Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: I automated half of my typing

github.com

Post reply on HN
Print view View original

41–50 of 331 posts

Re: Show HN: I automated half of my typing

#41

Post by JohnFen »

absoluteunit1 wrote: post #20
JohnFen wrote: post #17

I did this sort of thing quite a lot early in my career, but stopped quite a long time ago. The problem is that I started using multiple machines running different OSes, and not all of those machines are "mine", so I couldn't have these sorts of specializations everywhere. That turned a positive into a negative when using machines that didn't have these optimizations.

I haven’t done what you or the author have, but one workaround this is using something like QMK or ZMK. Especially the Leader key on QMK. Thanks for the post though, I need to do some analysis of my own typing habits to automate them

Interestingly enough, I have a macro keyboard I built myself that I use at home. It never occurred to me to use it professionally until you brought it up. Hmmm. Thank you!

Re: Show HN: I automated half of my typing

#42

Post by crazygringo »

This is a very clever idea.

However, I realize that I would never want to use something like this that would change over time. E.g. if I ran it every 6 months and last year "db" produced "debug", while this year it produces "database". Because talk about messing up my muscle memory and habits. And the language I write changes very much over time.

So I'd actually be much more interested in a "universal" version of this -- if you ran it across books and e-mails and text messages from thousands of authors covering diverse backgrounds and contexts, then what would most reliably help everyone?

E.g. expanding "t" to "the" seems like a no-brainer, just like "st" to "something". Is there a minimal set of, say, 200-500 of these that could simply be turned into a "standard keyboard" that everyone could learn?

Re: Show HN: I automated half of my typing

#43

Post by crazygringo »

jaxrtech wrote: post #37

I wish there was a way to just repurpose existing operating system support for CJK language IME for English.

Could you elaborate?

I'm aware of CJK input methods, but it's not immediately clear how/why you could repurpose them for English. Not to mention there are several types.

Re: Show HN: I automated half of my typing

#44

Post by zerojames »

I _love_ to see projects like this.

I was working on a similar problem this weekend, but with whole words instead of abbreviations I had made in a dictionary, and in the general case, fine-tuned on any given corpus of text.

I wanted to know: could I write an autocorrect that is "fine-tuned" on a given corpus of text? Use case: I write a lot of docs with long phrases (i.e. "data augmentation"). Could I automate them?

I arrived at:

    1. Calculate "surprisal" of unigrams and bigrams (entropy) from a general dataset (an NYT corpus), give a boost to words in the "fine-tuned" index;
    2. Create a trie data structure that is weighed by surprisals. The more surprising a word, the more weight it gets.
    3. Use that as advanced autocomplete.
I got a working solution here: https://github.com/capjamesg/autowrite/blob/main/autocomplet...

(No docs yet -- coming in the next few days. Leave a GitHub Issue if you want to chat about it!)

Re: Show HN: I automated half of my typing

#46

Post by MobiusHorizons »

quadrature wrote: post #28
tikkun wrote: post #21

Idea for extending this: keylogger (yeah, that's a downside...) that'll watch what you type, learn common phrases, identify common ones, then come up with shortcuts, then each time you type the full phrase it gently reminds you of the shortcut. Then anyone can install it, no friction, and over time it'll slowly start making you gently more productive.

Its not a keylogger in itself, its a tool to configure your keylogger.

I guess you are half right, the tool this repo configures is a shortcut engine (autokey). It has access to the key presses, but afaik does not log them.

Re: Show HN: I automated half of my typing

#48

Post by tshaddox »

input_sh wrote: post #36
jawns wrote: post #29

If you're going to use models, why not models all the way through? In other words, rather than using them merely to identify words to abbreviate and suggested abbreviations, why not use an LLM that can take a bunch of abbreviated text and infer what you're trying to abbreviate? Tht wy, u dn't hv 2 mem lsts f abbrs ahd f tm. ChatGPT was able to successfully guess that last sentence in abbreviated form: That way, you d…

Ah yes, I'd love to wait about five seconds every time I type some text for ChatGPT to return some output, worrying about boilerplate text around the actual answer, and to introduce unpredictability into the mix.

I'd probably try using ChatGPT to suggest the specific abbreviations for the most common phrases.

Re: Show HN: I automated half of my typing

#50

Post by calvinmorrison »

crazygringo wrote: post #42

This is a very clever idea. However, I realize that I would never want to use something like this that would change over time. E.g. if I ran it every 6 months and last year "db" produced "debug", while this year it produces "database". Because talk about messing up my muscle memory and habits. And the language I write changes very much over time. So I'd actually be much more interested in a "universal" version of thi…

Congrats you just invented shorthand!
Post reply on HN