@henrikenggaard said:
I'm arguing from my own experience -- but that is semantics
Ah, got you.
@henrikenggaard said:
But I will keep an eye on my use over the next few days, and if I find something worth sharing, I will write a post (probably also about graphs/networks since I use those a lot, too.)
I strongly disagree with the notion that backlinks are not useful and worse. And I really don't understand how Sascha of all persons would come up with such a view. Yes, I've read his explanation, but to me it seems, hm, "odd". Sorry.
The whole scientific world is interested in backlinks: they are called citations. Because when you look a paper A and know it has been cited 10 times vs paper B with 1000 citations, then you know something about paper A's popularity/reach at least, maybe even something about its quality.
Backlinks tell you, what the fan-in of a zettel is. And a zettel with a high fan-in to me is noteworthy. A zettel with a fan-in of 0 also might be noteworthy - for different reasons.
Fan-in and fan-out of a zettel tell you something about its role in the network of zettels.
And about the context: Sure, if a link from zettel P to zettel C is just sitting on a line with "see also" it's maybe not that useful when you view zettel C and see a backlink to P. Or maybe it is. Who knows what it tells the innocent viewer of zettel C, that of all zettels it's P that's linking to it? Well, that of course requires zettel P to have a meaningful title and not just a timestamp. But that to me that is how zettels are naturally named.
Now, add to that if in P the link to C is not just tagged with "see also", but embedded in a whole sentence. And now imagine, a tool would show - when looking at C - not only that P is linking to it, but also "in what context" the link is embedded.
When looking at C I get a list of other notes linking to it (backlinks). I get a count of such backlinks (fan-in). For each backlink (use of the link to C) I also get the context in which it sits. To me that's tremendously useful to get this "out of the box" from a tool like Obsidian.
It's free, it's without effort - and if I don't care, I switch the backlink view off. Or if I want to add context or effort, then I can explicitly go to C and enter a forwardlink to P. But why would I want to do that?
Sure, in Luhmanns Zettelkasten backlinks were very, very expensive to list/count. But with a software based Zettelkasten they are cheap. To me there is no reason why a modern Zettelkasten software should not provide them - at least for those who find them useful.
Imagine going to a libary, browsing the shelfs, looking through a book, you find it interesting, you check out the literature references to find more interesting books.
Ordinary.
Now imagine, the book also had automatic backlinking switched on. You'd see all the books referencing this book. Not ordinary anymore!
Don't you think there could be a treasure hidden in backlinks like in the forward references?
@henrikenggaard Looking forward to read your response posts! Sharing our experience and dissecting what we actually do know and what is worth exploring further is, I think, a (the?) path forward.
I btw don't have much to add in terms of dissectable experience, only a couple of anecdotes and self-observation. I navigated in my archive through the search for a very, very long time, and neglected forming proper networks of departments. When I browse through these orphaned notes, which share a very loose connection (see "Kinds of Ties") for most of the time, I am forced to make sense of the search results list on the fly. And that's so much worse than having better connected notes for similar topics because apart from (often far from perfect) note titles, I have no hints! So when I search for some detail in Ruby programming, I have to sort through a mess, while I can look at my more recent notes about text editor programming on the Mac and dive into the sub-department of my Zettelkasten through different entry points, though trails like How to make scrolling not lag that starts with a list of scroll performance inhibitors or similar concrete questions I could ask Google to answer for me. The list of un-annotated, incoming links to a note, aka backlinks, that you get when you just search for a note's ID, that's more like the 100 assorted Ruby tips I have collected in the early 2010s: it's a curiosity to look at, but it's work to deal with it, and there are other mechanisms that
Most here seem to be arguing pro optionality, while Sascha is pointing out the opportunity cost. E.g. leaving options open could be dangerous when they enforce bad habits because you allow yourself to be lax at times.
Like when you want to maintain your weight while working from home in 2020, but there's a bowl of your favorite candy on the table that wasn't a problem for your circumference until you were forced to sit next to it all day, every day, because of quarantine/lockdown. Slowly you pick up the habit of reaching for candy instead of noticing you are getting hungry and should prepare a meal rather sooner than later. Sure it's could be nice to have the option to treat guests, but having a bowl of candy on your table could destroy the rest of your eating habits, badly influence your blood sugar, concentration, and ruin your Keto diet, or whatever.
It's not like anyone is saying "Backlinks do not exist!". That'd be ridiculous. But is the warm and fuzzy feeling we get from browsing around with backlinks a good proxy for the utility of relying on them in our work? Do they actually make us better as much as we think they are? (And does it matter, or is the warm and fuzzy feeling all we're after? In that case there's no point in arguing, enjoy yourselves and that's it.)
Where's the addiction lurking? I don't get it. I don't need to do anything for backlinks; they just get created. I don't need to look at them. But when I see them, they might help to assess the context of a zettel. I'm probably too ignorant to understand what harm that might be able to cause.
But is the warm and fuzzy feeling we get from browsing around with backlinks a good proxy for the utility of relying on them in our work? Do they actually make us better as much as we think they are?
If "browsing around" gives you just a warm feeling and is not useful, does not help you achieve a goal, then backlinks won't make it worse.
If studying your network of zettels by following links helps you achieve a goal, then why would backlinks be links of lesser value?
I'd even say, there are no backlinks. There are just links, connections, relationships. Some are created explicitly, some automatically, some implicitly. Why limit connections to unidirectionality?
That's why some tools are offering to show not only backlinks, i.e. manually placed forward links viewed from the referred to zettel, but also the use of words/phrases which match zettel titles. Such a mention would be to use "liquidity" on some zettel w/o a link. And it still could be listed as a potential backlink on the Liquidity zettel.
What's not to like about that? How cool a feature so I can actually discover - instead of plan - relationships between zettels/topics!
Lack of a feature in a tool should not lead to making it a virtue not to want such feature. Backlinks are a technical possibility. They are here to stay. So I'd find it more helpful to explore how to use them best, instead of emphasising some harmfulness.
I don't think backlinks are like knifes. But even knifes are in widespread use, despite their potential to be used as a weapon. So, better make people conscious, responsible users of backlinks, than keeping them away from the feature.
I'd even say, there are no backlinks. There are just links, connections, relationships. Some are created explicitly, some automatically, some implicitly. Why limit connections to unidirectionality?
This discussion may be suffering from an apples to oranges comparison. I've been using Roam for a while, motivated by the ease and flexibility with which information can be stored and cross-linked. I thought these capabilities would be great for a Zettelkasten.
But, after several weeks it became clear to me that Roam was not well-suited for developing a Zettelkasten in the classic Luhmann sense -- as it is automated in The Archive.
It is possible to use Roam in that manner, but you have to cripple its functionality to do so. Or, you have to extend your definition of zettelkasten (lower case) to include any knowledge storage system incorporating links -- and that might not be useful.
Roam (and I believe Obsidian) works in fundamentally different ways from The Archive. For example, a link in The Archive is (a) created ON a note and is (2) separate from the information documented in the note. It might look like this:
[[202011200545]]
or maybe
[[202011200545 Title of this note]]
A link in Roam is typically entered (a) on a Daily Page in an outline style where it is (b) an organic part of whatever information is being entered. The link is not an add-on to the "note" - it is the "note".
A link in Roam might look like this:
[[Paper A]] was cited by [[Paper B]] because of the relevance of its Figure 6 found on page 14. #something, #something else, # third thing.
In Roam, the first-level, bi-directional link is typically from the Daily Page where information is recorded to a page (or pages) where that information can also be found--a much different structure from that used by The Archive.
The complexity that arises when trying to characterize the nature of links in The Archive does not arise in Roam because you can characterize links using plain language as part of the process of creating the Roam equivalent of a note.
Characterizing the nature of a link is not a limitation of language or the ability of the person creating the link to express his or her intention. It is a limitation of The Archive's method of creating a link. This limitation and its influence on the desirability of backlinks may be relevant to The Archive but not to Roam.
In the example above there are actually five links because this "block" (each bullet in an outline is called a "block") can now be found on the pages for the two papers, as well as the pages for the three tags. Multiple-link constructions like this are optional -- I find them convenient, empowering -- others may find them confusing. There is no requirement to use them.
Because Roam makes use of an outline metaphor for data entry, a link may be even more complex. The link example described above could contain child-blocks and grandchild blocks. Those "descendants" would be part of the link. if you go to the Page devoted to Paper A, you would find the top-level block date-stamped and displayed in a collapsed state -- with all child, grand-child, etc. sub-bullets immediately accessible by expanding the highest level block.
Is this a good system? Maybe no if you think that a properly constructed zettelkasten is really The Zettelkasten and that The Zettelkasten requires that specific rules be followed. Maybe yes if you think that a zettelkasten is one of several possible knowledge management systems, perhaps inspired by The Zettelkasten, but now offering alternative approaches.
Further complicating these discussions are other fundamental differences in approach between systems like Obsidian and Roam and The Archive -- the nature of links being only one, and IMO, not the most important.
In Obsidian you create zettels as files. It's explicitly based on (local) .md files. How you name those files is up to you. There is a plugin for naming them with timestamps. Or you use timestamps + a topic, or a topic only. Up to you.
In what way is that different from The Archive?
Yes, the display is different. The Archive is always just showing a single long list of zettels, if I remember correctly. If you want that, well, Obsidian also provides a long list of all the files/zettels you created. But who wants to look at a 10K list of zettels? So the essential functionality is search, I guess. And of course search delivers a long list of zettels in The Archive as well as Obsidian.
Yes, the "entry point" is different. The Archive has only one: the list. Obsidian provides you with two: the list or the daily entry. And I have to say, I have come to like the daily entry very much. I was skeptic at the beginning, but now I find it a differentiator compares to The Archive or Notion.
When I write a new zettel I do it in the daily entry by entering its link, e.g. [[My new zettel]]. Then I open the new zettel and write down whatever I want to take note of. For the next new zettel I do the same and so on.
The result is no different from The Archive: zettels accumulate in a long list, the folder where I want to store them.
But I get more than that: I also get a simple daily context. I can go to any day and see which zettels I created back then.
Also I can (and do) use the daily entry as a small diary of sorts. I can add stuff that does not need to go into a zettel. That's not essential of course for a zettelkasten. But nevertheless I find it an opportunity I'm using quite often. And of course when I search for something there is no difference between an official zettel and a daily entry.
Of course, if I'm looking at a zettel I found through search and want to create a related new zettel, I can go right ahead, enter its link, and move to the new zettel. Thereby the web is woven en passant.
Roam goes beyond that in that it does not use physical files to store data. It makes each and every paragraph a potential node in a network. Obsidian is somewhat moving in the same direction. I can link to headlines inside of files. But those links are kind of brittle, because in Obsidian they are relying on the actual title, not an underlying paragraph ID as in Roam. So one might be careful to use them.
The main links in Obsidian and The Archive, though, are file names. And Obsidian does a good job of updating links should I choose to rename a note or move it inside an Obsidian vault.
If you look at the resulting files there is no difference between what The Archive produces and Obsidian, I'd say. It's a long list, each named however you like it.
And if you choose so you can add folders to organize those files. Or tags. If you don't like that, then just don't. I just would not want to tell people that's "non-canonical".
If zettels are "information nuggets" (I want to avoid the term "note") stored in individual files for longevity, then The Archive and Obsidian are both tools to develop a zettelkasten. And I very much like the competition that has sprung up in the past 12 months in the zettelkasten tool arena. There is a lot to explore.
Is this a good system? Maybe no if you think that a properly constructed zettelkasten is really The Zettelkasten and that The Zettelkasten requires that specific rules be followed. Maybe yes if you think that a zettelkasten is one of several possible knowledge management systems, perhaps inspired by The Zettelkasten, but now offering alternative approaches.
This I'd change. The criteria are not based on ones mental model of a ZK but if the system actually makes the person more productive.
EDIT: I think @ralfw is correct. The Archive and Obsidian are more similar than Obsidian and Roam.
@ralfw - Obviously, my understanding of Obsidian is weak. Sorry for generating the confusion by assuming it aligned more closely with Roam. When I alluded to additional differences between Roam and The Archive I had the Daily Note in mind - an approach also used in Obsidian. I rely on it heavily and it is very different from making entries in The Archive.
As @Sascha correctly notes, the measure of value is productivity. Direct comparisons on a feature by feature basis (e.g., the value of backlinks) is hard because the overall structure of the applications is different. What works or is easy in one may well not work or be difficult in another.
@bjbarry: As I said before, I find the daily note a very useful addition to the general list of zettels. It provides another view - which one can use, but does not need to use. I find it easier not to use in Obsidian than in Roam.
Zettels benefit from context. To create context or to "encode it" in a zettel, is hard. To anchor new zettels in the daily view is a form of very natural context to me. And if the distinction between zettel and note gets blurred by that... then I'm quite happy. Zettels get stored in a dedicated Zettelkasten folder. But I can link to them not only from daily notes but also from any other note I care to store in Obsidian. And I can link to them through Obsidian links from all other applications as well, e.g. a calendar entry or a task or a Word manuscript for a magazine article I'm writing.
To me features inform the method, and the method informs tools to implement features. It's a co-evolution. Practice + theory, not one over the other or just one of them.
The criteria are not based on ones mental model of a ZK but if the
system actually makes the person more productive.
I agree with this. If one system encourages you interacting with the
system itself with positive feedback loop. It is the system for you.
As long as the user constantly reviews his/her methodology, it should
not be an critical issue.
The criteria are not based on ones mental model of a ZK but if the
system actually makes the person more productive.
I agree with this. If one system encourages you interacting with the
system itself with positive feedback loop. It is the system for you.
As long as the user constantly reviews his/her methodology, it should
not be an critical issue.
Constantly? Why? To what end? Wouldn’t constantly using your methodology prove its worth....or Not? And if not, it should become self evident. I guess I’m getting hung up on my perceived notion of wasting time reviewing rather than using.
The criteria are not based on ones mental model of a ZK but if the
system actually makes the person more productive.
I agree with this. If one system encourages you interacting with the
system itself with positive feedback loop. It is the system for you.
As long as the user constantly reviews his/her methodology, it should
not be an critical issue.
Constantly? Why? To what end? Wouldn’t constantly using your methodology prove its worth....or Not? And if not, it should become self evident. I guess I’m getting hung up on my perceived notion of wasting time reviewing rather than using.
Reviewing the Zettelkästen is fun for me. I set a goal to achieve
everyday, and I feel that I have more time concentrating on writing
atomic notes, which not necessary being productive on my work, but I
feel much better life satisfaction. I used to do daily reviews, weekly
reviews, and monthly reviews, but ending just weekly reviews and
monthly reviews now. During these reflection time, I will try my best
to decide the current methodology works or not.
To constantly using a methodology is the goal. It can be qualified by
how much time I spend in my Zettelkästen. I did this on daily basis
with time tracking tools. It strongly correlated to the goal that I
set, i.e. how many ideas I write every day.
I am not saying it is bad idea, but I feel that not having a single
determined goal will dilute my motivation. I feel just catching how
many notes I write is a better solution.
For me reviewing the Zettelkästen content is one of the way I use my
Zettelkästen. Being able to interacting my own thoughts is a flow
process for me.
The criteria are not based on ones mental model of a ZK but if the
system actually makes the person more productive.
I agree with this. If one system encourages you interacting with the
system itself with positive feedback loop. It is the system for you.
As long as the user constantly reviews his/her methodology, it should
not be an critical issue.
Constantly? Why? To what end? Wouldn’t constantly using your methodology prove its worth....or Not? And if not, it should become self evident. I guess I’m getting hung up on my perceived notion of wasting time reviewing rather than using.
I agree with @learning_ran in spirit, but perhaps not how to do it.
There are different levels of knowledge work:
The inner mechanisms and the nature of knowledge (philosophy, history and sociology of science, epistemiology, neuroscience etc.)
The methodology itself (How to write a Structure Note, how to place a link etc.)
The software
I base my refinments of the method on the inner mechanisms and the nature of knowledge.
Example (automated translation of an original ZK note I am currently working on, copyrighted etc):
Iain McGilchrists sees a difference in the brain hemispheres in that the left brain hemisphere focuses more on individual parts of experience and perception, while the right brain hemisphere focuses on the totality of perception.1.
The right half of the brain seems to focus more on the shape of an experience, while the left half seems to focus more on sequences, sequences and similar separating concepts.2
Patients with lesions in the right hemisphere of the brain lose sight of the big picture34 In particular, they lose the ability to form a visual image of an object, for example, if they have only touched it. Not because they have problems transferring haptic to visual. Rather, they lack an eye for the essential, the whole, from which the transfer would be possible.1
This problem rarely occurs in patients with lesions of the right hemisphere of the brain. However, problems are more likely to be manipulated (which is more in the domain of the left hemisphere).5.
A productive method allows for harmonious engagement of both hemispheres. Sadly, it is difficult to actually develop processes that prevent such problems. However, at least -- if understood -- you can use those insights as a filter against features and techniques that throw of the balance of the hemispheres. In Germany this is described as gehirngerecht (popularised by Vera Birkenbihl) which roughly translates to brain-friendly.
The nature of both hemispheres allows to assess one of the main obstacles in the collective thinking process. The left hemisphere tends to grab details and forces them together in the hope to make a whole. However, because it is not able to do so, the result is something akin to the Frankenstein Monster: This happens if there is no bigger picture behind the development of the software (or the method). This is what we observe as feature bloated software or "nice to have the option"-mentality (no dig!). Or on the individual level: As the inability to give life to something that is organically growing and nurturing it, and instead use systems, methods or whatever until use becomes misuse and results in abondenment and finally death (jumping from system to system or regularly starting all over again, resetting etc.).
The whole topic of knowledge work is a pyramid standing on a fundament of the inner mechanisms and the nature of knowledge, with a thick body of methodology and a small tip of software.
Iain McGilchrist (2009): The Master and his Emissary. The Divided Brain and the Making of the Western World, Totton: Yale University Press. ↩︎↩︎
J. L. Bradshaw and N. C. Nettleton (1981): The nature of hemispheric specialization in man, Behavioral and Brain Sciences 1, 1981, Vol. 4, S. 51--63. ↩︎
J. McFIE, M. F. PIERCY, and O. L. ZANGWILL (1950): VISUAL-SPATIAL AGNOSIA ASSOCIATED WITH LESIONS OF THE RIGHT CEREBRAL HEMISPHERE, Brain 2, 1950, Vol. 73, pp. 167-190. ↩︎
GEORGE ETTLINGER, ELIZABETH WARRINGTON, and O. L. ZANGWILL (1957): A FURTHER STUDY OF VISUAL-SPATIAL AGNOSIA, Brain 3, 1957, Vol. 80, pp. 335-361. ↩︎
J. MCFIE and O. L. ZANGWILL (1960): VISUAL-CONSTRUCTIVE DISABILITIES ASSOCIATED WITH LESIONS OF THE LEFT CEREBRAL HEMISPHERE, Brain 2, 1960, Vol. 83, pp. 243-260. ↩︎
I have to agree to disagree, Sascha. Your argument sounds plausible, but if you take a step back, you realise that any complex knowledge matrix is extremely difficult to navigate. Backlinks are simply one method of enabling knowledge managers to navigate the matrix. Indeed, they can serve as a valuable first step in constructing a more organised “knowledge-building” structure by reminding you of relationships you may have forgotten.
Consider: after using a given backlink, you can always incorporate it into a more explicit, more definitive structure (if it seems appropriate to do so). The best of the backlink-capable apps (e.g. Obsidian, NotePlan beta 3) I’ve used also provide the user with some context for the backlink, which undermines your distraction argument: I would argue that establishing context is a vital part of building knowledge.
Building any kind of knowledge base is extremely difficult, in that there is always considerable mental overhead involved. I have enormous respect for the Zettelkasten approach, and agree with your criticism of the innately linear/hierarchical Folgezettel sub-approach. But there’s nothing linear or hierarchical about backlinks, and they may highlight relationships that enable you to broaden your lateral-thinking ability (another vital part, I would suggest, of knowledge building). That backlinks (especially automatic backlinks) can become intrusive and annoying I wouldn’t deny – except that in practice, and based on thousands of words of notes, this rarely happens. Instead, my experience suggests that many backlinks eventually turn into intentional links, and may in fact result in more intentional links than one might otherwise have remembered to create.
As for your parallel with the Internet: well, the latter’s profusion of hyperlinks only becomes distractingly confusing if you use it in an undisciplined way. If you take a disciplined approach to hyperlinks, you can use them to build enormous webs of knowledge very rapidly (to take a very simple example: by opening links in separate tabs/windows rather than just clicking through). I do this all the time when conducting rapid web research in my capacity as a translator/copywriter, just as I use multiple search engines to find page sources in the first place. Much of this material will then make its way into dedicated knowledge bases – which I may choose to use individually or in combination, where knowledge in one particular field can help to clarify questions in another one. The same applies, for example, to legal cases or documentation, or various aspects of engineering. Discipline is also something that needs to become automatic ;-)
@MadaboutDana said:
I have to agree to disagree, Sascha. Your argument sounds plausible, but if you take a step back, you realise that any complex knowledge matrix is extremely difficult to navigate.
I cannot confirm this empirically. My Zettelkasten is very easy to navigate, it got easier the more complex it got and hit a ceiling since a certain point.
Backlinks are simply one method of enabling knowledge managers to navigate the matrix.
I will be more clear in a more general text on backlinks in the context of knowledge work: I am not arguing against backlinks as a navigation tool. I argue specificely against the the surplus of knowledge through connection that links (can) provide (if placed properly).
Consider: after using a given backlink, you can always incorporate it into a more explicit, more definitive structure (if it seems appropriate to do so). The best of the backlink-capable apps (e.g. Obsidian, NotePlan beta 3) I’ve used also provide the user with some context for the backlink, which undermines your distraction argument: I would argue that establishing context is a vital part of building knowledge.
Am I correct that you deem the note preview to be a link context? If yes, then:
I do not consider a note preview proper link context as a note preview does not highlight the nature of the connection. Proper link context gives you past and present reasons to follow the link. That means that one need to have written about the link nature in the first place. Otherwise, you will have to do work that already should have been done: Investigate the nature of the link.
Building any kind of knowledge base is extremely difficult, in that there is always considerable mental overhead involved. I have enormous respect for the Zettelkasten approach, and agree with your criticism of the innately linear/hierarchical Folgezettel sub-approach. But there’s nothing linear or hierarchical about backlinks, and they may highlight relationships that enable you to broaden your lateral-thinking ability (another vital part, I would suggest, of knowledge building). That backlinks (especially automatic backlinks) can become intrusive and annoying I wouldn’t deny – except that in practice, and based on thousands of words of notes, this rarely happens. Instead, my experience suggests that many backlinks eventually turn into intentional links, and may in fact result in more intentional links than one might otherwise have remembered to create.
Can't argue with experience. That would fall under the "not very useful" category. There is not no use. But the subpar quality of backlinks compared to other actions impose opportunity costs that decrease productivity (per time).
In the light of the following quote of yours: One part of the benefits of a ZK is that it allows you to focus on one thought at a time (technically, a link is a thought about the connection between two thoughts). The ZKM is not only about the setup and the thing "Zettelkasten". There are habits and processes baked in that lifts mental load like having to keep several items in your working memory more than necessary.
I need to formulate it more carefully to give it its due but to hint at it: There is a equivelant to the "distraction free" in "distraction free writers" that is part of the ZKM as I formulate it. From my experience few people have the working memory to hold much items necessary for some thinking processes. The aid a ZK gives you is partially by functioning as an extended working memory. There is a very good article by Anders Ericson called "Long-Term Working Memory" which gives a model on how the brain extends the working memory, especially in experts. To give rise to this effect the ZK-Environment (software or paper) needs to reduce the cognitive burden. One of these burdens are backlinks.
The issue is that you can't just switch them on and off how you like and expect your brain to work as if they are there or not. If you are aware of this function your brain will model your perception as if it was there. That is the reason why even switching of your smartphone is not enough to lift the negative effect of smartphones on your fluid intelligence and working memory. Only putting it in another room is enough, if one might assume that you get back to baseline (which I doubt). Another example of this problem can be felt. For a distraction free writer you even need to remove features from the software to reap all the benefits.
As for your parallel with the Internet: well, the latter’s profusion of hyperlinks only becomes distractingly confusing if you use it in an undisciplined way. If you take a disciplined approach to hyperlinks, you can use them to build enormous webs of knowledge very rapidly (to take a very simple example: by opening links in separate tabs/windows rather than just clicking through). I do this all the time when conducting rapid web research in my capacity as a translator/copywriter, just as I use multiple search engines to find page sources in the first place. Much of this material will then make its way into dedicated knowledge bases – which I may choose to use individually or in combination, where knowledge in one particular field can help to clarify questions in another one. The same applies, for example, to legal cases or documentation, or various aspects of engineering. Discipline is also something that needs to become automatic ;-)
This is an excellent discussion. I see this as 'Different strokes for different folks". And to me, a big mistake many productivity software developers and users make, is to assume the rest of the world thinks and works like they do.
I think one factor here is the length of time you'll need to use your links. When you use a link to jump from one item to another, you might use that link for the rest of your life or professional career (if for example, you teach or do academic research) or, you may only use that link for the next minute or so to snatch just one morsel of a thought, which probably is the case for most of the general population. Suppose on my computer or mobile device I need someone's telephone number quickly, or the location of today's meeting real fast. That's a link, and the backlink to return to my original thought or note may be nothing more than the back-arrow on a standard web browser. In my life and my personal organizer software, that may compose 90% or more of the links I use.
So the vast majority of links many people use may be needed for only a minute or so, to solve quick temporal questions on daily routine tasks and issues. For people like that, backlinks just aren't necessary, and as the original author noted, can be wasteful of time and productivity.
So frankly, I see backlinks as they are promoted in some productivity software today as just an overblown, unnecessary feature, probably useful for only a small subset of the general population.
I would agree with this article. Whilst I appreciate that backlinks may help others navigate your zettelkasten, it doesn't help me build mine. In the end my zettelkasten is for my use and as @Sasha importantly said, for my growth in knowledge. A computer cannot link knowledge, it can only link words and terms. I believe it is most dangerous to allow anything to create links in your zettelkasten other than yourself.
I'm sure if I looked at someone else's zettelkasten, I could easily see links that I think should be created. But those links do not necessarily demonstrate knowledge to the owner of the zettelkasten. In the same way I would not allow another person to create links in my zettelkasten, nor would I allow a computer to do so. It is my second brain, not a playground for an algorithm. Links created by anything/anyone else in your zettelkasten do not demonstrate a growth in your knowledge or a connection of ideas. They can only point to potential connections and if your zettelkasten is full of potential connections created by software, it ceases to be a second brain of what you actually know otherwise I could claim that the Internet is my zettelkasten.
I continue to be puzzled by the negative reaction that people have to back links. That may be because I am using Roam where two-way links, together with the Roam outlining metaphor, block-level links and use of a Daily Note Page are thought to create the strength of the application.
I don't think it is because back links are unsuitable for use with a zettelkasten. Roam instances can be "multi-player." There are currently 2,500 Roam users participating in a six-week "book club" using a Roam instance devoted to "How to Take Smart Notes."
This is the second such book club devoted to Ahrens book, the first had over 1,000 participants. Ahrens participated in the first and recently complimented the enthusiasm with which the second was being conducted.
So, either these thousands of people are delusional or it is perfectly possible to create a viable zettelkasten using software that employs back links.
If these back links are, in fact, structural weaknesses, then that is a real problem for Roam because interest in the Roam community for zettelkasten methods is intense.
But, leaving particular software features aside, I think that back links are important -- at least to me -- in other aspects of life.
For example,
An airline would not last long if it offered only one-way tickets.
Athletes practice and actors rehearse - "going back" and repeating after mistakes are made or when better options are discovered.
A chess player analyzes a position by considering different moves, a continuous process of "move here and go back" "move there and go back". That thinking might all occur in the player's mind, but it occurs.
Map applications provide directions from Point A to Point B, but also from Point B to Point A.
I use the ⌘-Z key combination with some frequency.
I would not use a web browser without a back button.
What happens in "real life" when a back link is not available? You might hear:
"I wish I hadn't said that to my boss."
"Why did I marry that guy?"
"I wish I had invested in Apple in 2006."
"Groan, why did I drink that last pitcher of beer?"
"Did my golf ball end up in the lake?"
"Now that I finally understand women I wish I were 18 again."
Back links in Roam are created automatically, are unobtrusive and useful, but are strictly optional. There is no requirement to use them, but most people do, and do so frequently.
@bjbarry said:
So, either these thousands of people are delusional or it is perfectly possible to create a viable zettelkasten using software that employs back links.
This is not an either/or. It's not reasonable to assume they're all attempting to build "pure" Zettelkästen, or trying to build Zettelkästen at all. It's also possible that they don't depend much on backlinking. It's most likely that they are not yet hitting the problems of scale that backlinking can generate when representing knowledge and argumentation.
I don't think that the argument was ever "a ZK built in a system with automatic backlinking cannot be viable". As I read it, there are two main prongs:
Backlinking generates information without representing knowledge, and that information becomes overwhelming at scale.
Therefore, depending on backlinking is a trap, because in a year it will be less useful, and in 5 will be nearly useless.
But, leaving particular software features aside, I think that back links are important -- at least to me -- in other aspects of life.
For example,
An airline would not last long if it offered only one-way tickets.
[...]
I don't think this is a very effective set of arguments because they don't model backlinking. An analog to backlinking would be:
An airline's website would be pretty awful if, when I indicated interest in flights departing LAX, it showed me a list of all flights arriving at LAX, at all times on all dates.
… which is exactly what Sasha was arguing happens to backlinks, at scale, in PKMs.
The difference between that and backlinks is that, to your point, backlinks are useful in some contexts. Sasha's point is that they are an "attractive nuisance" for someone building a lifelong body of thousands of notes. Structuring knowledge -- and knowledge in particular! -- with them seems useful until you hit real scale.
I'm reading all the discussion a second time after some months.
I use Obsidian, it has an automatic backlinking feature, but I don't use it.
It doesn't work for my mind.
I acknowlegde with some arguments pro backlinking emerged in the discussion, but in my experience I've found the same problems that Sasha has explained (before reading him :-) ) and are very heavy in my case and superior of pros.
1. backlinking doesn't provide context linking. I feel the exact problem of Sascha, I need to have a why, thinked before and represented after, for the links I put.
2. a list of backlinking links is flat. They are all equals. If there are many, it's a mess.
3. backlinking, above all automatic backlinking causes an overlinking effect in the notes, effect that make linking less effective. Linking is useful but I need to keep managed.
For me linking is an effort, a friction in note-taking, but is a good friction. Manual linking is thinking. Automatic linking is something like collecting.
After I wrote this, I think the discussion arises because the same simple mechanic of linking means and implements very different things. Enumerating some of them, links are usable for:
1. make relationships between things: concepts, ideas, and so on
2. make a browsing aid (the perfect example for me is a wikipedia-like linking implementation, with which I can easily move between pages)
3. make information architecture, or structure (the example here is using links to make structrure notes, outlines, or whatever)
4. and so on
Link purpose are many, and is different the importance and relevance of each use, too.
The problems or backlinking are more critical for some of these uses, less for others.
In the context of creating notes in Zettelkasten, the most important links are of type relationships, and in this case using backlink instead of thinked linking is a bad use, according to the principle already expressed "linking is thinking".
Despite being aware of this, anyway, I don't have in my notes only relationship links. I need relations between my notes, but also support for navigation. So, besides forming primary link between notes, I also make other layers of linking:
a wiki-like linking system, linking some key terms in the text to the note who defines them; for example, in my "Permanent note" note I've the first occurrence of Zettelkasten word made as a link to Zettelkasten note.
links that make structure and classification, such a "subject links" and "parent links". I use linking as tagging for subject classification. Psychology is not a tag, is a note and linking with Psychology replace tagging
structure notes
In some of these uses, having two links, forward and backward, between notes is useful.
I have four different ontologies of linking, I don't need the fifth provided by obsidian backlinks :-)
All the links I do are manual made. they are an effort, but this effort is thinking. For every link i do I have to think if, why, how and to which I make this link. A link placed must be "my link", not a link of a software.
The big problems of tools like obsidian, for me, is that don't differentiate between different kinds of links. So, I need to use some tricks that visually render different links. Using different positions, adding a prefix or suffix or texts, adding icons or callouts, even making an appropriate use of spaces.
In the discussion emerged semantic linking, is a relevant theme for me.
I find its automatic backlining of Obsidian useful for things like
" Serendipitous navigation", something that :-) , but I very rarely use that thing
for a temporary support for making real manual links, choosing few of proposed backlinks. For me automatic backlink mainly is a "build tool" for relationships, not the everyday implementation of the relationships between notes.
In the next years the question of automatic backinking will resurface in a different form, with the advent of the AI.
Automatic classification, automatic linking. they will be more powerful and invasive than backlinking.
For what I've written in the previous post, for me this will be a risk.
The big problems of tools like obsidian, for me, is that don't differentiate between different kind of these links. So, I need to use some tricks that visually render different links. Using different positions, adding a prefix or suffix or texts, adding icons or callouts, even making an appropriate use of spaces.
In the discussion emerged semantic linking, is a relevant theme for me.
If you are an Obsidian user, the plugins Supercharged links and Breadcrumbs provide some standardized tricks for indicating link types. Such plugins can help to implement and automatically represent the kind of conceptual schema of note types and link types that I described in a recent comment in another discussion.
The big problems of tools like obsidian, for me, is that don't differentiate between different kind of these links. So, I need to use some tricks that visually render different links. Using different positions, adding a prefix or suffix or texts, adding icons or callouts, even making an appropriate use of spaces.
In the discussion emerged semantic linking, is a relevant theme for me.
If you are an Obsidian user, the plugins Supercharged links and Breadcrumbs provide some standardized tricks for indicating link types. Such plugins can help to implement and automatically represent the kind of conceptual schema of note types and link types that I described in a recent comment in another discussion.
Yes, I already use supercharged links a bit. It is a good help.
I like the idea of breadcrumbs, but it has an ugly interface :-)
I'm always doubtful about depending heavily on plugins.
I want to maintain my notes for years, external features could not survive in the future.
A good tradeoff at the moment is annotating the notes with textual tricks.
For example, I use something similar just described in the comments above, i call them "weak tags", they start with a '@' and placed near some relevant links. One of the tag is @bible, indicating a link to a source note that is the most important source that I've used for create the note.
Today my insight has suggested me, wanting to insert in my notes what I've written tonight in this topic about backlinking, making a note title "What I've written about Backlinking (my)" and placing into the note the link at this page with tag '@where' :-)
Conventional positions, artifacts like weak tags, conventions in how writing links help me to differentiate links, at least in their rendering into the note. They are not really semantic constructs, they don't have a special treatment in Obsidian.
I'm always doubtful about depending heavily on plugins.
I want to maintain my notes for years, external features could not survive in the future.
A good tradeoff at the moment is annotating the notes with textual tricks.
The software dependency problem is not limited to Obsidian plugins; it applies to Obsidian itself and most other apps. The solution is to use syntax that can be converted later if needed.
If you learn (or already know) some tools for data wrangling text using regex, you can use Breadcrumbs link-type syntax in your notes now and then convert the syntax to something else later if other software comes along that uses different syntax. In other words, why not use the syntax that current software supports and then convert the syntax later if that software goes away, instead of using custom syntax that no software supports? That's a question to consider if you haven't considered it. It's not a question of semantic/conceptual schema, because you can use the same schema with either syntax.
I'm always doubtful about depending heavily on plugins.
I want to maintain my notes for years, external features could not survive in the future.
A good tradeoff at the moment is annotating the notes with textual tricks.
The software dependency problem is not limited to Obsidian plugins; it applies to Obsidian itself and most other apps. The solution is to use syntax that can be converted later if needed.
If you learn (or already know) some tools for data wrangling text using regex, you can use Breadcrumbs link-type syntax in your notes now and then convert the syntax to something else later if other software comes along that uses different syntax. In other words, why not use the syntax that current software supports and then convert the syntax later if that software goes away, instead of using custom syntax that no software supports? That's a question to consider if you haven't considered it. It's not a question of semantic/conceptual schema, because you can use the same schema with either syntax.
I could apply this reason if a plug in like breadcrumbs were a perfect suit for my needs and were strategic in this . But it doesn't, it implements only a partial set of link typing, hierarchical. I would pontential use only the parent relationship, thing that already do without plug-in in the use I have of that, using metadata.
Either breadcrumbs and "mine" are not perfect, but mine is more tailored and hasn't an external dependendency to manage.
For other plug-ins, like templater, I make other considerations.
I could apply this reason if a plug in like breadcrumbs were a perfect suit for my needs and were strategic in this . But it doesn't, it implements only a partial set of link typing, hierarchical. I would pontential use only the parent relationship, thing that already do without plug-in in the use I have of that, using metadata.
Either breadcrumbs and "mine" are not perfect, but mine is more tailored and hasn't an external dependendency to manage.
For other plug-ins, like templater, I make other considerations.
The Breadcrumbs link-type syntax is supported by other plugins too, like Dataview and Juggl. There is nothing explicitly hierarchical about the syntax (it just adds a property to the link that the various plugins interpret based on user preferences), and there is nothing about the syntax that requires any external dependencies: it's just plain text!
Personally, I would not use your link-type ("weak tag") syntax because it is the same as Pandoc's citation syntax and the pandoc-crossref filter syntax for cross-references to figures, equations, and tables. I make sure that all of my Markdown syntax is as compatible as possible with Pandoc's Markdown, which is a crucial tool for the kind of conversion I talked about in my previous comment. One could convert your custom syntax before running Pandoc, but that seems like an unnecessary step to me. If you're not familiar with Pandoc, I highly recommend learning about it.
Since I use what might be called backlinks, I don't have a regulation Zettelkasten. I have a Xettelkasten. My Xettelkasten, hereinafter XK, has alphabetic index notes. The note on Xettelkasten has the link [[0000.0000.0VWX]] V-W-X to the V-W-X alphabetic index note. When I visit the V-W-X note in Zettlr, the "related files" pane shows me the notes that link to it. This technique gives me an alphabetic index for free (except for adding the index note link)1. But the presence of such backlinks, intended to be traversed in reverse like the "what links here" link in Wikipedia, means that I cannot claim to have a ZK. I have an XK.
1. Not to mention the "emotional labor" I incurred by writing this comment, as if anyone asked.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
Since I use what might be called backlinks, I don't have a regulation Zettelkasten. I have a Xettelkasten. My Xettelkasten, hereinafter XK, has alphabetic index notes. The note on Xettelkasten has the link [[0000.0000.0VWX]] V-W-X to the V-W-X alphabetic index note. When I visit the V-W-X note in Zettlr, the "related files" pane shows me the notes that link to it. This technique gives me an alphabetic index for free (except for adding the index note link)1. But the presence of such backlinks, intended to be traversed in reverse like the "what links here" link in Wikipedia, means that I cannot claim to have a ZK. I have an XK.
"An alphabetic index for free": that's a great point, XettelDistraction. Also note that automatic backlinks give you a subject index for free if you have a note type analogous to subject headings when you link to notes of that type (a.k.a. "links as (subject) tags"). Admittedly, one can usually also get such alphabetic and subject indexes for free by using (non-link) tags. Perhaps those who condemn automatic backlinks opine that the people should use tags for such free indexes, not links as tags?
1. Not to mention the "emotional labor" I incurred by writing this comment, as if anyone asked.
@Andy said:
Also note that automatic backlinks give you a subject index for free if you have a note type analogous to subject headings when you link to notes of that type (a.k.a. "links as (subject) tags").
I have such features as well in my Xettelkasten.
Signed,
XettelDistraction.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
Comments
Ah, got you.
Nice!
I am a Zettler
I strongly disagree with the notion that backlinks are not useful and worse. And I really don't understand how Sascha of all persons would come up with such a view. Yes, I've read his explanation, but to me it seems, hm, "odd". Sorry.
The whole scientific world is interested in backlinks: they are called citations. Because when you look a paper A and know it has been cited 10 times vs paper B with 1000 citations, then you know something about paper A's popularity/reach at least, maybe even something about its quality.
Backlinks tell you, what the fan-in of a zettel is. And a zettel with a high fan-in to me is noteworthy. A zettel with a fan-in of 0 also might be noteworthy - for different reasons.
Fan-in and fan-out of a zettel tell you something about its role in the network of zettels.
And about the context: Sure, if a link from zettel P to zettel C is just sitting on a line with "see also" it's maybe not that useful when you view zettel C and see a backlink to P. Or maybe it is. Who knows what it tells the innocent viewer of zettel C, that of all zettels it's P that's linking to it? Well, that of course requires zettel P to have a meaningful title and not just a timestamp. But that to me that is how zettels are naturally named.
Now, add to that if in P the link to C is not just tagged with "see also", but embedded in a whole sentence. And now imagine, a tool would show - when looking at C - not only that P is linking to it, but also "in what context" the link is embedded.
When looking at C I get a list of other notes linking to it (backlinks). I get a count of such backlinks (fan-in). For each backlink (use of the link to C) I also get the context in which it sits. To me that's tremendously useful to get this "out of the box" from a tool like Obsidian.
It's free, it's without effort - and if I don't care, I switch the backlink view off. Or if I want to add context or effort, then I can explicitly go to C and enter a forwardlink to P. But why would I want to do that?
Sure, in Luhmanns Zettelkasten backlinks were very, very expensive to list/count. But with a software based Zettelkasten they are cheap. To me there is no reason why a modern Zettelkasten software should not provide them - at least for those who find them useful.
Imagine going to a libary, browsing the shelfs, looking through a book, you find it interesting, you check out the literature references to find more interesting books.
Ordinary.
Now imagine, the book also had automatic backlinking switched on. You'd see all the books referencing this book. Not ordinary anymore!
Don't you think there could be a treasure hidden in backlinks like in the forward references?
@henrikenggaard Looking forward to read your response posts! Sharing our experience and dissecting what we actually do know and what is worth exploring further is, I think, a (the?) path forward.
I btw don't have much to add in terms of dissectable experience, only a couple of anecdotes and self-observation. I navigated in my archive through the search for a very, very long time, and neglected forming proper networks of departments. When I browse through these orphaned notes, which share a very loose connection (see "Kinds of Ties") for most of the time, I am forced to make sense of the search results list on the fly. And that's so much worse than having better connected notes for similar topics because apart from (often far from perfect) note titles, I have no hints! So when I search for some detail in Ruby programming, I have to sort through a mess, while I can look at my more recent notes about text editor programming on the Mac and dive into the sub-department of my Zettelkasten through different entry points, though trails like
How to make scrolling not lagthat starts with a list of scroll performance inhibitors or similar concrete questions I could ask Google to answer for me. The list of un-annotated, incoming links to a note, aka backlinks, that you get when you just search for a note's ID, that's more like the 100 assorted Ruby tips I have collected in the early 2010s: it's a curiosity to look at, but it's work to deal with it, and there are other mechanisms thatMost here seem to be arguing pro optionality, while Sascha is pointing out the opportunity cost. E.g. leaving options open could be dangerous when they enforce bad habits because you allow yourself to be lax at times.
Like when you want to maintain your weight while working from home in 2020, but there's a bowl of your favorite candy on the table that wasn't a problem for your circumference until you were forced to sit next to it all day, every day, because of quarantine/lockdown. Slowly you pick up the habit of reaching for candy instead of noticing you are getting hungry and should prepare a meal rather sooner than later. Sure it's could be nice to have the option to treat guests, but having a bowl of candy on your table could destroy the rest of your eating habits, badly influence your blood sugar, concentration, and ruin your Keto diet, or whatever.
It's not like anyone is saying "Backlinks do not exist!". That'd be ridiculous. But is the warm and fuzzy feeling we get from browsing around with backlinks a good proxy for the utility of relying on them in our work? Do they actually make us better as much as we think they are? (And does it matter, or is the warm and fuzzy feeling all we're after? In that case there's no point in arguing, enjoy yourselves and that's it.)
Author at Zettelkasten.de • https://christiantietze.de/
Where's the addiction lurking? I don't get it. I don't need to do anything for backlinks; they just get created. I don't need to look at them. But when I see them, they might help to assess the context of a zettel. I'm probably too ignorant to understand what harm that might be able to cause.
If "browsing around" gives you just a warm feeling and is not useful, does not help you achieve a goal, then backlinks won't make it worse.
If studying your network of zettels by following links helps you achieve a goal, then why would backlinks be links of lesser value?
I'd even say, there are no backlinks. There are just links, connections, relationships. Some are created explicitly, some automatically, some implicitly. Why limit connections to unidirectionality?
That's why some tools are offering to show not only backlinks, i.e. manually placed forward links viewed from the referred to zettel, but also the use of words/phrases which match zettel titles. Such a mention would be to use "liquidity" on some zettel w/o a link. And it still could be listed as a potential backlink on the Liquidity zettel.
What's not to like about that? How cool a feature so I can actually discover - instead of plan - relationships between zettels/topics!
Lack of a feature in a tool should not lead to making it a virtue not to want such feature. Backlinks are a technical possibility. They are here to stay. So I'd find it more helpful to explore how to use them best, instead of emphasising some harmfulness.
I don't think backlinks are like knifes. But even knifes are in widespread use, despite their potential to be used as a weapon. So, better make people conscious, responsible users of backlinks, than keeping them away from the feature.
Some of the reasons are given in the article.
I am a Zettler
This discussion may be suffering from an apples to oranges comparison. I've been using Roam for a while, motivated by the ease and flexibility with which information can be stored and cross-linked. I thought these capabilities would be great for a Zettelkasten.
But, after several weeks it became clear to me that Roam was not well-suited for developing a Zettelkasten in the classic Luhmann sense -- as it is automated in The Archive.
It is possible to use Roam in that manner, but you have to cripple its functionality to do so. Or, you have to extend your definition of zettelkasten (lower case) to include any knowledge storage system incorporating links -- and that might not be useful.
Roam (and I believe Obsidian) works in fundamentally different ways from The Archive. For example, a link in The Archive is (a) created ON a note and is (2) separate from the information documented in the note. It might look like this:
[[202011200545]]
or maybe
[[202011200545 Title of this note]]
A link in Roam is typically entered (a) on a Daily Page in an outline style where it is (b) an organic part of whatever information is being entered. The link is not an add-on to the "note" - it is the "note".
A link in Roam might look like this:
In Roam, the first-level, bi-directional link is typically from the Daily Page where information is recorded to a page (or pages) where that information can also be found--a much different structure from that used by The Archive.
The complexity that arises when trying to characterize the nature of links in The Archive does not arise in Roam because you can characterize links using plain language as part of the process of creating the Roam equivalent of a note.
Characterizing the nature of a link is not a limitation of language or the ability of the person creating the link to express his or her intention. It is a limitation of The Archive's method of creating a link. This limitation and its influence on the desirability of backlinks may be relevant to The Archive but not to Roam.
In the example above there are actually five links because this "block" (each bullet in an outline is called a "block") can now be found on the pages for the two papers, as well as the pages for the three tags. Multiple-link constructions like this are optional -- I find them convenient, empowering -- others may find them confusing. There is no requirement to use them.
Because Roam makes use of an outline metaphor for data entry, a link may be even more complex. The link example described above could contain child-blocks and grandchild blocks. Those "descendants" would be part of the link. if you go to the Page devoted to Paper A, you would find the top-level block date-stamped and displayed in a collapsed state -- with all child, grand-child, etc. sub-bullets immediately accessible by expanding the highest level block.
Is this a good system? Maybe no if you think that a properly constructed zettelkasten is really The Zettelkasten and that The Zettelkasten requires that specific rules be followed. Maybe yes if you think that a zettelkasten is one of several possible knowledge management systems, perhaps inspired by The Zettelkasten, but now offering alternative approaches.
Further complicating these discussions are other fundamental differences in approach between systems like Obsidian and Roam and The Archive -- the nature of links being only one, and IMO, not the most important.
Sorry, @bjbarry, I don't follow:
In Obsidian you create zettels as files. It's explicitly based on (local) .md files. How you name those files is up to you. There is a plugin for naming them with timestamps. Or you use timestamps + a topic, or a topic only. Up to you.
In what way is that different from The Archive?
Yes, the display is different. The Archive is always just showing a single long list of zettels, if I remember correctly. If you want that, well, Obsidian also provides a long list of all the files/zettels you created. But who wants to look at a 10K list of zettels? So the essential functionality is search, I guess. And of course search delivers a long list of zettels in The Archive as well as Obsidian.
Yes, the "entry point" is different. The Archive has only one: the list. Obsidian provides you with two: the list or the daily entry. And I have to say, I have come to like the daily entry very much. I was skeptic at the beginning, but now I find it a differentiator compares to The Archive or Notion.
When I write a new zettel I do it in the daily entry by entering its link, e.g.
[[My new zettel]]. Then I open the new zettel and write down whatever I want to take note of. For the next new zettel I do the same and so on.The result is no different from The Archive: zettels accumulate in a long list, the folder where I want to store them.
But I get more than that: I also get a simple daily context. I can go to any day and see which zettels I created back then.
Also I can (and do) use the daily entry as a small diary of sorts. I can add stuff that does not need to go into a zettel. That's not essential of course for a zettelkasten. But nevertheless I find it an opportunity I'm using quite often. And of course when I search for something there is no difference between an official zettel and a daily entry.
Of course, if I'm looking at a zettel I found through search and want to create a related new zettel, I can go right ahead, enter its link, and move to the new zettel. Thereby the web is woven en passant.
Roam goes beyond that in that it does not use physical files to store data. It makes each and every paragraph a potential node in a network. Obsidian is somewhat moving in the same direction. I can link to headlines inside of files. But those links are kind of brittle, because in Obsidian they are relying on the actual title, not an underlying paragraph ID as in Roam. So one might be careful to use them.
The main links in Obsidian and The Archive, though, are file names. And Obsidian does a good job of updating links should I choose to rename a note or move it inside an Obsidian vault.
If you look at the resulting files there is no difference between what The Archive produces and Obsidian, I'd say. It's a long list, each named however you like it.
And if you choose so you can add folders to organize those files. Or tags. If you don't like that, then just don't. I just would not want to tell people that's "non-canonical".
If zettels are "information nuggets" (I want to avoid the term "note") stored in individual files for longevity, then The Archive and Obsidian are both tools to develop a zettelkasten. And I very much like the competition that has sprung up in the past 12 months in the zettelkasten tool arena. There is a lot to explore.
Some food for thought! Thanks @bjbarry
This I'd change. The criteria are not based on ones mental model of a ZK but if the system actually makes the person more productive.
EDIT: I think @ralfw is correct. The Archive and Obsidian are more similar than Obsidian and Roam.
I am a Zettler
@ralfw - Obviously, my understanding of Obsidian is weak. Sorry for generating the confusion by assuming it aligned more closely with Roam. When I alluded to additional differences between Roam and The Archive I had the Daily Note in mind - an approach also used in Obsidian. I rely on it heavily and it is very different from making entries in The Archive.
As @Sascha correctly notes, the measure of value is productivity. Direct comparisons on a feature by feature basis (e.g., the value of backlinks) is hard because the overall structure of the applications is different. What works or is easy in one may well not work or be difficult in another.
@bjbarry: As I said before, I find the daily note a very useful addition to the general list of zettels. It provides another view - which one can use, but does not need to use. I find it easier not to use in Obsidian than in Roam.
Zettels benefit from context. To create context or to "encode it" in a zettel, is hard. To anchor new zettels in the daily view is a form of very natural context to me. And if the distinction between zettel and note gets blurred by that... then I'm quite happy. Zettels get stored in a dedicated Zettelkasten folder. But I can link to them not only from daily notes but also from any other note I care to store in Obsidian. And I can link to them through Obsidian links from all other applications as well, e.g. a calendar entry or a task or a Word manuscript for a magazine article I'm writing.
To me features inform the method, and the method informs tools to implement features. It's a co-evolution. Practice + theory, not one over the other or just one of them.
I agree with this. If one system encourages you interacting with the
system itself with positive feedback loop. It is the system for you.
As long as the user constantly reviews his/her methodology, it should
not be an critical issue.
Constantly? Why? To what end? Wouldn’t constantly using your methodology prove its worth....or Not? And if not, it should become self evident. I guess I’m getting hung up on my perceived notion of wasting time reviewing rather than using.
Reviewing the Zettelkästen is fun for me. I set a goal to achieve
everyday, and I feel that I have more time concentrating on writing
atomic notes, which not necessary being productive on my work, but I
feel much better life satisfaction. I used to do daily reviews, weekly
reviews, and monthly reviews, but ending just weekly reviews and
monthly reviews now. During these reflection time, I will try my best
to decide the current methodology works or not.
To constantly using a methodology is the goal. It can be qualified by
how much time I spend in my Zettelkästen. I did this on daily basis
with time tracking tools. It strongly correlated to the goal that I
set, i.e. how many ideas I write every day.
I am not saying it is bad idea, but I feel that not having a single
determined goal will dilute my motivation. I feel just catching how
many notes I write is a better solution.
For me reviewing the Zettelkästen content is one of the way I use my
Zettelkästen. Being able to interacting my own thoughts is a flow
process for me.
I agree with @learning_ran in spirit, but perhaps not how to do it.
There are different levels of knowledge work:
I base my refinments of the method on the inner mechanisms and the nature of knowledge.
Example (automated translation of an original ZK note I am currently working on, copyrighted etc):
Iain McGilchrists sees a difference in the brain hemispheres in that the left brain hemisphere focuses more on individual parts of experience and perception, while the right brain hemisphere focuses on the totality of perception.1.
The right half of the brain seems to focus more on the shape of an experience, while the left half seems to focus more on sequences, sequences and similar separating concepts.2
Patients with lesions in the right hemisphere of the brain lose sight of the big picture34 In particular, they lose the ability to form a visual image of an object, for example, if they have only touched it. Not because they have problems transferring haptic to visual. Rather, they lack an eye for the essential, the whole, from which the transfer would be possible.1
This problem rarely occurs in patients with lesions of the right hemisphere of the brain. However, problems are more likely to be manipulated (which is more in the domain of the left hemisphere).5.
A productive method allows for harmonious engagement of both hemispheres. Sadly, it is difficult to actually develop processes that prevent such problems. However, at least -- if understood -- you can use those insights as a filter against features and techniques that throw of the balance of the hemispheres. In Germany this is described as gehirngerecht (popularised by Vera Birkenbihl) which roughly translates to brain-friendly.
The nature of both hemispheres allows to assess one of the main obstacles in the collective thinking process. The left hemisphere tends to grab details and forces them together in the hope to make a whole. However, because it is not able to do so, the result is something akin to the Frankenstein Monster: This happens if there is no bigger picture behind the development of the software (or the method). This is what we observe as feature bloated software or "nice to have the option"-mentality (no dig!). Or on the individual level: As the inability to give life to something that is organically growing and nurturing it, and instead use systems, methods or whatever until use becomes misuse and results in abondenment and finally death (jumping from system to system or regularly starting all over again, resetting etc.).
The whole topic of knowledge work is a pyramid standing on a fundament of the inner mechanisms and the nature of knowledge, with a thick body of methodology and a small tip of software.
Iain McGilchrist (2009): The Master and his Emissary. The Divided Brain and the Making of the Western World, Totton: Yale University Press. ↩︎ ↩︎
J. L. Bradshaw and N. C. Nettleton (1981): The nature of hemispheric specialization in man, Behavioral and Brain Sciences 1, 1981, Vol. 4, S. 51--63. ↩︎
J. McFIE, M. F. PIERCY, and O. L. ZANGWILL (1950): VISUAL-SPATIAL AGNOSIA ASSOCIATED WITH LESIONS OF THE RIGHT CEREBRAL HEMISPHERE, Brain 2, 1950, Vol. 73, pp. 167-190. ↩︎
GEORGE ETTLINGER, ELIZABETH WARRINGTON, and O. L. ZANGWILL (1957): A FURTHER STUDY OF VISUAL-SPATIAL AGNOSIA, Brain 3, 1957, Vol. 80, pp. 335-361. ↩︎
J. MCFIE and O. L. ZANGWILL (1960): VISUAL-CONSTRUCTIVE DISABILITIES ASSOCIATED WITH LESIONS OF THE LEFT CEREBRAL HEMISPHERE, Brain 2, 1960, Vol. 83, pp. 243-260. ↩︎
I am a Zettler
I have to agree to disagree, Sascha. Your argument sounds plausible, but if you take a step back, you realise that any complex knowledge matrix is extremely difficult to navigate. Backlinks are simply one method of enabling knowledge managers to navigate the matrix. Indeed, they can serve as a valuable first step in constructing a more organised “knowledge-building” structure by reminding you of relationships you may have forgotten.
Consider: after using a given backlink, you can always incorporate it into a more explicit, more definitive structure (if it seems appropriate to do so). The best of the backlink-capable apps (e.g. Obsidian, NotePlan beta 3) I’ve used also provide the user with some context for the backlink, which undermines your distraction argument: I would argue that establishing context is a vital part of building knowledge.
Building any kind of knowledge base is extremely difficult, in that there is always considerable mental overhead involved. I have enormous respect for the Zettelkasten approach, and agree with your criticism of the innately linear/hierarchical Folgezettel sub-approach. But there’s nothing linear or hierarchical about backlinks, and they may highlight relationships that enable you to broaden your lateral-thinking ability (another vital part, I would suggest, of knowledge building). That backlinks (especially automatic backlinks) can become intrusive and annoying I wouldn’t deny – except that in practice, and based on thousands of words of notes, this rarely happens. Instead, my experience suggests that many backlinks eventually turn into intentional links, and may in fact result in more intentional links than one might otherwise have remembered to create.
As for your parallel with the Internet: well, the latter’s profusion of hyperlinks only becomes distractingly confusing if you use it in an undisciplined way. If you take a disciplined approach to hyperlinks, you can use them to build enormous webs of knowledge very rapidly (to take a very simple example: by opening links in separate tabs/windows rather than just clicking through). I do this all the time when conducting rapid web research in my capacity as a translator/copywriter, just as I use multiple search engines to find page sources in the first place. Much of this material will then make its way into dedicated knowledge bases – which I may choose to use individually or in combination, where knowledge in one particular field can help to clarify questions in another one. The same applies, for example, to legal cases or documentation, or various aspects of engineering. Discipline is also something that needs to become automatic ;-)
I cannot confirm this empirically. My Zettelkasten is very easy to navigate, it got easier the more complex it got and hit a ceiling since a certain point.
I will be more clear in a more general text on backlinks in the context of knowledge work: I am not arguing against backlinks as a navigation tool. I argue specificely against the the surplus of knowledge through connection that links (can) provide (if placed properly).
Am I correct that you deem the note preview to be a link context? If yes, then:
I do not consider a note preview proper link context as a note preview does not highlight the nature of the connection. Proper link context gives you past and present reasons to follow the link. That means that one need to have written about the link nature in the first place. Otherwise, you will have to do work that already should have been done: Investigate the nature of the link.
Can't argue with experience.
That would fall under the "not very useful" category. There is not no use. But the subpar quality of backlinks compared to other actions impose opportunity costs that decrease productivity (per time).
In the light of the following quote of yours: One part of the benefits of a ZK is that it allows you to focus on one thought at a time (technically, a link is a thought about the connection between two thoughts). The ZKM is not only about the setup and the thing "Zettelkasten". There are habits and processes baked in that lifts mental load like having to keep several items in your working memory more than necessary.
I need to formulate it more carefully to give it its due but to hint at it: There is a equivelant to the "distraction free" in "distraction free writers" that is part of the ZKM as I formulate it. From my experience few people have the working memory to hold much items necessary for some thinking processes. The aid a ZK gives you is partially by functioning as an extended working memory. There is a very good article by Anders Ericson called "Long-Term Working Memory" which gives a model on how the brain extends the working memory, especially in experts. To give rise to this effect the ZK-Environment (software or paper) needs to reduce the cognitive burden. One of these burdens are backlinks.
The issue is that you can't just switch them on and off how you like and expect your brain to work as if they are there or not. If you are aware of this function your brain will model your perception as if it was there. That is the reason why even switching of your smartphone is not enough to lift the negative effect of smartphones on your fluid intelligence and working memory. Only putting it in another room is enough, if one might assume that you get back to baseline (which I doubt). Another example of this problem can be felt. For a distraction free writer you even need to remove features from the software to reap all the benefits.
I am a Zettler
This is an excellent discussion. I see this as 'Different strokes for different folks". And to me, a big mistake many productivity software developers and users make, is to assume the rest of the world thinks and works like they do.
I think one factor here is the length of time you'll need to use your links. When you use a link to jump from one item to another, you might use that link for the rest of your life or professional career (if for example, you teach or do academic research) or, you may only use that link for the next minute or so to snatch just one morsel of a thought, which probably is the case for most of the general population. Suppose on my computer or mobile device I need someone's telephone number quickly, or the location of today's meeting real fast. That's a link, and the backlink to return to my original thought or note may be nothing more than the back-arrow on a standard web browser. In my life and my personal organizer software, that may compose 90% or more of the links I use.
So the vast majority of links many people use may be needed for only a minute or so, to solve quick temporal questions on daily routine tasks and issues. For people like that, backlinks just aren't necessary, and as the original author noted, can be wasteful of time and productivity.
So frankly, I see backlinks as they are promoted in some productivity software today as just an overblown, unnecessary feature, probably useful for only a small subset of the general population.
I would agree with this article. Whilst I appreciate that backlinks may help others navigate your zettelkasten, it doesn't help me build mine. In the end my zettelkasten is for my use and as @Sasha importantly said, for my growth in knowledge. A computer cannot link knowledge, it can only link words and terms. I believe it is most dangerous to allow anything to create links in your zettelkasten other than yourself.
I'm sure if I looked at someone else's zettelkasten, I could easily see links that I think should be created. But those links do not necessarily demonstrate knowledge to the owner of the zettelkasten. In the same way I would not allow another person to create links in my zettelkasten, nor would I allow a computer to do so. It is my second brain, not a playground for an algorithm. Links created by anything/anyone else in your zettelkasten do not demonstrate a growth in your knowledge or a connection of ideas. They can only point to potential connections and if your zettelkasten is full of potential connections created by software, it ceases to be a second brain of what you actually know otherwise I could claim that the Internet is my zettelkasten.
I continue to be puzzled by the negative reaction that people have to back links. That may be because I am using Roam where two-way links, together with the Roam outlining metaphor, block-level links and use of a Daily Note Page are thought to create the strength of the application.
I don't think it is because back links are unsuitable for use with a zettelkasten. Roam instances can be "multi-player." There are currently 2,500 Roam users participating in a six-week "book club" using a Roam instance devoted to "How to Take Smart Notes."
This is the second such book club devoted to Ahrens book, the first had over 1,000 participants. Ahrens participated in the first and recently complimented the enthusiasm with which the second was being conducted.
So, either these thousands of people are delusional or it is perfectly possible to create a viable zettelkasten using software that employs back links.
If these back links are, in fact, structural weaknesses, then that is a real problem for Roam because interest in the Roam community for zettelkasten methods is intense.
But, leaving particular software features aside, I think that back links are important -- at least to me -- in other aspects of life.
For example,
What happens in "real life" when a back link is not available? You might hear:
Back links in Roam are created automatically, are unobtrusive and useful, but are strictly optional. There is no requirement to use them, but most people do, and do so frequently.
This is not an either/or. It's not reasonable to assume they're all attempting to build "pure" Zettelkästen, or trying to build Zettelkästen at all. It's also possible that they don't depend much on backlinking. It's most likely that they are not yet hitting the problems of scale that backlinking can generate when representing knowledge and argumentation.
I don't think that the argument was ever "a ZK built in a system with automatic backlinking cannot be viable". As I read it, there are two main prongs:
I don't think this is a very effective set of arguments because they don't model backlinking. An analog to backlinking would be:
… which is exactly what Sasha was arguing happens to backlinks, at scale, in PKMs.
The difference between that and backlinks is that, to your point, backlinks are useful in some contexts. Sasha's point is that they are an "attractive nuisance" for someone building a lifelong body of thousands of notes. Structuring knowledge -- and knowledge in particular! -- with them seems useful until you hit real scale.
I'm reading all the discussion a second time after some months.
I use Obsidian, it has an automatic backlinking feature, but I don't use it.
It doesn't work for my mind.
I acknowlegde with some arguments pro backlinking emerged in the discussion, but in my experience I've found the same problems that Sasha has explained (before reading him :-) ) and are very heavy in my case and superior of pros.
1. backlinking doesn't provide context linking. I feel the exact problem of Sascha, I need to have a why, thinked before and represented after, for the links I put.
2. a list of backlinking links is flat. They are all equals. If there are many, it's a mess.
3. backlinking, above all automatic backlinking causes an overlinking effect in the notes, effect that make linking less effective. Linking is useful but I need to keep managed.
For me linking is an effort, a friction in note-taking, but is a good friction.
Manual linking is thinking. Automatic linking is something like collecting.
After I wrote this, I think the discussion arises because the same simple mechanic of linking means and implements very different things. Enumerating some of them, links are usable for:
1. make relationships between things: concepts, ideas, and so on
2. make a browsing aid (the perfect example for me is a wikipedia-like linking implementation, with which I can easily move between pages)
3. make information architecture, or structure (the example here is using links to make structrure notes, outlines, or whatever)
4. and so on
Link purpose are many, and is different the importance and relevance of each use, too.
The problems or backlinking are more critical for some of these uses, less for others.
In the context of creating notes in Zettelkasten, the most important links are of type relationships, and in this case using backlink instead of thinked linking is a bad use, according to the principle already expressed "linking is thinking".
Despite being aware of this, anyway, I don't have in my notes only relationship links. I need relations between my notes, but also support for navigation. So, besides forming primary link between notes, I also make other layers of linking:
In some of these uses, having two links, forward and backward, between notes is useful.
I have four different ontologies of linking, I don't need the fifth provided by obsidian backlinks :-)
All the links I do are manual made. they are an effort, but this effort is thinking. For every link i do I have to think if, why, how and to which I make this link. A link placed must be "my link", not a link of a software.
The big problems of tools like obsidian, for me, is that don't differentiate between different kinds of links. So, I need to use some tricks that visually render different links. Using different positions, adding a prefix or suffix or texts, adding icons or callouts, even making an appropriate use of spaces.
In the discussion emerged semantic linking, is a relevant theme for me.
I find its automatic backlining of Obsidian useful for things like
In the next years the question of automatic backinking will resurface in a different form, with the advent of the AI.
Automatic classification, automatic linking. they will be more powerful and invasive than backlinking.
For what I've written in the previous post, for me this will be a risk.
@andang76 said:
If you are an Obsidian user, the plugins Supercharged links and Breadcrumbs provide some standardized tricks for indicating link types. Such plugins can help to implement and automatically represent the kind of conceptual schema of note types and link types that I described in a recent comment in another discussion.
Yes, I already use supercharged links a bit. It is a good help.
I like the idea of breadcrumbs, but it has an ugly interface :-)
I'm always doubtful about depending heavily on plugins.
I want to maintain my notes for years, external features could not survive in the future.
A good tradeoff at the moment is annotating the notes with textual tricks.
For example, I use something similar just described in the comments above, i call them "weak tags", they start with a '@' and placed near some relevant links. One of the tag is @bible, indicating a link to a source note that is the most important source that I've used for create the note.
Today my insight has suggested me, wanting to insert in my notes what I've written tonight in this topic about backlinking, making a note title "What I've written about Backlinking (my)" and placing into the note the link at this page with tag '@where' :-)
Conventional positions, artifacts like weak tags, conventions in how writing links help me to differentiate links, at least in their rendering into the note. They are not really semantic constructs, they don't have a special treatment in Obsidian.
@andang76 said:
The software dependency problem is not limited to Obsidian plugins; it applies to Obsidian itself and most other apps. The solution is to use syntax that can be converted later if needed.
If you learn (or already know) some tools for data wrangling text using regex, you can use Breadcrumbs link-type syntax in your notes now and then convert the syntax to something else later if other software comes along that uses different syntax. In other words, why not use the syntax that current software supports and then convert the syntax later if that software goes away, instead of using custom syntax that no software supports? That's a question to consider if you haven't considered it. It's not a question of semantic/conceptual schema, because you can use the same schema with either syntax.
I could apply this reason if a plug in like breadcrumbs were a perfect suit for my needs and were strategic in this . But it doesn't, it implements only a partial set of link typing, hierarchical. I would pontential use only the parent relationship, thing that already do without plug-in in the use I have of that, using metadata.
Either breadcrumbs and "mine" are not perfect, but mine is more tailored and hasn't an external dependendency to manage.
For other plug-ins, like templater, I make other considerations.
@andang76 said:
The Breadcrumbs link-type syntax is supported by other plugins too, like Dataview and Juggl. There is nothing explicitly hierarchical about the syntax (it just adds a property to the link that the various plugins interpret based on user preferences), and there is nothing about the syntax that requires any external dependencies: it's just plain text!
Personally, I would not use your link-type ("weak tag") syntax because it is the same as Pandoc's citation syntax and the pandoc-crossref filter syntax for cross-references to figures, equations, and tables. I make sure that all of my Markdown syntax is as compatible as possible with Pandoc's Markdown, which is a crucial tool for the kind of conversion I talked about in my previous comment. One could convert your custom syntax before running Pandoc, but that seems like an unnecessary step to me. If you're not familiar with Pandoc, I highly recommend learning about it.
Since I use what might be called backlinks, I don't have a regulation Zettelkasten. I have a Xettelkasten. My Xettelkasten, hereinafter XK, has alphabetic index notes. The note on Xettelkasten has the link [[0000.0000.0VWX]] V-W-X to the V-W-X alphabetic index note. When I visit the V-W-X note in Zettlr, the "related files" pane shows me the notes that link to it. This technique gives me an alphabetic index for free (except for adding the index note link)1. But the presence of such backlinks, intended to be traversed in reverse like the "what links here" link in Wikipedia, means that I cannot claim to have a ZK. I have an XK.
1. Not to mention the "emotional labor" I incurred by writing this comment, as if anyone asked.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.
@ZettelDistraction said:
"An alphabetic index for free": that's a great point, XettelDistraction. Also note that automatic backlinks give you a subject index for free if you have a note type analogous to subject headings when you link to notes of that type (a.k.a. "links as (subject) tags"). Admittedly, one can usually also get such alphabetic and subject indexes for free by using (non-link) tags. Perhaps those who condemn automatic backlinks opine that the people should use tags for such free indexes, not links as tags?
Thank you for your service, XettelDistraction.
I have such features as well in my Xettelkasten.
Signed,
XettelDistraction.
GitHub. Erdős #2. Problems worthy of attack / prove their worth by hitting back. -- Piet Hein. Alter ego: Erel Dogg (not the first). CC BY-SA 4.0.