Since my views have been discussed, perhaps I should clarify them. Let me say this again, out of the four baseline criteria Christian Tietze gives, I do accept three and have problems only with one. I do believe that direct links are more important than he seems to think they are. In particular, I take issue with the claim that: "if full-text search works, manually linking notes will work, too: just put the target’s identifier somewhere, copy it, search for the identifier, and open the resulting note."
This claim is not false. Manually linking notes by identifiers is always possible. In fact, if you use a paper-based system, manual linking is the only possible way of connecting notes. You can, of course, restrict yourself to this way of linking in an electronic version of a Zettelkasten, but I have no idea why you would want or need to. In fact, I believe it is a mistake to do so.
Christian, if I may, finds that my view amounts to favoring "[[WikiLink]] note connections because they are so darn fast to create if you know the title of the note you want to link to." His reason is: "That’s a situation I don’t ever find myself in. Instead, I have to fish for the correct title of a note with a search anyway." The initial quickness of the free-link approach makes ls only one of my reasons. I admit that, subjectively speaking, it is important to me.
However, it is not the most important reason for preferring free links. More important is what I take to be an objective reason, namely that direct clickable links are more effective than indirect links, even if they involve note "identifiers." If your notes have names like this: "201105160958" (as I gather they do from some your claims), then you could actually directly link to this note like this [] in an application that allows free links (and it dos not have to be a ConnectedText nor even a Wiki). So you would not need to think of a descriptive name. If you have back-link capability, as in ConnectedText, it would be trivial to change the
In any case, using unique IDs and links is not contradictory, and I never meant to suggest that they are. The two approaches can well be integrated, and my argument was not directed against unique IDs, but against the idea that indirect linking is just as good. It isn't. The free links make hard connections between topics that allow all kinds of other operations. We have not spoken of semantic links, categories, or other relations, and perhaps we should.
If indirect linking works for you, that is fine with me too, but I do think it is less than optimal. None of the arguments you present changes that fact: "just put the target’s identifier somewhere, copy it, search for the identifier, and open the resulting note." That is, you need at least three steps for something that takes one step with direct links. (I know that this introduces the notion of quickness again, but, again, I would say that quickness is not everything.)
Someone commented that I did not understand linking in Devonthink. I never claimed I did. In fact, I think I admitted limited acquaintance with the program. Still, the claim that Devonthink does not allow for free links remains true. Auto-linking does not substitute for direct linking based on a user's decision. One of the main troubles with auto-linking is that the link between a word or a phrase and the page to which it refers is broken, if the page is renamed. If a wiki page is renamed the referring link is also renamed and the link is retained.
Nor do you need a wiki for direct links. nValt and other applications (including Devonthink) do direct links. But they do so without backlinks, of course. Furthermore, links in these applications break, if the page to which the link referred is renamed or changed.
It is true, you don't need a car for some purposes. I walk as much as I can, and I hardly ever drive to go to the grocery store or anything that is within a mile radius. But when I visit my daughters who live more than 800 miles away it would be impractical to walk. In the same way, a text-based approach works for some purposes, but it is not sufficient for my note-taking needs.
Oh ... and to DutchPete: "sliced bread," in my view, was not really as good an invention as everyone seems to think it is.
I should perhaps also point out that I am grateful to Christian Tietze and those who responded for allowing me the opportunity to think further about these matters.
1. Corrected on Monday, March 30.