celerity

The plugin actually does support spaced repetition, but only an "exponential" algorithm. That is, you have to manually choose whether to multiply the current repetition period with 0.5, 1, 1.5 or 2.

This "exponential" repetition it's not as optimized as the kinds of algorithms you're describing, but I find it good enough and it comes with a few distinct advantages:

1) Spaced repetition algorithms are hard to get right with two buttons like remembered/forgot. Anki for example added more buttons (easy/normal/hard/forgot or something like that), but I could never really intuit what would happen if I clicked any of them. Having 4 explicit choices makes it easy to understand what will happen, and also lets you "manually" push back notes on axes like "how much do I still care about this" and not just "how well do I remember this".

2) I really wanted to keep the metadata fields stored in the note and easy for a human to edit and understand. Right now, you only need `repeat: spaced every N days` (and similar) and the plugin adds a `due_at` field that has an easy to read ISO timestamp. You can also add `hidden: true` but don't have to.

Hey, this is exactly why I made obsidian-repeat-plugin [1]. The goal was to have most of my notes surface every now and then, which removes the need for me to do any digging at all.

With that said, my non-repeating notes are far from "dead". First, I have a lot of useful self-made references. Second, Obsidian really makes it easy to spontaneously group tags with its backlink system. Every now and then, I go down a rabbit hole and catch up to notes that would otherwise be dead. I only have to maintain a few "entry points" that mostly have other notes linking to them.

[1]: https://github.com/prncc/obsidian-repeat-plugin

From the article here it looks like people will stay on payroll until early 2023. They won't work but have to abide by Twitter's internal agreements.

Does anyone know if part of these agreements includes not seeking other employment (a typical clause these days)?

If you want to get a taste of some of what Ramanujan seems to have been occupied with, the mathematical physicist John Baez explained "Ramanujan's easiest formula" in an approachable way https://johncarlosbaez.wordpress.com/2020/11/18/ramanujans-e...

Nice! We built something like this 3 years ago while I was still at Intoli. Instead of Puppeteer we used our own Web Extensions API based framework called Remote Browser [1], the core of which was written by my cofounder.

The tour is still up at [2]. The servers that actually run the Remote Browser have since gone down, but interestingly you can still run the tour. That's because if you don't change the code in the REPL window, you get cached results (except step 7/7 which scrapes Hacker News and won't work). To get those results, we built a little tour "recorder" that would be run on every release. If I remember correctly, we allowed some dynamic ES6 imports through a custom Babel compiler for the code that's input, which also allows first level async stuff, which still works :)

[1]: https://github.com/intoli/remote-browser [2]: https://intoli.com/tour/

A few aphorisms about Voja Antonić:

1. He was involved as a skeptic and wrote a well-received (among my friends at least) book debunking psychics and various kinds of nonsense. As a teenage boy in Serbia (in the late 90s?), I asked him to translate a portion of the book to English and put the translation on my website. He graciously allowed me to do so. Part of why I wanted to go through that massive effort was to convince an English-speaking girlfriend (whom I've met online!) that astrology is nonsense. You could say that relationship did not last long.

The book is now available as a free PDF on his website. [1] I don't know what happened to my website.

2. He moved to the US at 65 to work SV, and had some emotional things to say about the move. [2] It stuck with me.

(Both links are in Serbian.)

[1]: http://www.voja.rs/dpdl.htm

[2]: https://noizz.rs/intervju/voja-antonic-za-noizz-o-odlasku-u-...

I wonder if the Wayback Machine people are using a (potentially more modern) version of the AOPIC algorithm to decide what to archive. I wrote an article about that algorithm (which is similar to the original PageRank, but simpler IMO), and stated that a service "like the Wayback Machine would probably use something like AOPIC." It would be nice to remove that first like from the sentence!

[1] https://intoli.com/blog/aopic-algorithm/

To be fair, 200X was the Year of the Linux Desktop for us Linux users.

A lot of the comments here are arguing for or against the author's specific choices, but I would like to point out how great it is that Linux desktop environments are this customizable in the first place.

I recently switched back to Linux after using a Mac for a couple of years, and was blown away with how far Gnome has gotten in terms of customizability -- even if most of it is done through extensions. Moreover, if you can't tweak it to your liking, perhaps xfce or i3 or KDE will prove more accommodating...

My co-founder wrote a bit about Asciinema as part of a comprehensive comparison of terminal recorders [1]. There's a lot to love about it (wide availability, very nice JavaScript playback -- check out the Game of Life playback at [1]), but it's a bit of a chore to self host, if that's important for you.

[1]: https://intoli.com/blog/terminal-recorders/

Traffic to Lambda from the internet and between Lambda and S3 is free. The only thing you pay for are the transfer costs from S3 (at cents per GB).

The article doesn't mention it explicitly, but this is a nice example of how using Bayes theorem helps you ignore the hard-to-compute normalization term of the input space. In the article, this is the P(w) term of

P(c|w) = P(c)P(w|c)/P(w),

where c is a correction, and w is the original word.

The author does implicitly talk about this when he explains that P(c|w) conflates the two factors, but it's also not that hard to see that getting a handle on P(w) -- the probability space of misspellings -- is harder than getting a hold of P(c) -- the probability space of actual words, and Bayes lets us get rid of the former during optimization.

Selenium is just a tool built on top of the WebDriver API. One of its main disadvantages is needing to run a complicated proxy program (like geckodriver, ChromeDriver, etc.) built individually for each browser in order to drive your instance. As a result, users sometimes suffer from hard to debug edge cases and other pain points.

They also make interacting with JavaScript on the page a bit painful. For example, injecting JavaScript into the browser with Selenium can be quite an ordeal [1], so you're somewhat limited in what you can do by what Selenium's developers decided to focus on. It also complicates deployments by adding another moving part to the overall equation.

In contrast, the Web Extension API is now part of all major browsers, and makes interacting with different page contexts effortless. To give a sense of the project, we wrote an interactive tour of Remote Browser which runs browser instances on our backend.

[1]: https://intoli.com/blog/javascript-injection/

[2]: https://intoli.com/tour/1

It's been mentioned elsewhere in this thread, but Exodus is really great for this use case. You can just run

  exodus htop | ssh my-server-name

and your local machine's copy of htop will be installed in your home directory on the remote machine (along with all of the dependencies).