I think this is going in the wrong direction. While cool URIs don't change [0], many URLs that you'd like to bookmark are not cool. So for bookmarks to be useful long-term you need to store much more data than just the URL. At the very least you need a timestamp to be able to find the resource you wanted to bookmark on the Internet Archive in the future. But better would be to save a snapshot of the site alongside the bookmark instead. It's a shame that no browser cares to integrate that or other relatively simple usability enhancements instead of blindly copying Chrome's UI (which usually means removing features).
I don't think this project is necessarily going in the wrong direction, but I agree that keeping track of web URLs requires more data than just the URL and title.
At the very least, I personally need tags. Tree hierarchies are too limiting, but tags allow adding free-form metadata which can be used to quickly find anything. Often I don't remember the actual URL or title of a web site, but I'm looking for a general category, and am usually able to find what I'm looking for. This does require a certain discipline from the user to use tags consistently, but they work well IME.
These days with ML tools it's easier than ever to assign relevant tags to URLs automatically, though I would personally not rely on them because of their propensity to hallucinate and not follow conventions.
And, like you say, web sites change and disappear constantly, so snapshots are a must as well.
I've been meaning to try SingleFile/ArchiveBox/Wallabag for months now, but I'm currently too comfortable with Pinboard. Pinboard has been working great for me for many years, contrary to the negative buzz around it, but I don't have confidence that it will stick around forever. I just hope that I can export and migrate all my snapshot data as well...
Bookmarks are something I have grown tired of. From the days of del.icio.us, I started collecting bookmarks and it kind of trickled along almost until Pinboard went on life support (or something else if you'd prefer to call that).
The thing is - I just saved bookmarks, I never really utilised them ever, to find something, to go back to. I can remember once or twice and either I couldn't find anything among my bookmarks or the sites were long gone. I really don't think I personally had to consult my thousands of bookmarks (which I have now dutifully migrated to Raindrop of course, because why the hell not) in any useful sense ever. I paid for a couple of archiving services as well before realising "nah, I don't really need that, nor this recurring outgoing payment in my life".
So like a lot of things on the Internet, I guess I did "bookmarking things" just for the sake of doing "bookmarking things".
That reminds me of note-taking. There was a time when I used to do "note-taking exploration and research" and never really took any notes or, hell, even needed them. When I started note-taking, while I still keep an eye out for a decent app, I just pick a decent or half-decent note-taking app and I just take notes. Oh, backup and sync tools and services. Those too - there was "explore and research" and now there's "just use something damnit". "TODOing" to, yes! I am sure this tool (or philosophy? style? bookmarking architecture?) is very nice and novel.
This is not at all reflecting on why or why not one should do such "things", I absolutely believe this is good and sometimes in fact results in tools/services massively good, I am just talking about this out loud wondering whether it's just me or this kind of fatigue really sets in for other people as well.
> The thing is - I just saved bookmarks, I never really utilised them ever, to find something, to go back to.
You're not describing a bookmarks issue. You're describing a personal organization issue, which is reflected on how you manage bookmarks.
You're voicing the exact same sort of complains often directed at todo lists. In fact, from your description you're implicitly treating bookmarks as ad-hoc Todo lists, and you're complaining your To-do backlog is growing.
Like others, you can blame bookmarks and Todo lists for your growing backlog of things you want to do but never get around to doing. Those are not the problem though, and only reflect a symptom caused by the actual problem.
> So like a lot of things on the Internet, I guess I did "bookmarking things" just for the sake of doing "bookmarking things".
You're describing a symptom of your problem. The fact that it extends beyond bookmarks is a telltale sign.
> This is not at all reflecting on why or why not one should do such "things", I absolutely believe this is good and sometimes in fact results in tools/services massively good, I am just talking about this out loud wondering whether it's just me or this kind of fatigue really sets in for other people as well.
I believe you're expressing the same issues expressed by those who have trouble managing their task queue. Your problem reflects on bookmarks, on personal notes, on productivity software, etc. This means your problem is not bookmarks, or personal notes, or productivity software. It's something else that is reflected across tools and systems.
I have a huge bookmark collection, but I don't care if I saved a bookmark that I never opened. I configured my browsers to exclusively use bookmarks in recommendations, thus serving as an ad-hoc search engine of noteworthy links I visited or want to visit. If I don't visit any of the links, that's too bad. Why would it present a problem?
Bookmarks are useful for address bar autocomplete back to a site I know I visit semi-regularly, but for whatever reason disappeared out of my browser's autocomplete history. E.g. this might happen for social media sites that I only ever visit inside an incognito tab to try (probably in vain) to mitigate being tracked across visits. But also it can happen when I try a new browser or my phone resets itself or whatever. Then I just export the bookmarks.html from my laptop and import it on whatever other device.
I never bothered to look at the schema of the bookmarks.html, because feels like it's worked the same for 20+ years. I used to care a lot about housekeeping the structure, but it doesn't really matter, as long as they're in the bucket the browser will use them for autocomplete suggestions, so...
Totally agree with you I have had the same experience. Most of the bookmarks I saved over the years, I never went back to. These days I usually just copy content, run a quick summary through ChatGPT, and if its useful I keep it as a note. That way I dont have to keep deferring things in an endless bookmark pile.
Now I mostly keep two kinds of bookmarks: quick-access ones for work (like repos I contribute to or PR sections I need to check often), and then more organized notes for ideas, projects, or interests I want to revisit later. To make that easier, I use a little tool I put together (beavergrow.com) where I can group bookmarks into blocks and keep notes alongside them—it’s been handy for giving some structure without overcomplicating things.
They have always been useful to me. Whenever I had problems, it turns out because I hadn't bookmarked enough. In general I bookmark for anything that I knew once and found difficult to search for the next time when I forgot about it.
For some people, they have no need for bookmarks. For other people, bookmarks may be useful but the implementation is not.
Reading the author's description made me realize how unbookmark-like bookmarks actually are. The current implementations are somewhat akin to creating a list of books that you like at the library. It's not so much a pointer to the information you found useful, as it is a list of books you found useful. You still have to do some digging when you go back for the book. If the book is lost, you end up having a reference to something that you cannot obtain. And if you just add books to the end of your list, you still end up having to search through the list. The only way around that is to spend time organizing your list. It's no wonder why bookmarks are useless to so many people.
The author doesn't really solve the problem with bookmarks, except for one. The last one. By sticking a bookmarks file in a project directory, at least you're only searching through a list of bookmarks relevant to the project. If you are no longer interested in the project and delete it, you're also getting rid of bookmarks that you (hopefully) no longer need. It also addresses the portability of bookmarks. As far as I can tell, the only way to move bookmarks between any of the major browsers involves the use of special software or network services. Look at moving bookmarks from one Firefox installation to another: you either use online sync, export to HTML to import from HTML, or import the database (which replaces your current bookmarks with the ones being imported).
> You still have to do some digging when you go back for the book.
There is widespread browser support for linking directly to text fragments[1] which makes it possible to link to arbitrary parts of documents even when the author hasn't marked up some nearby element with an id to target, like so: https://news.ycombinator.com/item?id=45047572#:~:text=You%20...
Unfortunately, Firefox provides no convenient way of creating such links, but Chromium has a "Copy link to highlight" context menu entry when you have text highlighted. Neither Firefox nor Chromium provide a convenient way to create a bookmark to a text fragment.
> As far as I can tell, the only way to move bookmarks between any of the major browsers involves the use of special software or network services.
Both Firefox and Chromium support importing/exporting bookmarks from/to "HTML" (really, Netscape bookmark file format[2]).
Both browsers also provide the means to organize bookmarks into folders, and Firefox lets you add arbitrary tags to them as well. Alas as you say, the only way is to spend time manually organizing them. Automatically suggesting folders/tags (preferring ones you already have created) seems like an ideal use case for LLMs or other NLP tools. Ideally browsers would offer an option to save a snapshot of the page together with the bookmark, that would guard against link rot and enable full text search. We have the technology, it's really only a matter of improving the UI and linking the two features together. Too bad hamstringing adblockers, gimping sites that rely on XSLT, and implementing WebBluetooth or whatever has higher priority.
I see your point. Also, maybe to add to it, I should have been rather judicious in collecting bookmarks.
And — possibly to also literally keep them inside the browser’s default bookmarks/favourites whatever browser one uses. Not on some fancy service with AI and what not.
I bookmark things so I can look them up when I search in the addressbar. If it’s bookmarked, the result is higher than my search engine results— making them quicker to find again.
This resonated because I realized my relationship to bookmarks was different - I don't save bookmarks, I save or want to remember sentences and saving a bookmark wasn't the best way to do that.
I try to read little I am not looking to apply, or be conscious it's for pleasure/interest
If I bookmark something, I consider it unread. If I read something, I make sure I bookmark and annotate it and tag it to make my mind more actively work with what I'm reading (and make it easer to find.)
The result? 10-15 years of every link I've ever saved, organized and annotated by me. Chronological, sorted, I can see what I was paying attention to chronologically, or by topic, and at any time search any of my highlights and notes.
This is the nice part because it isn't an AI tool, but maybe something that can feed into an AI tool quite nice. My curation, where relevant, as input.
Best of all, it just works. It's not heavy or tedious, anything that has my attention, gets my attention.
The one thing a text only approach will not solve is that URIs while universally defined will not perpetually stay online.
Diigo, and other tools like it allow you to save your own cache, or perhaps submit to a public cache that page so once it invariably goes offline, it doesn't.
There's lots of tools out there to help with this each person's way, I liked diigo.com, but lately think tools like logseq with a few basic plugins are offering a lot of promise to directly save a bookmark, whatever snippets are relevant, and they are always and instantly searchable.
Not really the same, but I find bookmark folders to be pretty useful. I have a "daily", "weekly", "taxes", etc., and it's just one click to open them all up in one go. They could easily be converted into a cron job with a plaintext URL list like TFA.
and then write a program which detects a randomly opened bookmark and closes it because i'm in the middle of something important and can't be bothered right now.
To combat link rot I use an applet which looks up a link at archive.org and when I visit an interesting site (of supposedly longer lasting value) I check if it's archived already. If not I ask the way back machine to store it.
I spend most of my days wading through the web. When i come across something that excites me I share the link to apple
Notes. I dont write anything. Its just a bookmark. In February I spent a week sorting 5000 links into folders. It was quite satisfying to guess why i saved things, find patterns, create an organizational structure etc. the activity yielded both motivation and direction for my personal projects.
Given that the aim is to still visit the URL, this appears to be browser bookmarks with extra steps. I am not seeing an advantage to the external storage mechanism and the overhead it brings.
Browsers have been doing an excellent job of managing bookmarks, you can tag and search for them from the address bar itself which is very convenient.
Well, actually I could see a use for this in specific context and use cases, for instance if you happen to have different dev environments you're able to just move from machine to machine while keeping all the bookmarks in the repository.
If you organise them you can even reference them from the codebase, or the documentation to avoid clutter. The format is simple and dumb enough so that a simple bookmark.txt can be converted into a dictionary, array that can be used in the program if some URLs are supposed to be used there.
It's not revolutionary by any means, but I have to confess that it didn't occur to me that's a great per repo documentation reference tool or per folder.
Is there a reliable standard to do this across browser vendors and versions and not have to rely on a proprietary cloud implementation on somebody's server?
we do something similar. IntelliJ (ultimate) has a text-based http client included. Like postman but you can commit .http files to the repo. Then this client has variables which can come from a json file. So basically all URLs that our software uses, are in that file. Eventually this led to the json file being used by scripts too.
It has been very useful. URLs are super easy to modify, super easy to share, super easy to open, add notes etc. Having a gist adds the ability to share a set or URLs and people can comment on them.
I could (if i want to) use github API and a browser extension to put bookmarks directly in a gist as plaintext
(Probably skip if you don't want to read a ~rant..)
Questions:
- New "format"? It is a CSV with spaces as separators and no header? Why?
- Global file located in the home directory? Why would you want another file to pollute the $HOME? Why not just $BOOKMARKS_FILE or something?
- Local bookmarks file - why don't just reuse README.md? Add 'Resources' section and chuck markdown links in it, parsing that is not (much) harder, and you get nothing more with a separate 'bookmarks.txt' file. As some other comments pointed out, there are no tags, no (semantic) grouping, no content archiving, etc... This is just a list of URL with optional title attached to them, this is not a distinct 'concept'...
I never really used browser bookmarks until I discovered Firefox's bookmark keyword feature [1]. It basically eliminates the search step. For example, I set "pr" to the GitHub pull request page. Just press Alt + D to focus the address bar, type the keyword, and hit Enter.
I configured Firefox to only ever auto-complete to bookmarks or open tabs, never some random thing from my browser history. That also makes it very easy to find specific pages and I do not have to memorize any tags.
Mozilla-based browsers used to (before Firefox 3, I believe) store bookmarks as HTML (bookmarks.html) instead of using an SQLite database. It still uses a single HTML file when exporting or importing bookmarks.
I have my own Chrome/Firefox extension that shows bookmarks in the new tab page ordered by recency with a search field on top, it's pretty good for me. Every time I open a new tab, I see the latest bookmarks I've added. Granted, I don't usually bother to browse older ones, but at least I see ~15 new ones on (desktop monitor) every day.
Also, I sometimes add a snippet of text from the content to the title of the bookmark, making it easier to search afterwards.
I think that is a good idea to keep URLs as text. I use the Tab Session Manager extension which allows me to export saved sessions (title, URL, with a little bit of extra stuff) as text (json actually, a little bit too verbose) and I also use Eagle Filer to save entire web pages, just in case it disappears from the web, which happens quite regularly. What I want is the title of the web page associated with its URL and maybe extra keywords. I don't use classic browser bookmarks much anymore.
I have become somewhat disillusioned with maintaining a list of bookmarks, as I often discover that the URLs are no longer functional when revisiting pages that have been bookmarked for an extended period.
I think that people who say they prefer keeping a simple file has never found a good app for bookmarking. There is still a space for killer app with bookmarks.
I think that simple URL bookmarking is just wrong. It simply will not work for big bookmarking data sets. The key is using tags, and rating system, and automatic update which checks if URLs are even valid any more.
I also thought that we miss a killer RSS app.
That is why I created my own self-hosted app.
- it can store bookmarks
- it gather news through RSS
- it provides tags (I can search bookmarks by tags)
- it provides user ratings (I can filter using it too)
- I can filter, or order by link, date of publish, date of creation, etc. etc.
- It checks if links are rotten (and marks them)
- I can mark link to read it for later
- I can see how many times I have visited a link
- I can check 'related links' to jump to things I have jump before from this link
On the other hand, I am quite certain that I use it, because it is 'tailored for me'. I am not that interested in the looks. I know how it works
I've made my own private bookmark browser extension that syncs e2ee across my devices. Because it's an extension I can open bookmarks automatically in a private tab and much more.
After some years of using this extension every day I decided to make it available for others as well. You can find it here, free forever: markbook.io This was just a side project I did in about 6 months of my spare time, definitely not a polished product that's trying to become the next billion dollar enterprise.
Just in case, If you want to index the content of these bookmarks for searching maybe my little wip tool face is useful to you. As it’s also a local server you can use a keyword search in browsers for it. https://github.com/svandragt/fafi
I needed to keep the highlighted text as well. With ChatGPT I have created an extension [1] that lets me capture highlights from any webpage and send them directly to a Notion database along with the page URL, capture date, and a per-selected project tag.
Scrapbook was (is?) great, since it does everything bookmarks do plus saves a copy of the page just the way it currently looks and makes that available in the browser.
I switched to saving pages using SinglePage instead, that saves the current page as a single stand-alone HTML file. It loses the bookmark-like features, but I can sort those saved files easier in my file system to keep them like any other downloaded documents on various topics. Each file also by default has a comment near the top with its original URL, so it would be easy to write a script to find all of those and build something like bookmarks.html.
This is needed, TIL but only till recently did Firefox remove their bookmarks limit, it was something like 6000 about 3 years ago.
(For those curious about why one needs so many bookmarks, similar to maybe some other people, I use the bookmark feature as a "like" to save articles/URLs I find interesting).
I actually do this. I put urls and files in a text file. Which I parse with rofi. The reason why I do this is so I can just open rofi with a shortcut and type out the name of the file or url and instantly open it.
Same, been doing this since the start of the web. I have decades of web pages I've found interesting over the years, and its all offline/local searchable and will be there for me when the grid goes down.
I'm part-way through getting the 90,000+ PDF files collected in this manner, analyzed by an LLM so I can .. query it about my own interests, I guess? ;)
I don't find that saving URL's is very productive - they are the dangling pointer of the web. Far better to have your own cache of docs to refer to imho ..
For my CTO newsletter I use raindrop.io to store interesting articles I encounter, export them to CSV (like bookmarks.txt), sort, filter and remove 90%, convert them to my own format, write my content and convert them to Markdown and then to HTML with Hugo.
I’m using a similar format to keep my links organized on an iPhone. A bookmarks.txt note kept in Apple Notes with markdown # headings to separate each category, allowing me to not keep open 500 tabs on Safari.
Let me check... 1331 open tabs in Safari on my iPad. Several text files with thousands of bookmarks, with various attempts at grouping. Since I cannot save myself, I save them.
I've been doing exactly this for ten years, keeping the bookmarks file in Git. I added 70 bookmarks today, which is more than average. But rather than just a "title" I have whatever seems like I might want to search for later, including if the link rots. So it's more of a summmary. With hashtags. In Markdown (though without newlines).
Each new day gets a header.
Here's Sunday, with blank lines added to get reasonable formatting on HN:
https://youtu.be/Nn5v59l2Xec?t=64 VEMAG brand double-screw extruder screw pump #mechanisms pumping M&Ms and ground meat. #video footage of the parts being washed.
https://github.com/3b1b/manim Manim math #animation software for morphing equations and plots and stuff into each other, using FFmpeg, OpenGL, LaTeX, and Pango
https://docs.kernel.org/filesystems/f2fs.html "f2fs" is the #Flash friendly #filesystem for #zoned-storage. “F2FS is a file system exploiting NAND flash memory-based storage devices, which is based on Log-structured File System (#LFS). The design has been focused on addressing the fundamental issues in LFS, which are snowball effect of wandering tree and high cleaning overhead.”
———⁂———
Here's today, ten years ago, so you can see how my bookmarking style has developed:
links from 2015-08-28:
http://www.excamera.com/sphinx/article-j1a-swapforth.html A free-software #Forth operating system running on the J1a open-source CPU running on a Lattice #FPGA thanks to Project #IceStorm’s reverse engineering and the resulting free-software #synthesis and programming toolchain, running on the Lattice iCEstick evaluation board with 8K of RAM. #J1 #hardware
https://www.youtube.com/watch?v=uRVaLUQUmA8&list=PLACB124F79... A 2009 #video lecture series on #mecheng manufacturing processes by “nptelhrd”, Prof. Inderdeep Singh at IIT Roorkee. Some problems with the audio, seems to be a bit clipped. However, the video is entirely a talking head and a bunch of PowerPoint slides, so the only way you would watch this video to learn about powder metallurgy instead of reading a book is if you are dyslexic or have a beard fetish. At least it doesn’t have shitty elevator music.
http://minerals.usgs.gov/minerals/pubs/commodity/titanium/mc... #Titanium #pricing in the US has gone from US$9.62/kg in 2010 to US$11.20/kg in 2014. Nearby pages use dollars per pound, but the prices per kg show a sharp reduction from its peak of US$17.28 in 2005, but it had a low of US$6.50 in 2003, US$9.70 in 1995, and US$8.26 in 1992–3. This means that the FFC Cambridge Process is not in production yet.
http://pubs.usgs.gov/sir/2012/5188/ A #PDF with a bunch of metal #materials #pricing for 1970–2010. Comprehensive, covers nearly all metals and some semimetals. Unfortunately at least some prices use folk units like pounds instead of SI.
Perhaps surprisingly, of those 8 links from ten years ago, only one has really linkrotted; unsurprisingly it's Solnit's essay. But it survives in the Archive. The titanum PDF redirects to the general titanium Mineral Commodities Summary page.
Emacs full-text search is surprisingly often effective at finding relevant information. Grep (or M-x occur, its moral equivalent) works even more often. F6 opens the URL in my browser. Emacs C-s can search those 12000 bookmarks faster than I can type.
At some point in the past I wrote a Chrome extension that would show me all the links for a given hashtag with their descriptions (formatted as Markdown), with the hashtags being links to other similar hashtag pages, and also let me bookmark pages from within the browser interface, but I don't know where it is now. It wouldn't work with Manifest v3 anyway.
[0] https://www.w3.org/Provider/Style/URI
At the very least, I personally need tags. Tree hierarchies are too limiting, but tags allow adding free-form metadata which can be used to quickly find anything. Often I don't remember the actual URL or title of a web site, but I'm looking for a general category, and am usually able to find what I'm looking for. This does require a certain discipline from the user to use tags consistently, but they work well IME.
These days with ML tools it's easier than ever to assign relevant tags to URLs automatically, though I would personally not rely on them because of their propensity to hallucinate and not follow conventions.
And, like you say, web sites change and disappear constantly, so snapshots are a must as well.
I've been meaning to try SingleFile/ArchiveBox/Wallabag for months now, but I'm currently too comfortable with Pinboard. Pinboard has been working great for me for many years, contrary to the negative buzz around it, but I don't have confidence that it will stick around forever. I just hope that I can export and migrate all my snapshot data as well...
Logseq, etc, are tools that help facilitate this, albeit in not plain text format, but close.
The thing is - I just saved bookmarks, I never really utilised them ever, to find something, to go back to. I can remember once or twice and either I couldn't find anything among my bookmarks or the sites were long gone. I really don't think I personally had to consult my thousands of bookmarks (which I have now dutifully migrated to Raindrop of course, because why the hell not) in any useful sense ever. I paid for a couple of archiving services as well before realising "nah, I don't really need that, nor this recurring outgoing payment in my life".
So like a lot of things on the Internet, I guess I did "bookmarking things" just for the sake of doing "bookmarking things".
That reminds me of note-taking. There was a time when I used to do "note-taking exploration and research" and never really took any notes or, hell, even needed them. When I started note-taking, while I still keep an eye out for a decent app, I just pick a decent or half-decent note-taking app and I just take notes. Oh, backup and sync tools and services. Those too - there was "explore and research" and now there's "just use something damnit". "TODOing" to, yes! I am sure this tool (or philosophy? style? bookmarking architecture?) is very nice and novel.
This is not at all reflecting on why or why not one should do such "things", I absolutely believe this is good and sometimes in fact results in tools/services massively good, I am just talking about this out loud wondering whether it's just me or this kind of fatigue really sets in for other people as well.
You're not describing a bookmarks issue. You're describing a personal organization issue, which is reflected on how you manage bookmarks.
You're voicing the exact same sort of complains often directed at todo lists. In fact, from your description you're implicitly treating bookmarks as ad-hoc Todo lists, and you're complaining your To-do backlog is growing.
Like others, you can blame bookmarks and Todo lists for your growing backlog of things you want to do but never get around to doing. Those are not the problem though, and only reflect a symptom caused by the actual problem.
> So like a lot of things on the Internet, I guess I did "bookmarking things" just for the sake of doing "bookmarking things".
You're describing a symptom of your problem. The fact that it extends beyond bookmarks is a telltale sign.
> This is not at all reflecting on why or why not one should do such "things", I absolutely believe this is good and sometimes in fact results in tools/services massively good, I am just talking about this out loud wondering whether it's just me or this kind of fatigue really sets in for other people as well.
I believe you're expressing the same issues expressed by those who have trouble managing their task queue. Your problem reflects on bookmarks, on personal notes, on productivity software, etc. This means your problem is not bookmarks, or personal notes, or productivity software. It's something else that is reflected across tools and systems.
I have a huge bookmark collection, but I don't care if I saved a bookmark that I never opened. I configured my browsers to exclusively use bookmarks in recommendations, thus serving as an ad-hoc search engine of noteworthy links I visited or want to visit. If I don't visit any of the links, that's too bad. Why would it present a problem?
I never bothered to look at the schema of the bookmarks.html, because feels like it's worked the same for 20+ years. I used to care a lot about housekeeping the structure, but it doesn't really matter, as long as they're in the bucket the browser will use them for autocomplete suggestions, so...
Now I mostly keep two kinds of bookmarks: quick-access ones for work (like repos I contribute to or PR sections I need to check often), and then more organized notes for ideas, projects, or interests I want to revisit later. To make that easier, I use a little tool I put together (beavergrow.com) where I can group bookmarks into blocks and keep notes alongside them—it’s been handy for giving some structure without overcomplicating things.
Reading the author's description made me realize how unbookmark-like bookmarks actually are. The current implementations are somewhat akin to creating a list of books that you like at the library. It's not so much a pointer to the information you found useful, as it is a list of books you found useful. You still have to do some digging when you go back for the book. If the book is lost, you end up having a reference to something that you cannot obtain. And if you just add books to the end of your list, you still end up having to search through the list. The only way around that is to spend time organizing your list. It's no wonder why bookmarks are useless to so many people.
The author doesn't really solve the problem with bookmarks, except for one. The last one. By sticking a bookmarks file in a project directory, at least you're only searching through a list of bookmarks relevant to the project. If you are no longer interested in the project and delete it, you're also getting rid of bookmarks that you (hopefully) no longer need. It also addresses the portability of bookmarks. As far as I can tell, the only way to move bookmarks between any of the major browsers involves the use of special software or network services. Look at moving bookmarks from one Firefox installation to another: you either use online sync, export to HTML to import from HTML, or import the database (which replaces your current bookmarks with the ones being imported).
There is widespread browser support for linking directly to text fragments[1] which makes it possible to link to arbitrary parts of documents even when the author hasn't marked up some nearby element with an id to target, like so: https://news.ycombinator.com/item?id=45047572#:~:text=You%20...
Unfortunately, Firefox provides no convenient way of creating such links, but Chromium has a "Copy link to highlight" context menu entry when you have text highlighted. Neither Firefox nor Chromium provide a convenient way to create a bookmark to a text fragment.
> As far as I can tell, the only way to move bookmarks between any of the major browsers involves the use of special software or network services.
Both Firefox and Chromium support importing/exporting bookmarks from/to "HTML" (really, Netscape bookmark file format[2]).
Both browsers also provide the means to organize bookmarks into folders, and Firefox lets you add arbitrary tags to them as well. Alas as you say, the only way is to spend time manually organizing them. Automatically suggesting folders/tags (preferring ones you already have created) seems like an ideal use case for LLMs or other NLP tools. Ideally browsers would offer an option to save a snapshot of the page together with the bookmark, that would guard against link rot and enable full text search. We have the technology, it's really only a matter of improving the UI and linking the two features together. Too bad hamstringing adblockers, gimping sites that rely on XSLT, and implementing WebBluetooth or whatever has higher priority.
[1] https://developer.mozilla.org/en-US/docs/Web/URI/Reference/F... [2] https://learn.microsoft.com/en-us/previous-versions/windows/...
And — possibly to also literally keep them inside the browser’s default bookmarks/favourites whatever browser one uses. Not on some fancy service with AI and what not.
I try to read little I am not looking to apply, or be conscious it's for pleasure/interest
If I bookmark something, I consider it unread. If I read something, I make sure I bookmark and annotate it and tag it to make my mind more actively work with what I'm reading (and make it easer to find.)
The result? 10-15 years of every link I've ever saved, organized and annotated by me. Chronological, sorted, I can see what I was paying attention to chronologically, or by topic, and at any time search any of my highlights and notes.
This is the nice part because it isn't an AI tool, but maybe something that can feed into an AI tool quite nice. My curation, where relevant, as input.
Best of all, it just works. It's not heavy or tedious, anything that has my attention, gets my attention.
The one thing a text only approach will not solve is that URIs while universally defined will not perpetually stay online.
Diigo, and other tools like it allow you to save your own cache, or perhaps submit to a public cache that page so once it invariably goes offline, it doesn't.
There's lots of tools out there to help with this each person's way, I liked diigo.com, but lately think tools like logseq with a few basic plugins are offering a lot of promise to directly save a bookmark, whatever snippets are relevant, and they are always and instantly searchable.
Browsers have been doing an excellent job of managing bookmarks, you can tag and search for them from the address bar itself which is very convenient.
If you organise them you can even reference them from the codebase, or the documentation to avoid clutter. The format is simple and dumb enough so that a simple bookmark.txt can be converted into a dictionary, array that can be used in the program if some URLs are supposed to be used there.
It's not revolutionary by any means, but I have to confess that it didn't occur to me that's a great per repo documentation reference tool or per folder.
Mine is online (rendered client-side) so I can access it from anywhere. https://start.oinam.com
Just in case, anyone wants to look the source is at https://github.com/oinam/start
It has been very useful. URLs are super easy to modify, super easy to share, super easy to open, add notes etc. Having a gist adds the ability to share a set or URLs and people can comment on them.
I could (if i want to) use github API and a browser extension to put bookmarks directly in a gist as plaintext
[1] https://support.mozilla.org/en-US/kb/bookmarks-firefox#w_how...
Saves me a middleman request.
Saving the earth one request (less) at a time!
The file it generates has:
Also, I sometimes add a snippet of text from the content to the title of the bookmark, making it easier to search afterwards.
[1] https://darekkay.com/static-marks/
reading through this thread I’m seeing people use bookmarks to save dozens of new URLs per day, which is very surprising to me
I think that simple URL bookmarking is just wrong. It simply will not work for big bookmarking data sets. The key is using tags, and rating system, and automatic update which checks if URLs are even valid any more.
I also thought that we miss a killer RSS app.
That is why I created my own self-hosted app.
- it can store bookmarks
- it gather news through RSS
- it provides tags (I can search bookmarks by tags)
- it provides user ratings (I can filter using it too)
- I can filter, or order by link, date of publish, date of creation, etc. etc.
- It checks if links are rotten (and marks them)
- I can mark link to read it for later
- I can see how many times I have visited a link
- I can check 'related links' to jump to things I have jump before from this link
On the other hand, I am quite certain that I use it, because it is 'tailored for me'. I am not that interested in the looks. I know how it works
- https://rumca-js.github.io/search - demo search
- https://rumca-js.github.io/music - demo music
- https://rumca-js.github.io/bookmarks - demo bookmarks
- https://github.com/rumca-js/RSS-Link-Database - database of bookmarks
- https://github.com/rumca-js/Internet-Places-Database - link meta information
- https://github.com/rumca-js/Django-link-archive - main crawling engine using for all databases
After some years of using this extension every day I decided to make it available for others as well. You can find it here, free forever: markbook.io This was just a side project I did in about 6 months of my spare time, definitely not a polished product that's trying to become the next billion dollar enterprise.
[1]https://github.com/sea2ocean/keeper
I switched to saving pages using SinglePage instead, that saves the current page as a single stand-alone HTML file. It loses the bookmark-like features, but I can sort those saved files easier in my file system to keep them like any other downloaded documents on various topics. Each file also by default has a comment near the top with its original URL, so it would be easy to write a script to find all of those and build something like bookmarks.html.
No link rot + it's available without internet.
(For those curious about why one needs so many bookmarks, similar to maybe some other people, I use the bookmark feature as a "like" to save articles/URLs I find interesting).
For frequent websites I can remember their TLD and navigate to the panel I wanted
For content saving a URL is not enough. Gotta save the full page or it might be gone very soon.
I'm part-way through getting the 90,000+ PDF files collected in this manner, analyzed by an LLM so I can .. query it about my own interests, I guess? ;)
I don't find that saving URL's is very productive - they are the dangling pointer of the web. Far better to have your own cache of docs to refer to imho ..
For my CTO newsletter I use raindrop.io to store interesting articles I encounter, export them to CSV (like bookmarks.txt), sort, filter and remove 90%, convert them to my own format, write my content and convert them to Markdown and then to HTML with Hugo.
Each new day gets a header.
Here's Sunday, with blank lines added to get reasonable formatting on HN:
links from 02025-08-24:
https://archive.org/details/TheDesignOfSwitchingCircuits/pag... My childhood book on digital logic (including a little #electronics) as an #ebook. #hardware
https://en.wikipedia.org/wiki/High-intensity_discharge_lamp sodium-arc and mercury-arc lamps are also HID lamps. #hardware
https://archive.org/details/ge-glow-lamp-manual-1966/page/n1... scan of GE’s glow-lamp manual for neon lamps from 01966. #ebook #hardware #electronics #history
https://youtu.be/Nn5v59l2Xec?t=64 VEMAG brand double-screw extruder screw pump #mechanisms pumping M&Ms and ground meat. #video footage of the parts being washed.
https://www.radiomuseum.org/r/aleksandro_arz_51_arz51.html very pretty old radio
https://www.nature.com/articles/nmat2141 "Superlenses" to overcome the diffraction limit, #optics #paper from 02008
https://en.wikipedia.org/wiki/Electrical_length#Electrical_l... electrically long and short #antennas and loading coils and whatnot. #electronics #communications #radio
https://www.ornl.gov/publication/evaluation-power-fluidic-pu... #fluidics for pumping in #molten-salt #nuclear-reactors (just an abstract)
https://www.hopefulmons.com/p/in-defense-of-tech-trees trying to use "tech trees" to understand the #history of technological development
https://en.wikipedia.org/wiki/Fresnel_diffraction near-field #diffraction is "Fresnel diffraction" #optics
https://psi329.cankaya.edu.tr/uploads/files/Lewis-PernFascis... Was #Perón a fascist? #fascism #history #PDF #paper #toread
https://lwn.net/Articles/1030818/ #Treacherous-Computing for “confidential VMs” #privacy despite #Linux #virtualization #toread
https://news.ycombinator.com/item?id=44950482 discussion of alternative ways to run graphical apps inside #Docker, including maybe drawing in a web browser
https://news.ycombinator.com/item?id=44991638 my post about how #Rust’s approach to handling #Unicode introduces unnecessary bugs into command-line programs
https://www.youtube.com/watch?v=WApL1EL2GMk #Frondizi speaks about #Perón. #history #Argentina #video
https://github.com/3b1b/manim Manim math #animation software for morphing equations and plots and stuff into each other, using FFmpeg, OpenGL, LaTeX, and Pango
https://nvmexpress.org/wp-content/uploads/September-2020_NVM... in ZNS #zoned-storage a zone must be a power-of-2 number of LBAs, unlike in ZAC and ZBC #PDF #toread
https://docs.kernel.org/filesystems/f2fs.html "f2fs" is the #Flash friendly #filesystem for #zoned-storage. “F2FS is a file system exploiting NAND flash memory-based storage devices, which is based on Log-structured File System (#LFS). The design has been focused on addressing the fundamental issues in LFS, which are snowball effect of wandering tree and high cleaning overhead.”
———⁂———
Here's today, ten years ago, so you can see how my bookmarking style has developed:
links from 2015-08-28:
http://www.excamera.com/sphinx/article-j1a-swapforth.html A free-software #Forth operating system running on the J1a open-source CPU running on a Lattice #FPGA thanks to Project #IceStorm’s reverse engineering and the resulting free-software #synthesis and programming toolchain, running on the Lattice iCEstick evaluation board with 8K of RAM. #J1 #hardware
https://www.youtube.com/watch?v=uRVaLUQUmA8&list=PLACB124F79... A 2009 #video lecture series on #mecheng manufacturing processes by “nptelhrd”, Prof. Inderdeep Singh at IIT Roorkee. Some problems with the audio, seems to be a bit clipped. However, the video is entirely a talking head and a bunch of PowerPoint slides, so the only way you would watch this video to learn about powder metallurgy instead of reading a book is if you are dyslexic or have a beard fetish. At least it doesn’t have shitty elevator music.
http://minerals.usgs.gov/minerals/pubs/commodity/ The USGS gives #pricing information on a variety of mineral #materials, including some historical stuff. #minerals
http://minerals.usgs.gov/minerals/pubs/commodity/titanium/mc... #Titanium #pricing in the US has gone from US$9.62/kg in 2010 to US$11.20/kg in 2014. Nearby pages use dollars per pound, but the prices per kg show a sharp reduction from its peak of US$17.28 in 2005, but it had a low of US$6.50 in 2003, US$9.70 in 1995, and US$8.26 in 1992–3. This means that the FFC Cambridge Process is not in production yet.
http://pubs.usgs.gov/sir/2012/5188/ A #PDF with a bunch of metal #materials #pricing for 1970–2010. Comprehensive, covers nearly all metals and some semimetals. Unfortunately at least some prices use folk units like pounds instead of SI.
http://rebeccasolnit.net/essay/a-rape-a-minute-a-thousand-co... Rebecca Solnit talks a bit about #rape and #intimate-partner-violence. #feminism
https://www.federalregister.gov/articles/2015/08/06/2015-184... The FCC is proposing to outlaw #free-software for wireless firmware.
https://codegolf.stackexchange.com/questions/4732/emulate-an... Programmers competing to do Intel 8086 (subset) #emulation in different languages, including a C program in 348 lines.
———⁂———
Perhaps surprisingly, of those 8 links from ten years ago, only one has really linkrotted; unsurprisingly it's Solnit's essay. But it survives in the Archive. The titanum PDF redirects to the general titanium Mineral Commodities Summary page.
Emacs full-text search is surprisingly often effective at finding relevant information. Grep (or M-x occur, its moral equivalent) works even more often. F6 opens the URL in my browser. Emacs C-s can search those 12000 bookmarks faster than I can type.
At some point in the past I wrote a Chrome extension that would show me all the links for a given hashtag with their descriptions (formatted as Markdown), with the hashtags being links to other similar hashtag pages, and also let me bookmark pages from within the browser interface, but I don't know where it is now. It wouldn't work with Manifest v3 anyway.