrument that the reporter uses is much less important. And trials have the audio recorded anyway.
I know that when my uncle started in the early 90s, there were still some court reporters that used shorthand with pencil and paper. I don't think there is a requirement that they use a steno machine. Their job isn't to record what happened, their job is to generate a single authoritative transcript that everyone works from. In theory, they could probably not take notes at all, and just work from the recording later, but it is probably easier to take the first round of notes live.
When a court reporter generates a transcript, they sometimes have to go back through the audio recordings sometimes and actually clarify in the official transcript exactly what was said. They have to make it clear in the transcript if people were talking over each other, or to try to clarify what exactly was said if it might be ambiguous.
I think it is a good idea for a trial to have an official record that has been processed by a human, instead of just throwing a recording up and calling it a day.
Think of how inefficient it would be otherwise, where all of the parties would essentially have to act as their own court reporters and generate their own transcripts.
What if different parties disagreed with the interpretation of the transcripts? Perhaps the court would then have to appoint a neutral expert to make an authoritative determination of what was said; and we've now re-invented the court-reporter role.
I think there is a tendency for people to look at this only as a technical problem to be solved. Like with voting, it isn't just a matter of securely tallying votes, there is much more to it. The process requires maximizing transparency and minimizing room for anyone to contest it.
As a teenager / college-aged kid, I provided "IT assistance" to family, friends, and friends of family/friends for a fee.
One of my best clients was middle-aged mother of three - let's call her Pat, because that's close enough - whose home IT setup always needed a lot of extra love. Weird viruses on old machines, ancient printers that had to be coaxed to work well when shared over a network, "what laptop should I buy my son for college" - the usual.
She was a court reporter, by trade. I had to help her with her laptop and court reporting software / hardware from time to time - it was very interesting. She had a custom "keyboard" that only had 10-15 unlabeled keys that really looked more like long and thin paddles than anything else; it worked with a program that recorded audio and tracked it to her inputs.
So she could wind the report back to any given point in time, and it would highlight what she'd typed as it went. So if she needed clarification on something - she could enable audio playback that showed what she had transcribed, not unlike a karaoke machine.
It was all very interesting stuff. I was a little sad when I had to tell her that I couldn't do work for her anymore, because I didn't have the time to do side work due to a developing career and social life.
There is a product called "LiveScribe" which is a pen for handwritten notes that also records the audio. It uses a special "GPS" paper with a location pattern in it and you can tap your note and it seeks back to the relevant audio. I think it also digitises the actual notes but I am not 100% on the exacts of that.
>I think it is a good idea for a trial to have an official record that has been processed by a human, instead of just throwing a recording up and calling it a day.
especially if someone points at someone else or uses other means of communication. Of course the judge can then say what was communicated so there is a verbal record in the recording but it is probably better that the court reporter is there in that case.
Transcripts are really important and they need to be proofed and edited by humans. However, they can be produced efficiently by sending digital audio recordings to a team of fast typists who can work remotely, work concurrently, adjust the playback speed, skip pauses, and pause and replay tricky sections. These days, automated transcription software can further accelerate the process by generating a very rough draft. I don't think stenographers are used to produce court transcripts outside the United States.
I don't see how that's any faster or cheaper. All stenographers are is fast typists. So I honestly don't even know if you're describing a distinct process.
There are literally "Steno Masks," a sound-dampening mask that fits over your lower face that you can use to repeat whats being said, clearly and quietly, to transcription software on your computer (along with formatting notes and the identities of speakers.) Stenographers aren't required to use a stenotype, they're required to produce an accurate transcript of the proceedings, and to be able to read back what was recently said. If you come up with a system that does that more efficiently, put it into action and take the profit.
The methods used are defined by their efficiency. This is a thread about discussing people who can type faster and more accurately than one would imagine possible, and easily better than current machine performance.
Stenotype might very well be on the way out, but I still want every trial transcript to be edited by an actual human with expert training and who is strictly regulated and held responsible.
I'm fine with lawyers using software to optimize their work. But I don't ever want to replace lawyers themselves. I can understand the temptation to create a law AI, since so much of what lawyers do is systematic and procedural. I still want a licensed lawyer to review and sign off on any arguments.
Stenographers aren’t just fast typists, they are specialists trained to produce a written (shorthand) record of the hearing in real time. Audio recordings make this obsolete. Certified full transcripts are still needed, but they can be produced more quickly using concurrent typists and proof readers, and more cheaply because you don’t need to hire specialists trained in the use of expensive stenotype machines, or pay them to sit around in the courtroom when nobody is speaking. It’s not a novel idea, which is why I mentioned that as far as I know, stenotype is now used only in the United States. https://www.bbc.com/news/magazine-13035979
I think it's also that court reporters are legal officers of the court. They have some legal training, and are highly regulated. They can be held responsible for what happens, in the same way as other court officers, like court clerks.
Also, they witnessed what happened in the court room first hand, which is helpful.
While you could certainly improve training for a team of outsourced stenographers, I think most people would feel safer knowing that the person generating the transcript is closely connected with the court, and not just a team of anonymous replaceable people in another country who didn't see the trial, and may only have access to portions of it.
What is the real benefit? It's debatable whether any money would even be saved, paying several people in a developing country less than the salary of a single court reporter. And it introduces all sorts of new liabilities and complications.
Personally, I can't see the cost savings really being worth it. The court reporter is hardly the most expensive part of a trial. I get nervous enough outsourcing my voicemail transcription to anonymous humans, so I can't imagine trial transcripts.
The thing though that I think most people miss, is that this is far more than a technical problem. The courts are an instrument of civil society, and they need to be held to really strict standards. Legal procedures took centuries to evolve, so that people will accept them and time is not wasted re-inventing procedures. This is so much more important than any optimizations that could be introduced. And if it becomes too opaque and complicated, it loses any value because people need to be able to trust it.
This over-optimization attitude can be a real problem for society. I am worried that silicon valley will try to "solve" this problem for us.
It is bad enough that this "move fast and break things, bro" culture is going after our voting machines and our nuclear missile silos.
It is totally amazing what almost-instant 100% correct (including puns and jokes that I would easily miss as a non native and that automatic captioning would definitely miss) texting does to live streams.
Source: I saw it at an Angular conference in London three years or so ago.
One interesting bit for nerds: those people study up on different topics (law, programming etc) to customize their dictionaries and get it correct every time.
Re. languages: The issue with the Plover type stenography is that it heavily relies on a dictionary. If a dictionary does not exist for your language, it becomes close to useless.
Many European languages use a form of orthographic stenography instead. This is perhaps a tiny bit less efficient, but makes up for it with flexibility in spades.
I started working on a computer implementation of that, but sadly life got in the way. If there was any project I could be paid to continue work on, I would want it to be this. https://github.com/kqr/qweyboard/
The language itself is a bottleneck, English for instance is very redundant. I suppose this is an error correcting mechanism of sort. So reinventing the language could increase the bandwidth 7th fold. Alternatively an intelligent model could cut away this by means of compression, look into Dasher for more on that.
Cool, I've always wondered how the stenography machines worked. Also, I'm stupid because I read through many pages before realizing I was mispronouncing 'Chording' in my mind as 'Chow-ording' and was wondering where that word came from.
Chording like playing a chord in music aka playing multiple notes at the same time. Stenography involves pressing multiple keys at the same time to type entire words or syllables at a time.
I've reached a point with typing where I can ~95% touch-type. This is fairly recent for me, let's say in the last two years. However, most of what I do with a keyboard is code, and I'm not convinced that touch typing speeds up my coding that much.
However, if I'm wrong I'd like to put in the effort to reach the final 5%, or even maybe switch to a different keyboard layout, or teach myself stenography.
Does anyone have any data or insight on touch typing/stenography and coding speed rather than typing speed?
I would guess that coding speed is also tied up with knowing the features of your editor well. There's really efficient ways of moving blocks around with vi, or applying regex patterns for example. I've certainly watched over the shoulder of people flying through refactoring, even though their actual typing speed wasn't notable.
No data, but I don't think you generally need to type all that fast for coding. You just need to type well enough that it's not an impedement or a distraction. It's ok to look for some of the symbols you don't use that often.
Fast typing helps a smidge with the easiest problems on topcoder. Of course, that's after you commit to single character variable names.
I find touch typing to be most valuable to me in pair programming with coworkers, or live programming for teaching.
I don’t need to think about what my fingers are doing, and I can focus on what I’m speaking or what I’m going to do/talk about next.
This reduces the frequency and duration of pauses, and makes the presentations more engaging. The goal is to be modifying text just a bit slower than people can keep up with, and taking explanatory pauses after dense changes.
Obviously helps if you can navigate your editor sans mouse.
I read through some of the site. From what I can tell, this would using stenography works best for typing actual words. I wonder how hard it would be to configure the stenography software to be able to output strings that I actually type as a programmer, given that a nontrivial amount of them are made up on the spot.
There's some examples in the article, but it seems like the gist is you assign a chord to various symbols. Like a curly brace { might use a modified chord for the word "brace"
But that seems to defeat the purpose if typing a single character maps to a chord. I really don't understand how chording could be used to make programming faster given how you have things like someLongVariableName, and especially since good editors already provide excellent support for things like templates.
Overall, I'd say that coding speed doesn't really change as typing fast is not what makes coding fast.
There are some real advantages that I find difficult to quantify, though. I switch between stenography and typing for both coding and writing depending on whether I'm at my desk and I find it hard to express clearly why coding in stenography feels natural and nice. I suppose: there's a certain fluidity when you break things down into semantic words rather than simply symbols.
How do you do the arrow keys and navigate the page?
When coding I seem to expend most of my keystrokes just moving around the page. I take it you just setup some short chords for each arrow key, pgup/pgdn, ctrl, alt... etc.
It is hard enough getting around a desktop with just keyboard shortcuts as it is!? I expect I would waste an inordinate amount of time fiddling with my dictionary trying to optimize keystrokes for the OS and apps that I spend the most time in.
I've always wanted to learn how to do this but I've also been curious how you can type special symbols like ones that are needed in code. I doubt a court stenographer is ever writing many curly braces or parentheses.
Nim has style - you could just force it. You could also add a new-word-in-identifier sound, along with a sound at the beginning denoting the style (e.g. camel case vs snake case). If your language has well established standards for it (Rust comes to mind as using Camel case, snake_case and SCREAMING_SNAKE_CASE), then it could just be auto corrected to the correct style.
While writing lisp code in classic editors, you already shouldn't be editing or navigating code with classic commands, instead you will be using structured editing.
Of course it isn't limited to lisp, but languages with C-like syntax are harder to use for structured editing.
I highly recommend watching this video, where the speaker uses voice commands to write both in emacs-lisp and in python.
I feel typing speed is rarely my bottleneck for coding/writing/anything I do on the computer. Not sure who needs to type this fast besides actual stenographers?
Same, even when bashing out an HN post I spend more time thinking than typing. When coding I spend far, far more time thinking about it (and the better I get at coding, it seems, the more this ration swings towards thinking and away from typing!)
If I could type that fast, I would immediately try to use the keyboard to think on screen, like thinking out loud (or subvocalizing).
Writing as I think is like climbing with a rope. I don't get mental superpowers, but I do have an easier time mapping out ideas, as I don't have to rely so much on memory. Then faster typing means faster, easier ideation. Maybe I would journal more.
This is not a new concept: Braille users type out Braille (mechanically or digitally) with Perkins-style 6 or 8-key (in reference to the Braille dots being encoded) keyboards, only using chords, which are known as Braille chords.
These Perkins-style keyboards, which are also colloquially referred to as Braille keyboards, are often integrated into refreshable Braille displays (and are sometimes standalone devices that are known as Braille notetakers, which also allow one to utilize various Braille codes better), which is how one reads digital material. Not only can these keyboards (and chords) be used to type, they are used to navigate, via screen reader, mobile and physical computer terminals, along with other interfaces.
When it comes to typing out something in math code, for example, Braille users have the ultimate advantage over anyone who uses LaTeX/MathML, if they know which Braille code to learn. Right now, that Braille code is currently Lambda. They can type out, for example, infty, with a single Braille chord, if they know what they are doing. Such codes also allow for the spatial encoding of the integral sign and its components, in a linear manner, along with the back-translating of that into how it should be displayed visually to a sighted person.
But, if one learned Braille as an adult, they tend to prefer to at least keep a standard Bluetooth keyboard around to type in QWERTY/QWERTZ. They cannot help falling back on it from time to time.
The people who learned Braille in school as a young child tend to prefer Braille chords over standard keyboards, even for standard typing. It can help prevent typing errors too, so it is a very useful skill to have.
One can practice these chords on the vast majority of standard keyboards, including on laptops.
But, there are also a couple of Braille Bluetooth keyboards that are being sold, but they are typically overpriced for what they are.
It reminds me of Chinese characters. Sure, you could use latin characters to represent a word or phrase. OR you could write a tiny little pictogram that captures the same thing in a fraction of the time and space.
The only problem is that you now have to memorize thousands of unique symbols instead of just ~26. I imagine stenographers must go through a similar process and memorize the "shape" of thousands of words and phrases.
As someone who went through Chinese grade school where I had to copy each character many, many times, I have to say those pictograms are a PITA to write :)
But I did get curious. I just tried to copy a few sentences from a UN document in both Simplified Chinese and English, and the time I spent are surprisingly similar. I imagine Traditional Chinese would be worse though.
This seems similar to the fuzzy-filtering found in many text editors... I’d never type “IInstantiationService”, I’d just do something like “IIntSer<Tab>” (And get the module auto-imported too if needed!)
Indeed, I was thinking that generating a chording dictionary based on an optimal number of bits of information that would reduce the error rate on typos could potentially speed up overall throughput.
I actually own a TAP (gen 2). It's surprisingly quick to pick up, especially with the apps, and once you get used to them they work surprisingly well.
One of the reasons I got it was actually how appealing the alternate mouse inputs were.
I tend to "collect" dexterity-based hobbies, so ymmv, but I found it pretty fun and satisfying to pick up. I'm completely out of practice now (new baby and wfh sucked up a bit more of my free time) but might just pick it back up anyways.
This is really fascinating... But is it worth the investment when there's cheap storage for high quality audio and video recordings, and text-to-speech does a decent job? What am I missing?
> I'm a programmer who uses stenography in my everyday work. Coworkers often come to my desk to chat or ask questions, not knowing that I use a steno machine to program and type. As I start typing in front of them, they'll watch as words and symbols appear in bursts, faster than they've ever seen anyone type before. Then, confused, they'll look down at my hands to see how I can possibly move that fast and see the reduced keyboard layout that I'm typing on, which always leads to more questions.
Would be really interesting to know how exactly they use steno for writing code. Also imagine using a keyboard that looks like this [0]
> The dictionary is customizable. As the English language develops and new words are created, the stenographer can add them to their dictionary. This is critical for anyone writing jargon, complex terminology, or programming languages
Honestly, I would only learn chording (steno) if you find it interesting/fun, and not because you think it will save time in the long run after the initial time investment. This isn't a trivial weekend adventure -- it's a legitimate skill that requires hours and hours of consistent practice and training.
That being said, I greatly encourage anyone to try it out. Typing using chords is so satisfying! If you're looking for a cheap dedicated steno keyboard, take a look at the Georgi [1].
It's unlikely-bordering-impossible that one's writing bottleneck is actually getting the words into a computer, except on very small timescales (maybe tens of minutes?)
Thinking and editing. There's a legitimate difference between being blocked conceptually (writer's block) and thoughtfully taking the time to find just the right words to express your intentions succinctly and elegantly. I've tooled over a three page manuscript for days--not because I was suffering from lack of creativity, but rather, in the interest of continual refinement.
I type at a little more than 70 WPM, but even when writing this comment, I'm writing far below that as I'm trying to gather my thoughts, as well as editing as I'm going along.
Now, when I sit down to write a page for my thesis, I'm faaarrr below the that 70 WPM, since those thought are even harder to sort in my mind.
Clearly, I'd be served by writing handwritten drafts, etc..., but I really think Steno probably only shines when taking dictations or typing up someones notes they've written.
I'm surprised that there seem to be no examples anywhere of anyone seriously using chorded typing for programming. After cheap chorded keyboards entered the market and the associated software became free, I assumed it would only be a matter of time before something like Travis Rudd's incredible demo of voice coding [1] came out. Perhaps that it hasn't speaks to how difficult it is to get real speed improvements with a chorded setup?
I'm the author of Art of Chording—I program full-time with steno in JavaScript (working mainly with React.)
I'd love ideas on how to demonstrate coding in steno. I struggle with it sometimes because the slowest part about coding is not the input rate… it's the brain. I guess if people are looking to code "quicker"… it's not the rate of input that one would want to explore. I will say that writing comments became a lot easier when the words started to just flow onto the screen.
I would imagine that one of the difficulties with use chording for programming is that so many of the terms used are not normal words. I suppose you can just create new chords for language keywords or common APIs but it just seems like there are so many possible unique terms you might need to type at any given moment. How do you handle long function names, snake case, camel case? Has it changed the way you name your own variables and functions?
This is interesting, how does it work with different languages, Every dictionary for every language?? If it's different languages, how is it switching when using different languages.
This is an important topic of development for Open Steno. Machine shorthand has existed in many languages and is used in many countries.
However, not all of them are "computer" or "realtime" compatible, which roughly means that you wouldn't be able to distinguish in your writing between homophones like "their" and "there" and "they're."
I'd be very curious to see relative rates of repetitive stress injuries in stenographers vs regular typists, controlling for amount of time spent typing, etc.
On the one hand, when I see something like this I think: Maybe this would be worth looking into to reduce chances of RSI.
On the other hand, when I look at videos of people typing like this, it strikes me as being ergonomically weird. Very low keyboards, hands locked into a relatively small area of space, etc.
I see the author is here. Would love to hear about this from an insider!
Disclosure: full-time programmer, working jazz drummer, amateur pianist who is always terrified that it's all gonna catch up with me one of these days.
The claim I see being made is that stenography is mostly arm motion driven as opposed to normal typing being key driven, due to stenography basically being just around the home row.
Fascinating! In a time of ML algorithm flourished, I'm wondering if there is a way to combine as few as possible keystrokes and GPT-3 style machine intelligence to make typing both easy to learn and fast enough to match the speed of thought. As far as I know, Chinese Pinyin input methods have seen a lot of improvements in this direction in the past 10-20 years. If one could reduce the size of keyboard to only 10-20 keys, there will be many innovations in the computing devices that we use everyday. After all, the size of physical keyboard is one of the main constraints that limits the form factor of our phones and portalbe computers.
88 comments
[ 4.8 ms ] story [ 140 ms ] threadThat makes him about as fast as a 300 baud modem. And probably around 1.8x the fastest keyboard typists.
Edit: Curious if this as "optimized" as it can be. Is there a theoretically faster way to get words from your hands to something digital?
I'm also curious how this works for court cases where a language other than English comes into play, alongside English. Canada might be a good example, where French and English dialogue might often co-exist.
The related Penti-Chorded keyboard apparently works fairly well for German [0], and Japanese Kanji [1]. I'd imagine that using the arpeggio to shift the characters would work similarly with the steno.
[0] https://software-lab.de/penti.html
[1] https://picolisp.com/wiki/?TermuxPentiPicolisp
For the sake of, err, discussion, I can envision some sort of wearable glove like input device. I know I can play notes on a guitar far faster than he is inputting characters because of the technique involved. I feel like it could be possible to come up with a faster method playing around with that idea.
But the ultimate would be hands free, no? Makes one wonder if this is yet another skill/job about to be blown away by automation.
Offline speech recognition with perfect accuracy will probably come standard with every phone in a couple of years.
I wonder if there's a blind stenographer that can really fly, but just isn't well known.
I bet there are some unknown talents in the closed captioning space.
Unrelated, I've seen some interesting typos there, especially with strong English accents like Scottish or Irish. And you can see them improving as episodes progress. The "RedRock" Irish soap is a good example. They get much better over time.
Sounds like the perfect instrument for Steno Hero: https://killscreen.com/previously/articles/steno-hero-might-...
Half the speed.
If I remember correctly, just to get his license he had to hit 270w/m. He said he can get up to about 300w/m.
The raw transcripts are composed of shorthand codes, and letters that represent syllables and other components of speech. I saw that many words were just represented by one or a few letters. So, by "words", we are often talking about codes of only a few letters. They are printed on thermal receipt width paper, and floppy disk (this was about 15 years ago).
> I'm also curious how this works for court cases where a language other than English comes into play, alongside English. Canada might be a good example, where French and English dialogue might often co-exist.
It shouldn't matter what language it is, because the stenographer has their own custom shorthand dictionary. The raw transcript that they generate is for the stenographers internal use. Only after the trail do they transcribe it into human readable form. Lawyers then have to pay the stenographer a fee for their transcripts. My uncle told me that it is common for multiple lawyers to each pay him a fee that he charges by the page.
He, like most court reporters, outsources their transcription to other people via Email. It requires human transcription, because the raw transcripts need to be interpreted. It is not a something that you can just run through a filter. His scopists[1] will email him back and fourth for clarification on parts of the transcript they don't understand. This is important, because their transcripts are the official legal record of what happened in a trial.
[1]: https://en.wikipedia.org/wiki/Scopist [2]: https://en.wikipedia.org/wiki/Shorthand
I know that when my uncle started in the early 90s, there were still some court reporters that used shorthand with pencil and paper. I don't think there is a requirement that they use a steno machine. Their job isn't to record what happened, their job is to generate a single authoritative transcript that everyone works from. In theory, they could probably not take notes at all, and just work from the recording later, but it is probably easier to take the first round of notes live.
When a court reporter generates a transcript, they sometimes have to go back through the audio recordings sometimes and actually clarify in the official transcript exactly what was said. They have to make it clear in the transcript if people were talking over each other, or to try to clarify what exactly was said if it might be ambiguous.
I think it is a good idea for a trial to have an official record that has been processed by a human, instead of just throwing a recording up and calling it a day.
Think of how inefficient it would be otherwise, where all of the parties would essentially have to act as their own court reporters and generate their own transcripts.
What if different parties disagreed with the interpretation of the transcripts? Perhaps the court would then have to appoint a neutral expert to make an authoritative determination of what was said; and we've now re-invented the court-reporter role.
I think there is a tendency for people to look at this only as a technical problem to be solved. Like with voting, it isn't just a matter of securely tallying votes, there is much more to it. The process requires maximizing transparency and minimizing room for anyone to contest it.
One of my best clients was middle-aged mother of three - let's call her Pat, because that's close enough - whose home IT setup always needed a lot of extra love. Weird viruses on old machines, ancient printers that had to be coaxed to work well when shared over a network, "what laptop should I buy my son for college" - the usual.
She was a court reporter, by trade. I had to help her with her laptop and court reporting software / hardware from time to time - it was very interesting. She had a custom "keyboard" that only had 10-15 unlabeled keys that really looked more like long and thin paddles than anything else; it worked with a program that recorded audio and tracked it to her inputs.
So she could wind the report back to any given point in time, and it would highlight what she'd typed as it went. So if she needed clarification on something - she could enable audio playback that showed what she had transcribed, not unlike a karaoke machine.
It was all very interesting stuff. I was a little sad when I had to tell her that I couldn't do work for her anymore, because I didn't have the time to do side work due to a developing career and social life.
Interesting in any case.
especially if someone points at someone else or uses other means of communication. Of course the judge can then say what was communicated so there is a verbal record in the recording but it is probably better that the court reporter is there in that case.
There are literally "Steno Masks," a sound-dampening mask that fits over your lower face that you can use to repeat whats being said, clearly and quietly, to transcription software on your computer (along with formatting notes and the identities of speakers.) Stenographers aren't required to use a stenotype, they're required to produce an accurate transcript of the proceedings, and to be able to read back what was recently said. If you come up with a system that does that more efficiently, put it into action and take the profit.
The methods used are defined by their efficiency. This is a thread about discussing people who can type faster and more accurately than one would imagine possible, and easily better than current machine performance.
I'm fine with lawyers using software to optimize their work. But I don't ever want to replace lawyers themselves. I can understand the temptation to create a law AI, since so much of what lawyers do is systematic and procedural. I still want a licensed lawyer to review and sign off on any arguments.
Also, they witnessed what happened in the court room first hand, which is helpful.
While you could certainly improve training for a team of outsourced stenographers, I think most people would feel safer knowing that the person generating the transcript is closely connected with the court, and not just a team of anonymous replaceable people in another country who didn't see the trial, and may only have access to portions of it.
What is the real benefit? It's debatable whether any money would even be saved, paying several people in a developing country less than the salary of a single court reporter. And it introduces all sorts of new liabilities and complications.
Personally, I can't see the cost savings really being worth it. The court reporter is hardly the most expensive part of a trial. I get nervous enough outsourcing my voicemail transcription to anonymous humans, so I can't imagine trial transcripts.
The thing though that I think most people miss, is that this is far more than a technical problem. The courts are an instrument of civil society, and they need to be held to really strict standards. Legal procedures took centuries to evolve, so that people will accept them and time is not wasted re-inventing procedures. This is so much more important than any optimizations that could be introduced. And if it becomes too opaque and complicated, it loses any value because people need to be able to trust it.
This over-optimization attitude can be a real problem for society. I am worried that silicon valley will try to "solve" this problem for us.
It is bad enough that this "move fast and break things, bro" culture is going after our voting machines and our nuclear missile silos.
Source: I saw it at an Angular conference in London three years or so ago.
One interesting bit for nerds: those people study up on different topics (law, programming etc) to customize their dictionaries and get it correct every time.
Many European languages use a form of orthographic stenography instead. This is perhaps a tiny bit less efficient, but makes up for it with flexibility in spades.
I started working on a computer implementation of that, but sadly life got in the way. If there was any project I could be paid to continue work on, I would want it to be this. https://github.com/kqr/qweyboard/
However, if I'm wrong I'd like to put in the effort to reach the final 5%, or even maybe switch to a different keyboard layout, or teach myself stenography.
Does anyone have any data or insight on touch typing/stenography and coding speed rather than typing speed?
Fast typing helps a smidge with the easiest problems on topcoder. Of course, that's after you commit to single character variable names.
I don’t need to think about what my fingers are doing, and I can focus on what I’m speaking or what I’m going to do/talk about next.
This reduces the frequency and duration of pauses, and makes the presentations more engaging. The goal is to be modifying text just a bit slower than people can keep up with, and taking explanatory pauses after dense changes.
Obviously helps if you can navigate your editor sans mouse.
My thoughts on going down to one-symbol-per-chord:
- Due to the small steno layout, you don't need to stretch your hands to far symbols on the keyboard. - You're not limited to what's on the keyboard. Symbols like ÷ and © and any emoji are now first-class citizens. - There are cases where you get multiple symbols per chord. For example, calling a function `()` is one chord. Writing an arrow like `=>` is also one chord.
Overall, I'd say that coding speed doesn't really change as typing fast is not what makes coding fast.
There are some real advantages that I find difficult to quantify, though. I switch between stenography and typing for both coding and writing depending on whether I'm at my desk and I find it hard to express clearly why coding in stenography feels natural and nice. I suppose: there's a certain fluidity when you break things down into semantic words rather than simply symbols.
Hope that helps!
When coding I seem to expend most of my keystrokes just moving around the page. I take it you just setup some short chords for each arrow key, pgup/pgdn, ctrl, alt... etc.
It is hard enough getting around a desktop with just keyboard shortcuts as it is!? I expect I would waste an inordinate amount of time fiddling with my dictionary trying to optimize keystrokes for the OS and apps that I spend the most time in.
The thing that’s still a question for me is variable names. How does one steno out snake_case or CamelCase?
You can also do a stroke-by-stroke basis. For example, I have strokes for prefix "is" and "on" followed by a capital. So
would be "onSubmit". You could also fall back to forcing an attached, uppercase word. So:While writing lisp code in classic editors, you already shouldn't be editing or navigating code with classic commands, instead you will be using structured editing.
Of course it isn't limited to lisp, but languages with C-like syntax are harder to use for structured editing.
I highly recommend watching this video, where the speaker uses voice commands to write both in emacs-lisp and in python.
https://youtu.be/8SkdfdXWYaI?t=540
Writing as I think is like climbing with a rope. I don't get mental superpowers, but I do have an easier time mapping out ideas, as I don't have to rely so much on memory. Then faster typing means faster, easier ideation. Maybe I would journal more.
It's the whole waterfall-vs-agile on a micro scale.
The other thing that I sometimes end up doing with stenography is just a stream-of-consciousness-style writing where I then have to edit it down.
These Perkins-style keyboards, which are also colloquially referred to as Braille keyboards, are often integrated into refreshable Braille displays (and are sometimes standalone devices that are known as Braille notetakers, which also allow one to utilize various Braille codes better), which is how one reads digital material. Not only can these keyboards (and chords) be used to type, they are used to navigate, via screen reader, mobile and physical computer terminals, along with other interfaces.
When it comes to typing out something in math code, for example, Braille users have the ultimate advantage over anyone who uses LaTeX/MathML, if they know which Braille code to learn. Right now, that Braille code is currently Lambda. They can type out, for example, infty, with a single Braille chord, if they know what they are doing. Such codes also allow for the spatial encoding of the integral sign and its components, in a linear manner, along with the back-translating of that into how it should be displayed visually to a sighted person.
But, if one learned Braille as an adult, they tend to prefer to at least keep a standard Bluetooth keyboard around to type in QWERTY/QWERTZ. They cannot help falling back on it from time to time.
The people who learned Braille in school as a young child tend to prefer Braille chords over standard keyboards, even for standard typing. It can help prevent typing errors too, so it is a very useful skill to have.
One can practice these chords on the vast majority of standard keyboards, including on laptops.
But, there are also a couple of Braille Bluetooth keyboards that are being sold, but they are typically overpriced for what they are.
It reminds me of Chinese characters. Sure, you could use latin characters to represent a word or phrase. OR you could write a tiny little pictogram that captures the same thing in a fraction of the time and space.
The only problem is that you now have to memorize thousands of unique symbols instead of just ~26. I imagine stenographers must go through a similar process and memorize the "shape" of thousands of words and phrases.
https://en.m.wikipedia.org/wiki/Chinese_character_classifica...
But I did get curious. I just tried to copy a few sentences from a UN document in both Simplified Chinese and English, and the time I spent are surprisingly similar. I imagine Traditional Chinese would be worse though.
edit: also, written Chinese has been traditionally thought of as the widespread adoption of a stenographic script: https://en.wikipedia.org/wiki/Clerical_script
One of the reasons I got it was actually how appealing the alternate mouse inputs were.
I tend to "collect" dexterity-based hobbies, so ymmv, but I found it pretty fun and satisfying to pick up. I'm completely out of practice now (new baby and wfh sucked up a bit more of my free time) but might just pick it back up anyways.
> I'm a programmer who uses stenography in my everyday work. Coworkers often come to my desk to chat or ask questions, not knowing that I use a steno machine to program and type. As I start typing in front of them, they'll watch as words and symbols appear in bursts, faster than they've ever seen anyone type before. Then, confused, they'll look down at my hands to see how I can possibly move that fast and see the reduced keyboard layout that I'm typing on, which always leads to more questions.
Would be really interesting to know how exactly they use steno for writing code. Also imagine using a keyboard that looks like this [0]
[0] https://www.artofchording.com/introduction/how-steno-works.h...
> The dictionary is customizable. As the English language develops and new words are created, the stenographer can add them to their dictionary. This is critical for anyone writing jargon, complex terminology, or programming languages
That being said, I greatly encourage anyone to try it out. Typing using chords is so satisfying! If you're looking for a cheap dedicated steno keyboard, take a look at the Georgi [1].
[1]: https://www.gboards.ca/product/georgi
If you write a lot, about what amount of writing do you think it can in the pay back the initial time investment?
Now, when I sit down to write a page for my thesis, I'm faaarrr below the that 70 WPM, since those thought are even harder to sort in my mind.
Clearly, I'd be served by writing handwritten drafts, etc..., but I really think Steno probably only shines when taking dictations or typing up someones notes they've written.
For the keyboard shortcut, I use "FEFK" which is default in the Plover dictionary.
[1] https://www.youtube.com/watch?v=8SkdfdXWYaI
I'd love ideas on how to demonstrate coding in steno. I struggle with it sometimes because the slowest part about coding is not the input rate… it's the brain. I guess if people are looking to code "quicker"… it's not the rate of input that one would want to explore. I will say that writing comments became a lot easier when the words started to just flow onto the screen.
Here are my existing videos:
Unscripted: https://www.youtube.com/watch?v=711T2simRyI
Scripted: https://www.youtube.com/watch?v=RBBiri3CD6w
Looking forward to any suggestions on how to improve or even just the big open questions you'd want answered on this subject.
https://www.youtube.com/watch?v=z_01ha1uS6Y https://www.software-lab.de/penti.html
Does his development work on a tablet with a chorded keyboard software.
However, not all of them are "computer" or "realtime" compatible, which roughly means that you wouldn't be able to distinguish in your writing between homophones like "their" and "there" and "they're."
Here's a list of some of the languages that have been developed or ported to Plover: https://github.com/openstenoproject/plover/wiki/Chorded-Syst...
There is a Plover plugin to switch on-the-fly between different steno systems.
There is also a plugin to switch between enabled dictionaries on-the-fly.
Finally, there is the idea of bilingual dictionaries, but I haven't seen it implemented well yet.
There's also the problem of some languages having drastically different layouts.
So far, the most multilingual and successful stenographer I've seen is Stanley Sakai. Here's him writing in a Spanish theory that he developed: https://www.youtube.com/watch?v=eGZ43TID9jU&t=90s
I also know that there are bilingual stenographers in Canada who write both English and French, but I haven't seen it in action yet.
In summary, I think that multilingual stenography is critical for the adoption of steno, but it's currently not easily accessible or widely used.
For this system, You’d need an input device that sends its input to a dictionary and then sends that through the normal input?
Like another poster here, vim’s modal editing bit with chording makes me wonder what could be achieved with a normal keyboard
On the one hand, when I see something like this I think: Maybe this would be worth looking into to reduce chances of RSI.
On the other hand, when I look at videos of people typing like this, it strikes me as being ergonomically weird. Very low keyboards, hands locked into a relatively small area of space, etc.
I see the author is here. Would love to hear about this from an insider!
Disclosure: full-time programmer, working jazz drummer, amateur pianist who is always terrified that it's all gonna catch up with me one of these days.
If there where to be a heads up display as a reference, the learning curve would be cut down quite a lot.
Check out contact form if anyone is interested.
http://tbf-rnd.life/blog/2019/06/16/0-learning-curve-chorded... http://tbf-rnd.life/contact/