Links vs searches

cobblepot

I want to build on a previous popular thread and a exchange that @ctietze and I had on the Zettelkasten Discord about the practice of referring to automatic searches in software as "links" even though they are actually searches.

I wrote:

@ctietze I think it would really lessen confusion if you stopped calling them "links" - which they are not! They are saved searches for UIDs (or DTIDs, or just IDs). It is only in TheArchive that these act as link substitutes. They may be better than links, but calling them links confuses people and discussions. Rather than talking about them as links and then occasionally mentioning that they "actually (!) execute a search", just say up front that TheArchive doesn't use links, but uses saved ID searches instead.

Now it turns out (I think) that the popular ZK software Zettlr also implements "links" in this way. But since so many of these conversations are about workflow details and recommendations, it is a huge waste of time for there to be misunderstandings about these terms.

@ctietze responded:

@cobblepot overlooked this -- I see where you're coming from, but am not sure if I agree about the essence of what a link is, and what it's not

First, let's all agree on a few things. First, definitions are arbitrary and we can define things however we like. Second, there will be some gray area counterexamples to any definition, but that does not make defining terms pointless, nor does it mean that all definitions are equally good. There is no unassailable definable "essence" of a link that won't have some counterexamples, because language is messy and people use words in inconsistent ways. But concepts are useful for communicating with other people and community and definitions provide a starting place for assumptions. (If you don't agree with me on this, please continue in another thread - I respect your position but don't want to have that argument here.)

Here's a proposal for using the terms "link" and "saved search" in the ZK community when referring to connections between separate notes in the ZK. I am not arguing that links are better than searches or vice-versa, and I am not arguing that anyone is using these terms incorrectly. I just want to get clarity so that those of us who are primarily interested in improving our systems (rather than debates or language games) can talk to each other without wasting time miscommunicating. These proposed definitions don't aim to describe links in contexts other than between ZK notes, such as file or web links, although to me they seem to cover those cases decently well also.

Link: text in a note (the body content) that refers to a specific, individual, distinct note (whether that note is another file, another individual entry in a database, etc.)

Search: text used for a search (whether implicit/explicit, automatic/manual, etc.) that returns a list of files (however ordered) that include this text (whether in the filename or in the note content)

Saved search: a method for automatically showing the results of a specific search (e.g. clicking on text with specific syntax such as [[20200101011100]])

To me, links and searches are clearly different things and it creates problems to talk about them as if they are the same thing. They may accomplish similar goals, they may be equally good or one may be better than the other, but they are different ideas. If [[20200101011100]] executes a search that can return one or many items, but [[20200101011100.org]] executes a command to show one specific note, they are different. The former is not a link and should not be talked about as such. The latter is and should.

Please comment and let me know whether you agree, disagree, etc. and why!

edit: sorry this was put in the wrong channel - I'll ask the mods to move it.

ctietze

@cobblepot said:
Link: text in a note (the body content) that refers to a specific, individual, distinct note [...]

Saved search: a method for automatically showing the results of a specific search (e.g. clicking on text with specific syntax such as [[20200101011100]])

To me, links and searches are clearly different things and it creates problems to talk about them as if they are the same thing.

With this, you did mix the level of textual content with the level of affordances provided by the software. A saved search can be the way you execute a link.

There can be no saved searches inside a text, not in the way that you can "have" saved searches in your email or file management app. In a text, there's only text.

sbicknel

However they are implemented there are differences that distinguish a link from a search. To me a link is a record of a cognitive connection between two ideas and a search is a question that asks what other thoughts might be related in some way to a particular thought or to each other and that might be candidates for linking together.

I don't like the idea of relying on my software to tell me what's relevant. It feels too much like letting it do my thinking for me.

So when I do a search I want to find things to link together but the search results are not good enough on their own to establish that relevance. I need to be able to see a connection. Then I can document it as a link.

nistude

@cobblepot said:
Now it turns out (I think) that the popular ZK software Zettlr also implements "links" in this way. But since so many of these conversations are about workflow details and recommendations, it is a huge waste of time for there to be misunderstandings about these terms.

Actually, Zettlr does both. If you click on a note-link it shows you the respective note (thus it's a link) but it also executes a search for the respective ID and shows the results in the sidebar (thus it's a saved search).

cobblepot

@ctietze , let's bracket the debate about whether a link is an action or relationship, and whether the text level is different than the affordance level. We can address that but it's a bit of a semantic debate.

Let's get to the bottom line: do you really not see a difference between text that always refers to only one item compared to text that refers to a set of matching items?

ctietze

How does the text itself refer to sometimes one thing, and sometimes another, though? You're still secretly talking about the implementation in software. It's not text that does anything here. (Your whole point is that you can put the same text in 2 apps and get different outcomes when you click on a string of characters in double brackets. So it's not the text that changes.)

A saved search menu in an email client is an actual affordance to the user for a very specific feature. But we're talking about a string of characters here. It doesn't do anything, it just is. -- I do interpret stuff that looks like this [[202006040902]] to denote a reference to another Zettel. I used not to; instead, I used the § as a prefix and interpreted §202006040903 to denote a reference. For years, neither was click-able in my life. I did the lookup manually. Was this not a reference? Is Luhmanns absolutely unclickable paper note not containing links?

---

Spoiler: The answer is "yes." Luhmann's weird 1b34a/d scribbles are, in fact, links. If you show a single note to someone and don't explain anything, they won't know. It looks like garbage.

cobblepot

@ctietze you said many things but you carefully avoided answering my "bottom line" question. Please do so: "Do you really not see a difference between text that always refers to only one item compared to text that refers to a set of matching items?" (I assume your statement "The answer is 'yes'" refers to your own question about Luhmann).

There are many subtle issues you can talk about, and do, regarding the theory of information and reference, but they are not directly relevant to the distinction I am trying to draw. I was very careful in my first post to note that I am not claiming absolutes in definitions. I also pointed out that clever attempted counterexamples (like your Luhmann point) are NOT useful for making the discussion clearer. We can debate those later. But why fight this so hard? Do you not see a difference between links and searches?

Let's use normal language as the "implementation" to explain the difference.

The letters "cat" are a sign that refer to a type of animal. "Cat" refers to a large group of animals. But if I say "look at that cat" while pointing to a specific cat, I am referring to a specific animal.

The difference between links and searches is the difference between saying "The former president of the United States Barack Obama" (which, like a link, refers to one specific person) and "The former president of the United States" (which, like a search, is a descriptor that matches a group of people even if my goal is to eventually refer to one person among that group).

Yes, you can get all into "well, when you say it, what if I hear something slightly different?" or "what if I mispronounce Obama? does it still refer?" and a bunch of other semantic discussions that have nothing to do with actual workflow.

(Your whole point is that you can put the same text in 2 apps and get different outcomes when you click on a string of characters in double brackets. So it's not the text that changes.)

No, that's not my point. Is that even true? Regardless, I'm talking about different software using different basic models/methods of connecting different notes. My point is that it is confusing to refer to links and searches as the same thing when they are not. Links are one-to-one references and searches are one-to-(between zero and infinity) references. Very basic. So referring to them as links (actually (!) executing a search) is confusing and has lead to lots of previous confusion on the forum. Just look at the discussions about whether you can change "link" titles without breaking the connection to other notes, etc.

ctietze

I'm not even fighting hard, because I still don't believe we are talking about the same thing

Now I'm programming on stuff for a couple of years now and my sense of reality may be warped. I notice parts of this in conversations. Maybe I don't get your point.

Here's my answer to your questions: A link is not a search.

When I search for "any phrase at all" on the web, I get search results. I perform a phrase-based search. When I paste a URL into the search box of DuckDuckGo, like "https://csswizardry.com/2020/05/the-fastest-google-fonts/", and always click the first item below ads in the results page, what do I get in return? It's not a search for any phrase, I use it as a means to lookup a page on the web directly. -- I bring up this example not to be some modern era sophiste, but because non tech-savvy people have been using their browser's search box (aka "the Google box") for for years. for both run-of-the-mill phrase searches, and to enter URLs -- because Google got their back when they mistyped part of a URL, while the URL/regular location bar would just throw a "Not Found" or "Server Does Not Respond" error at you. That was before the advent of Omnibars in Chrome and later Firefox etc.

So when I type in a URL from a flyer into my DuckDuckGo search box, do I search for the page, or do I want to get to this one specific page and flat-out ignore the rest of the search results page? Is that any different from a link? In what important manner is that so?

Notational Velocity introduced the pattern to bring up the best search match in the editor automatically, based on literal title matches. Searching (!) for bread would not bring up banana bread, while a search for banana would. If you click on a button with "banana" on it and your app shows the "banana" note, why should people not conceptualize this as a link?

Remember that I do agree that all this is about implementation, which I think you should fall back to as well If user @fooMcFooFace decides to have a Zettelkasten on her teletyper, stored on tapes, having to print Zettel onto paper to read them, is she not able to link? Is the only able to link, say with the §ID convention, if the tape address/file name is exactly what comes after the §? What about file extensions? Is it a link only when she writes §someID.txt, but not if she writes §someID? After all, she has to locate files on the tape manually anyway, so there's no software that cares how she instructs her future self to look up a note.

(I don't even know how a teletype and tapes can be connected. I know that you can wire a teletype to a Linux server aka "run linux on a teletype" to some extent, but then again, that's already a hack.)

You didn't post this into the The Archive category, so this question is not in the context of The Archive's features and their implementation and the app advertising -- you're framing this in a more general way, and I think virtually everyone will be "D'oh, of course links are not searches, what kind of question is this?"

Our brave Foo McFooFace teletype user can link notes, Luhmann was able to link notes on paper, and I am able to link notes and look up the link target manually in my file browser. Is me eyeballing the file listing a search? Is it a link if I cannot click it? Did I not build a hypertext until 2 years ago when links in my life became clickable?

@sbnickel said it well:

However they are implemented there are differences that distinguish a link from a search. To me a link is a record of a cognitive connection between two ideas and a search is a question that asks what other thoughts might be related in some way to a particular thought or to each other and that might be candidates for linking together.

The software doesn't matter. I don't care that when I click a link, other stuff turns up below my target note. When you use title-based links in your software of choice, and don't get where you want to go, you have to switch apps or change your approach. Simple as that. I don't even care what my own piece of software might prefer, so I use Ids and am free to link notes

Basil

When I see @ctietze -- the developer of one of my most loved and important applications -- get bogged down in semantic discussions such as this one, I can't help but feel a little sad. Don't get me wrong, I think it's great that this community has grown so much in the last year or so, but there seems to be a strong negative correlation between the number of daily forum posts and how much The Archive is progressing, and discussions such as this one sure don't help in that regard. Maybe we can find a professional hairsplitter who can serve as a stand-in?