Ted Nelson had many pioneering ideas, but we need to acknowledge that many of them _didn't work_. Not because Tim Berners-Lee lacked the time to code them, but they just couldn't _ever_ work.
Nelson was 30 years late delivering Project Xanadu when Berners-Lee jettisoned the _impossible_ parts of Ted's vision and ate his lunch with the WorldWideWeb. [0]
Ted dreamed of these bidirectional links. Not just that you got _notified_ if someone linked to you, for which we have pingbacks (which are massively exploitable for spam if you auto-publish them), but if someone changed their document's address or title, they'd have the power to update _your_ document with the link to it. No bitrot! All it takes is one entity having complete executive control over the entire WWW, and probably the ability to change anything on any page without your knowledge, or even better, give remote control of your documents to anyone who controls the server you happen to link to.
He had other ideas, like you'd write a page and get automatically paid micropayments by any people reading it, and if anyone wanted to excerpt a section or an image and transclude it in their page, they could do that and you'd get paid your share. Universally monetizing the web would have strangled the information revolution in the crib. It'd be like putting a coin slot on your computer terminal a-la Ubik. People would _avoid_ reading pages to avoid getting billed, they'd be easily swayed to centralised providers offering free access to their pages (because what they really want is power and influence) like Facebook Zero cynically did to India. There'd also need to be, for billing, a perfect record of absolutely everything that anyone ever read, and anyone ever published. It'd be an authoritarian's wet dream.
If you haven't read The Curse of Xanadu, [0], I urge that you do, and be willing to ditch ideas that don't work so people can benefit all the sooner from ideas that do, and other people don't beat you to the punch _thirty years_ after you started.
The idea is that you don't, you preserve existing links because you respect that people have already linked there. See Cool URIs don't change for example [0]
If you absolutely have to change the link, and if a million people link to it, and are still interested in whatever your link is, they'll update the links themselves.
If you removing the link for good, people will have to link to the Wayback Machine instead; Wikipedia bots do this automatically. [1]
Bidirectional links ultimately aren’t very interesting except to people who make CMSs. Normal readers just do not care “what links here”. People prefer to feel like they are moving forward, not backing up.
Website owners collectively spend a lot of money to get this information and the unidirectional nature of links on the web today helps to reinforce the moat that Google has in search.
I use the Emacs package org-roam which I guess is a free clone of roam. I've never used roam, but org-roam is fantastic. Org is, among other things, a kind of markdownish language. I have daily diaries and a whole knowledge base of notes I've built up. One feature I lean on heavily is org-babel, which is basically jupyter style notebooks in org. The backlink feature is super useful because it will take me to the dailies where I was referring to whatever note or notes. Those dailies will also include links to the particular code I've been working on. All in all it's great for someone with an aging brain that just doesn't have the recall it once did.
Roam is not just bidirectional links though. Like most successful innovation, its attraction is in putting together multiple (at least 3) simple but powerful ideas :
- Outlining: or the ability to represent a tree of (separate data) blocks on the simple classic text page
- In-line querying (+ block references/embed) : or the ability to visually reference/reuse distant blocks anywhere else within the (outline) tree.
- Bidirectional linking
So It is first about visualizing tree networks (=outlining) that you can freely recompose/reuse (=query/references), then (and only then), on top of it, building that other (virtual) hyperlinked network (=bidirectional linking)
In a (limited) way, it does : In (media)wiki pages are often hierarchical and this is indeed a tree network. you can also embed some content (templates...) and list pages (tables...).
I would say that the fundamental difference is that in (Media)wikis the pages are the nodes in the networks, while in outliners, it is the (sub) blocks that constitute them.
Also, in Mediawiki features like embedding and listing, like Bidirectional linking (as shown in this article), are almost an afterthought (limited, hidden…), while in Roam they became more integrated and advanced while being simple to use (and on that more granular block level).
You can transclude any page into another page, not just templates. If you want to compose content in MediaWiki atomically and transclude them into articles or books, that's built in. https://www.mediawiki.org/wiki/Transclusion
That doesn't include the built-in Lua scripting that can create content modules with more power than transclusion alone.
> and list pages (tables...)
Dynamic Page List extensions let you query content by namespace, category (including category intersection), and search results. You can also template the output, for instance to generate an image gallery from a query that returns `File:` namespace results.
You can do what Roam does in MediaWiki, but only if you're comfortable with MediaWiki markup, which can be miserably clunky with a steep learning curve when you get into templating. Roam's value is focusing on and surfacing those features and creating a UI around them.
You can't do MediaWiki in Roam, and Roam would be a worse product if you could. If you use Roam and wish it could do X or Y, you can get there with MediaWiki.
Not only Mediawiki but also wikis specifically for managing personal information like Zim Wiki and Tiddlywiki have backlinks. It’d be interesting to see an article exploring why Roam and other recent entrants seem to have been more successful than these wikis. (Assuming they are. I don’t have hard user counts to compare.)
I was a user of TiddlyWiki for many years before I started working on Obsidian.
TiddlyWiki is an amazing piece of software, but it has too many points of friction. The first being that you have to run a local server, which eliminates 99% of users.
Up until recently, most wiki software was modeled on Wikipedia, designed for consumption as a website. As an app, Obsidian can have more freedom when it comes to UI, e.g. multiple panes, tabs, command prompt, etc. All of which can make the experience feel richer and snappier than a website.
That said, whenever someone non-techy asks me what Obsidian is, I always start by comparing it to Wikipedia. People understand the idea that Wikipedia is organized via relationships/links rather than chronologically/hierarchically.
> TiddlyWiki is an amazing piece of software, but it has too many points of friction. The first being that you have to run a local server, which eliminates 99% of users.
This was not always true. One reason TW became popular was because you could save your data without running a web server on Firefox. Then Firefox decided to pull that functionality, and TW suddenly became a lot less exciting. You could still download apps or use the Github saver, but those were far less appealing than the old approach with Firefox.
At least for me: Speed. Roam is _fast_. I couldn't imagine using a Wiki that has lots of page loads in all sorts of flows because even from localhost, that just takes too long and feels too disruptive, this is one case where a fast SPA really shines. Also, they got things like autocomplete pretty much right, which many Wikis don't have at all. I still find Roam a bit cumbersome (why can't I set my graph to case-insensitive titles?) but the Wikis I've tried have been a lot less streamlined, and as I need to take notes pretty much all the time, little delays and annoyances add up. I'm told org-mode in emacs would be a better experience still, but trying to learn emacs has been one of the most unpleasant things I've ever done with a computer, so I guess I'll stick with Roam or whatever comes around to take its place. It's the first notes app product that feels somewhat right-ish to me.
There is a free version with a limited feature set available for personal use. Unfortunately, some essential features are missing (like tags and thought/link types), but at least you are not limited in the number of thoughts (= nodes) you can create, so it may be good enough for some people.
I’ve been using TheBrain (previously called “PersonalBrain”) for more than a decade now and got a lot more value out of its graph-based navigation and spatial linking concept than I would have with any of the text/outliner-based tools that have been popping up recently (although I use some of them too, for different reasons).
I built a website called “reversible.org” approx. 20 years ago that basically listed on every page the pages that linked to it, as a sort of self-organizing directory. If you were a blog about frogs, you would link to reversible.org/blogs/frofs and there was a list of everyone who categorized themselves as such. Of course, it was spammed to hell and back. I was so dejected about the failure I went off and designed tags on delicious without hierarchy.
They’ve had a lot of employee turnover, with many leaving for Tana. Having used both, Tana is just more polished. I have really enjoyed Roam and am grateful for Conor’s contributions to the space, but he really needed to hire someone with experience managing an engineering team that ships. The innovation has been in the RoamJS extensions. A former developer on the team told me he left because the founder was trying to “reinvent engineering management from first principles”. It seems like that effort was at cross purposes with shipping speed. ¯\_(ツ)_/¯
Roam singlehandedly made me love note taking. Before, it was so tedious and you were presented a wall of text, but it made content digestible and fluidly organizable using back links. Now I use Logseq, but it had a big part in getting my knowledge base to where it is now.
At least in my circles Roam has been absolutely impossible to miss even if living under a rock. Before Roam, people around me used either Evernote or OneNote or some kind of text files in directories scheme, all of which apparently sucks for other people just as much as they do for me.
Most are now on Roam or Obsidian, and I hear that at least in some in scientific circles Roam has made an even bigger splash. I've tried Roam and Obsidian and a few copycats and stuck with Roam, and it's lightyears ahead of what I had before, but there seems to be no good way to interface with the graph using code and that's limiting things severely for me. Still looking for something better, but IMO it's absolutely fair to say that Roam has had a huge impact.
If your circles were into onenote or evernote they were pretty tight to begin with. I use Apple notes app for everything. Going beyond a default OS notes app is a big step. I keep all my subscriptions in there. Monthly cash flows. Any interesting links I find. Though the links I almost never go back to
Roam (and Obsidian, Evernote, OneNote, ...) are aimed at people who work with large-ish amounts of facts and knowledge and sources and need to organize that somehow, I guess most people in such jobs will use something like that. Roam and Obsidian make it (relatively) easy to build a sort-of queryable graph structure, which I guess is a pretty niche thing to want. Apple Notes is great too, I use it for travel to do lists and the like, it's great for that. I don't keep my private everyday notes in Roam, that would be a bit overkill. They don't really solve the same problem, more like adjacent problems.
Seems an unjustified over generalisation or you’re taking the ‘personal’ in personal information too literally. I use obsidian to manage notes for my job, which is research and writing.
I think it’s probably the opposite — the data is obviously there — you encounter vast amounts of new information every day, though most of it useless, but there clearly remains a lot of useful information; you’re just not capturing it.
MediaWiki hasn’t been sufficiently developed and renovated over time to blame the PKM community for not using it. The most recent one I set up was in 2020, and it made me sad to see how annoying and ugly it was out of the box compared to contemporary tools. Sad because it has enormous utility but it hasn't done the work to help users and developers want to access it.
I’ve been following MediaWiki development since about 2005 and have contributed to it, so I’m part of the problem.
I recently set it up. What I disliked was that it was hard to get pre-built extensions (there was some kind of dynamically generated link to get an extension). I wanted to build a docker image that has my mediawiki build but it ended up being too tedious and the extensions which needed db schema changes scared me a bit in this endeavor. I wish it was just a single binary, a config file and a directory to put in extensions.
I like the new mediawiki ui though, a lot of people don't seem to but i think it's nice. I just wish it was responsive and mobile friendly. It's quite crazy that I need an extension to make the platform display correctly on mobile devices.
Sure, the sidebar doesn’t make it very prominent, but it doesn’t sound like something that should be. Even now knowing about it, I can’t see myself ever using it.
Nelson was 30 years late delivering Project Xanadu when Berners-Lee jettisoned the _impossible_ parts of Ted's vision and ate his lunch with the WorldWideWeb. [0]
Ted dreamed of these bidirectional links. Not just that you got _notified_ if someone linked to you, for which we have pingbacks (which are massively exploitable for spam if you auto-publish them), but if someone changed their document's address or title, they'd have the power to update _your_ document with the link to it. No bitrot! All it takes is one entity having complete executive control over the entire WWW, and probably the ability to change anything on any page without your knowledge, or even better, give remote control of your documents to anyone who controls the server you happen to link to.
He had other ideas, like you'd write a page and get automatically paid micropayments by any people reading it, and if anyone wanted to excerpt a section or an image and transclude it in their page, they could do that and you'd get paid your share. Universally monetizing the web would have strangled the information revolution in the crib. It'd be like putting a coin slot on your computer terminal a-la Ubik. People would _avoid_ reading pages to avoid getting billed, they'd be easily swayed to centralised providers offering free access to their pages (because what they really want is power and influence) like Facebook Zero cynically did to India. There'd also need to be, for billing, a perfect record of absolutely everything that anyone ever read, and anyone ever published. It'd be an authoritarian's wet dream.
If you haven't read The Curse of Xanadu, [0], I urge that you do, and be willing to ditch ideas that don't work so people can benefit all the sooner from ideas that do, and other people don't beat you to the punch _thirty years_ after you started.
[0] https://www.wired.com/1995/06/xanadu/