Emacs Is My Browser
Discussion 87 comments
You can actually get other close-to-browser experiences with stuff like xwidget and embedded WebKit (https://emacsdocs.org/docs/emacs/Embedded-WebKit-Widgets) but generally it’s not worth the fight for commercial sites these days; everything is vibed and only tested against Chrome, so if you’re lucky they’ll load in other Chromium browsers and _maybe_ Firefox.
Quite a lot of sites work well with other browser engines too. I have tried PaleMoon, Netsurf. Users of Apple mobile devices seem to be fine with Safari.
Really light or unusual browsers might be a problem. Dillo, for example (its very fast though).
It sounds like the dark mode thing is coming in the next Safari release.
If I open a brand new FF profile, some of these sites do work. But I guess the combination of FF, on Linux, with my addons is triggering something.
I tried xwidget-webkit as main browser briefly, but very quickly gave up. When a collaborator sends me a Google Doc, for example, I really still need a browser to work on it. I still use xwidget-webkit daily -- it's a great PDF viewer, and really nice to be able to have Jupyter notebooks open side by side with code buffers -- but for most tasks involving the web I'll switch to a browser.
Not long ago reddit killed old.reddit.com effectively by forcing log ins to it. That totally defeated the point of it so I stopped using reddit. I also no longer read it since the platform is so censorship-heavy.
I believe some of the old IPFS folks moved to Iroh
His motive is not mindfulness, but ideology.
I love my Emacs. And yet here I am completely dumbfounded by this interaction!
EDIT: eww can render SVG quite well!
EDIT 2: but no SVG animations...
Then, recently, it hit me - w3m-mode has both tabs and session management. It somehow even displays images more reliably than Dillo. I wouldn't use it as my main browser on the desktop, but for my specific use case, it's perfect.
Paired with a local on-device model would be pretty sweet.
Feels like we were already almost there because of ad blocking, without even considering AI agents and summaries. And I take no offense in ad blockers, I use them myself.
Sad thing about current public web "content" apps (be it newspapers or other formats) is that the paid versions often share all tracking features with the ad-financed version of the same service, when both are behind a login wall. Only fewer ads, at least hopefully no ads from other vendors / no intrusive brokered ads.
Self-advertisement going through the roof when accepting an offer for a paid subscription is another frequent annoyance though.
not great for everything, and it does, unfortunately need an LLM call to generate the translation program, but if you truly want to live in emacs, I've found it as a good stop-gap for certain pages that EWW cannot effectively render.
I think the idea that you do something in a more difficult or less convenient manner for the sake of "mindfulness" or "focus" is well-established, and what I hope for (and am always disappointed by the lack of) when reading such blog posts is an actually useful feature of the unconventional method in question.
The title of this blog post promises some usefulness from using Emacs as your browser, and upon reading, you discover that the "usefulness" is the inconvenience and impracticality of using it that way.
People write these things, and use these methods, specifically because they don't conceive of usefulness the same way you do.
In this case, the justification offered doesn't mention all the various keybinds that have been pushed down to the "how" section. But clearly the simplicity is a feature:
> D will download images locally, b goes back a page, a will save bookmarks. All of this makes browsing in eww truly enjoyable.
The point is the same as usual for TUIs: you don't have to switch over to the mouse and hunt out a button, as the necessary key is always literally at hand.
People who find this sort of thing "truly enjoyable" (what clearer statement is there of featurefulness!) commonly take it for granted and don't explain the virtues of it very well. That sort of explanation is found in think pieces about TUI design more generally.
TFA describes a goal of avoiding distraction. That is not at all the same thing as "inconvenience and impracticality". Clearly the author finds these keybinds quite convenient and practical. It's just that they don't cover a bunch of web interactions that you perhaps value, but the author doesn't.
The keybindings aren't mentioned in the "Why?" section of this blog post, and clearly aren't the point of the post. You can get extensions to add Emacs bindings to full-featured browsers, and there's even the Nyxt browser which has Emacs keybinds and Lisp configuration.
The core reason listed in this blogpost for browsing this way is the very fact that it is less featureful and less convenient. As someone who uses Emacs, I know that eww isn't as convenient as a fully-fledged web browser and I know that sometimes the hard way of doing things can be beneficial. I found this blog post disappointing because it doesn't afford you any new information, nor tell anyone about any actually useful features about browsing in eww.
You’ll have people who demonstrate their whole set up and interactions and hot keys and shortcuts and everything and then when you say OK well why would you do it like that it all boils down to:
“This extremely precise implementation works for only me and my specific workflow”
I found that it’s almost more useful to think about emacs users showing off their org mode bindings are basically like Warhammer players showing off their painted figurines
That's their value.
But also, consider this: maybe they just have very specific, unusual needs that you're unable to relate to because they're very different from your lived experience. I'm not saying this as a criticism. I feel like most people struggle to comprehend just how different someone else's lived experience can be from their own. I certainly do. Usually, when I read these emacs setup blog posts, I'll dislike most of their setup, but often there will be some nugget in there I'll absolutely love, and then I might add that nugget to my init.org.
Let me try to explain my singular point of view. I have very severe ADHD-C, and GUIs always felt horrible for me to use. Especially big complex GUIs. Things like MS Office, Photoshop, big IDEs etc.
Why? Multiple things:
1. Visual processing:
ADHD can include strange visual processing or spatial reasoning problems. It's very hard for me to describe this one in a detailed manner, but suffice it to say, Icons don't make sense to me. Text does. I am capable of learning some icons, certainly, but certain ones never stick. For example, the protonmail web client has the reply and forward button displayed as arrows pointing in opposite directions, I always click the wrong one. Every time. And everytime I'm positive it's the right one and click it immediately. As I'm sitting here now, looking at an HN textbox, I can reason my way into understanding that to the western mind things flow from left to right, and so left should be reply and right should be forward. But when I'm actually there wanting to reply to an email, I hit forward every time without thinking. If on the other hand the buttons just said "reply" and "forward", this wouldn't happen. I can't explain it even to myself really. That's just how my brain brains.
2. I lose my train of thought extremely easily:
You know how sometimes when you walk into a room, you suddenly forget why you're in there? This supposedly happens to most people occasionally. For me it's more like 15 times a day, not just with rooms, but any sudden sensory transition. Like a pop up window. Ah yes. I'm in some stupid baroque GUI program. I realise I need to change a setting. I locate the settings button from the dropdowns, which every program seems to have in a slightly different place, big settings window pops up covering most of the main window. I immediately forget why I'm in the settings. I try to remember why for half a minute, and fail. I close the settings window and try to remember what I was doing before. 2 minutes later I remember which setting it was. I open the settings. I forget why I opened the settings again. Ok, annoying, but not that big a deal right? Oh but you see opening the settings is just the first roll of the dice. The settings window has 15 different tabs. And some of the tabs have buttons that spawn new sub windows, and even some of the subwindows have tabs(looking at you, Microsoft. What is wrong with you people?!). I call it menu purgatory.
3. Extraneous information distracts me very easily.
This one should be self-explanatory
4. My hand eye coordination is not fantastic, so I often click the wrong thing, which then feeds back into the wrong thing, which then feeds into 2 and 3 again. Keybindings don't require coordination, only muscle memory.
On a bad day, with a shitty piece of software like that, I can spend literally half the day locked in a cognitive battle of attrition with the software instead of getting work done. That's not an exaggeration.
Because I'm so hampered by what to other people is just minor annoyances, I need software that I can configure completely arbitrarily. No change can be off limits. That is almost never the case with GUI software, at least without cloning the source and patching it, figuring out the build system, possibly learning the implementation language. With Lisp software, the config file is really just a hotpatch. And Lisp is my favourite language, so yay. I would prefer Common Lisp, but beggars can't be choosers and Elisp isn't that bad once you learn to avoid the footguns. Certainly better than vimscript, which I've always found incomprehensible.
So that's why my setup is roughly this:
Never use GUIs with the exception of Firefox(I want to use Nyxt but I could never quite get past the learning curve, and there are still key features missing) and occasionally mpv(which is barely a GUI anyway). Everything else, find a good TUI. If you can't find one, find a shitty one for emacs and bend it to your will. This keeps (auto-) defenestration at a minimum.
It wouldn't be unconventional if most found it useful. That doesn't mean the author themself doesn't find it useful:
>> I have seen in my own use of the modern browser (chromium/firefox) that it is far too easy to get distracted
They find EWW useful to cut distractions.
>> In addition to this, my belief is that we should make strides toward leaving behind Layer 8 of the internet - the limiting frontends of social platforms and locked away corners of the net that limit actual discourse (Discord, I’m looking at you). [...] The internet is just a delivery mechanism.
They also find it a useful way to stick to their ideals which I understand to be that they should be in control of the front-end and the internet is only to be treated as a delivery mechanism of information.
The main utility I see in browsing the web through emacs is the potential to integrate it into your workflow. You can have your code/notes open in one buffer and have the website in another, which is useful for documentation, or an article you're following along, etc.
With gptel, it's easy to highlight part of a blogpost and ask an llm a question about it, or get it to rewrite or annotate the whole page.
With macros you can do basic automation, like this webscraping example here: https://nickdrozd.github.io/2018/10/17/web-scraping.html
An added bonus is that your keybindings for the rest of emacs are also valid here. (Though for that I've recently started using Glide, a firefox fork, and set all of its keymap to match my emacs bindings)
I’m curious what the difference is between that and having pinned windows in Ubuntu
For example I’ll have sublime text, one terminal, a notepad, and a firefox window with multiple tabs, pinned to the four corners of my desktop
Everything You describe is possible without emacs so I’m genuinely trying to understand the appeal
The difference is the text consumption. Can you easily retrieve any text you're seeing on the screen into your editor?
Inserting any piece of text I see on the screen into my notes? A couple of keystrokes. I can grab text in any form, even if it's a frame in a YT video - I'd just select the area, snapshot it - Emacs OCRs the text out of it, which takes milliseconds. I fully control the playback of said video directly from my editor - rewind, pause, mute, speed up, extract transcript, etc. It is enormously liberating and empowering.
I have complete and utter control over text - I can send a query to an LLM in the middle of typing this very comment, without ever missing a beat. I can ask for anything - fact-check it, spell check it, reword it, explain it, give me different variations of the same sentence. I can replace part of it with diffing in-place. I can even start changing the behavior of Emacs itself, right there on the spot.
I can directly grab any part of the log of a process running in the terminal. Grep through it, semantically search through it with an LLM, whatever. Similarly, I can directly grab a piece of any text in Emacs and send it over to another app, e.g., as a command to run in the terminal.
> Everything You describe is possible without emacs
Everything is possible with Turing-complete languages, but Turing-completeness is a trap. It's all about affordability. What Emacs grants as a Lisp-REPL loop with a built-in text editor is lowering that affordability bar to the lowest imaginable minimum.
Have you ever thought: "What if I could change the 'Open File' dialog in Sublime so I can browse GitHub repos directly instead of folders and files?" This kind of thinking has probably never occurred to you - for an Emacs user, that is some prosaic, just-another-Wednesday kind of thought.¹
---
¹ https://www.youtube.com/watch?v=BDn8ak-8yIs
By the way fishing that url out of my browser history took a keystroke - of course Emacs can search through my browser history too - https://github.com/agzam/browser-hist.el
The example you describe is something I would rarely do so building a process around it would be inefficient for me but if that’s where people want to spend time then great
It's just a drop in a vast ocean of things. Your close-minded thinking just sees one thing and makes generalized assumptions.
> where people want to spend time
You're reading it completely backwards. This is the thing that I never understood about the majority of programmers. For whatever reason they just readily accept the status quo. Here's a typical example:
So often do I feel like rolling my eyes whenever I pair with my colleagues - I'd do something trivial, like fetching a list of PRs related to a specific ticket number at the cursor. They'd be like "whoa, that's cool", and then never do anything about it - they stick to their "learned helplessness" paradigm - copy the ticket number, switch to browser, navigate to Jira, pass through SSO, look around for the phone just to push the button, paste the ticket number, find the linked PRs, etc. The other teammate watching all that may say something like: "you could do it faster directly on GitHub"... And here I am - pressing a key, voila - the list. Why the heck don't they do anything about it, I just don't get it. Trained engineers, they spend years dealing with cranky software, why in the world are they unwilling to do anything about these seemingly small yet everyday annoyances? Why do so many programmers treat software as if it's magic? And I think I know the answer: because the software you use shapes your affordances. If there's no downloadable get-me-list-of-prs-for-a-jira-ticket VSCode extension and no simple way to build it, it would never occur to them to even get annoyed. In Emacs, I can write a picker in a scratch buffer by hand and it would take me minutes. Today with LLMs, I can just spend fifteen seconds, not even typing, speaking my thoughts and it goes to build it, run a sequence of e2e tests and hand it to me on the spot.
Sure, go ahead, talk about time. Make me chuckle. The irony is that the time you spend using a Lisp-based system never feels wasted; it only gets better, it has a compounding effect. It's the time you waste convincing skeptics that it is in fact enormously pragmatic and time-saving that feels like tilting at windmills. Why am I even still talking to you here, if it was clear from the beginning that you didn't in fact have good-will curiosity, only simply wanted to find evidence for your already settled and stiffened mind.
You think people choose and stay with Emacs, because their parents baptize them into it? It's a hands-down the best known system to deal with text of any kind. It is not a text-editor, text-processor, IDE or coding tool. It's an effing text orchestrator. It is capable of handling multitude kinds of text-related tasks in anything in its computational vicinity - local and remote computers, apps, and services. And programming is all about text. That is the essence that every computer user has to conquer first and foremost; everything else is secondary.
The advantage is that it’s a single program with multiple windows with a shared list of buffers. So you can save transform your layout and save it, manipulate the content without any real boundary, unified bookmarks,…
Your WM can’t access sublime and firefox internals. And the last two can’t talk with each other.
But why would I want them to as the default interaction?
I can ctrl-c > ctrl-v from one to the other trivially and ffx has console and inspector internals visible with one Function key press
Although I don't browse the web with Emacs, I see the utility.
Tell me: How would you select some text, and with a keystroke, send the selected text to yourself via email, with the URL included and the title of the page in the subject line?
Some find it useful. One can write code in vim or sublime, the only differences are the specific actions to get it done and the user experience. People will lean towards one another, not for the capability to write code, but for how it’s done.
So what I’m trying to showcase is not the capability, but the Emacs ways of getting it done. You can also implement something with a WM setup or tmux (with w3m). But it’s a different feel and effort, so people like ones and dislike the others.
Last I heard, Apple has each version of macOS certified for POSIX/SUS compliance. Did they stop doing that?
I didn't mean any snark or insult by my original comment btw, just meant non MacOS Unixes
But emacs always felt like a whale.
You can use software like Kanata to accomplish this if you don't have a programmable keyboard.
if you happen to use a thinkpad and don't use the trackpoint, you can remap the two mouse buttons above the trackpad to ctrl and alt using keyd on linux. this completely solved RSI on my laptop.
I've never had this. I'm either lucky, or I know to use both Ctrl keys on the keyboard.
If you're typing C-x, you should use the right control key. The standard ergonomics advice is to use both hands when doing multiple key combinations.
Also, just use the usual cursor keys for navigation!
Do you touch type? You’re meant to use the modifier keys on the opposite side (so use both hands) rather than stretching your pinky to press them. Also there are packages such as god-mode which help.
But the most helpful thing is doing navigation by page instead of line which reduce the use of keybindings.
Well, he has a point - Google controlling so much of the www is bad.
On the other hand, it's not all bad only. While more and more youtube videos are crap, there are useful youtube videos. Many people use it for learning too and often that works as well. Using emacs as browser really does not fix the issue. Also, emacs needs a proper editor if it ever wants to be a good operating system.
I have yet to figure out how to get around those issues.
And a big plus, eww is great for resource limited systems like old 32bit laptops.
I have a very easy solution for that: zero sound output from my main desktop/workstation computer. I mean: technically I leave the sound drivers and the sound on, but there's nothing connected to the output (so speakers / no headphone).
The deep underlying philosophical question being: "If a PC outputs audio to nothing plugged into it, does it make a sound?"
The answer doesn't really matters though. What matters is that it's a great way to not get distracted.
My main desktop emits, on purpose, no sound since... 20 years?
For the rest, I've got laptops.
I have more hope in projects like lem which implement new emacsen from scratch.
Not true in practice. It is absolutely useful in real, battle-tested scenarios. If this were a deal-breaking problem, we'd keep seeing it discussed and eventually fixed. Like the issue with long lines, which was a problem in Emacs for decades and is now mitigated. Nobody complains about it anymore.
Lem is very cool, yet too young to be pragmatically compared with the 40-year package ecosystem of Emacs. Let's be honest, with all the "better" stuff, we're not witnessing herds of Lispers switching to it. Another decade may pass before we see it make more than a blip on the radar. Aside from a number of devoted Common Lispers, there's not much usage of it in the real world.
Meanwhile, Emacs is seeing another uptick. First it was Spacemacs, then Doom, now LLMs are making it very appealing, and we see HN submissions every other week. Don't get me wrong, I'm betting on both of these horses, even on Neomacs, and whichever wins, I'd be happy to see it. Emacs for now remains the uncontested king of Lisp-based text editors.
There are also various multi-processing configurations for even more flexibility: multiple Emacs processes that can coordinate, Emacs modules over a C API which can create native threads outside of Emacs and so on. What Emacs lacks is parallelism inside a single Emacs process, which is arguably not needed and would be a net negative from a safety, ease of development and debugging point of view.
Look at the recently introduced canvas API (https://monadicsheep.org/blog/an-introduction-to-canvas-in-e...) for some examples of what's possible via that flexibility.
Lem is nice but nowhere near as nice as modern Emacs (and probably never will be). Emacs Lisp isn't Common Lisp but has evolved significantly and the remaining differences don't really matter in practice because Emacs momentum, massive user base and ecosystem are a tremendous advantage.
[1] https://www.reindeereffect.com/0002
If you want to put the multi-column mode on massive steroids, check out an old Vim add-on called microViche for inspiration -- "text editing on an infinite pane": https://www.youtube.com/watch?v=9YNiPUTGO28 and https://www.youtube.com/watch?v=xkED6Mv_4bc
The addon itself, FWIW (no idea if it works on a current Vim): https://www.vim.org/scripts/script.php?script_id=4835
I remember trying to use this one as well, because the paradigm/concept is really cool and IMO unusual. I couldn't really get used to it, but it was a fun and inspirational way to make one's brain explode.