My random Ted Nelson encounter: by chance ended up sitting in the seat next to him on a plane to Portugal (we were both going there for a conference, but not the same one). Didn’t recognize him. He noticed I was doing something in vim so asked if I was a programmer, and I said, well sort-of, I’m a computer-science grad student (as I was at the time). I asked what he did, and he said something like “I do a lot of things, but what I’m famous for is the word ‘hypertext’”, which was a good hint to who he was.
The main thing he was working on at the time and wanted to demo was ZigZag, which is intriguing. It’s a take on knowledge representation and browsing for structured but schema-less and extensible data, vaguely in the space of what (in the personal space) people use mind-maps or org-mode for, but with ambitions to also be used in larger-scale software.
Out of curiosity, does it seem concrete to you what ZigZag is? I’m fairly interested in data structures, especially ones that are pitched as foundational. I don’t see how it is that, since it relies heavily on bidirectional links, which presumably need to be implemented as pairs of pointers, and you can’t build arrays out of it. But the paper doesn’t formalize what observable properties its three primitive types are supposed to have.
I’m trying to figure out whether there’s something useful here that’s simply being communicated badly - there’s a great deal of rhetoric on the site and in the paper about how it’s normal to not understand what’s being said, so I take it as undisputed that it’s bad communication, and just want to know what it’s trying to say. :)
I think his main interest is in conceptual data structures, i.e. “the structuring of data” in the knowledge-representation sense, rather than in data structures like red-black trees or hash tables and such. As far as I can tell, ZigZag comes mostly out of his dissatisfaction with other structures for data, such as spreadsheets, relational databases, and (especially) tree-structured hierarchies. Especially given the initial tool, which lets you build up these linked data structures and then pivot on different fields to give views that look array-like or other-things-like, that’s why I see it as more of a competitor to mind maps or org-mode, than to data structures in general.
It’s true he does also position it as a foundational data-structure in the CS sense, too, and I think he also believes that. My read of that is that it’s in the tradition of classic Lisp (which he’s influenced by), which positioned the cons cell as the foundational data structure that you can build anything from, for certain senses of “anything”. He thinks this is more complex than a cons cell, but still fairly simple. He has a kind of political/aesthetic opposition to anything particularly complex that can’t be understood by normal people. His 1974 classic book Computer Lib / Dream Machines is more or less a polemic against the idea of professionalized technologist, and a call for people to take back computers for themselves.
His style of writing definitely is odd. I think it works well in some cases and not others. It’s a sort of west-coast counterculture anti-authoritarian style, which is very well suited to a book like Computer Lib. I think his article about ZigZag would benefit from a somewhat different style, but that’s how he writes.
That’s all pretty fair. I appreciate the perspective check. :) The comparison to cons cells explains where it’s coming from, and especially why he wanted to talk about foundational structures. From a knowledge-representation standpoint, I really like this, but for the kinds of applications that come to mind for me, it just wouldn’t scale. It would work as an abstraction provided by an API, and possibly as the subject of a relational-ish query language, but not as a low-level encoding.
I can get past the style, it’s refreshing to be exposed to people who think differently. But at some point I do want to map it to the way I think!
My random Ted Nelson encounter: by chance ended up sitting in the seat next to him on a plane to Portugal (we were both going there for a conference, but not the same one). Didn’t recognize him. He noticed I was doing something in vim so asked if I was a programmer, and I said, well sort-of, I’m a computer-science grad student (as I was at the time). I asked what he did, and he said something like “I do a lot of things, but what I’m famous for is the word ‘hypertext’”, which was a good hint to who he was.
The main thing he was working on at the time and wanted to demo was ZigZag, which is intriguing. It’s a take on knowledge representation and browsing for structured but schema-less and extensible data, vaguely in the space of what (in the personal space) people use mind-maps or org-mode for, but with ambitions to also be used in larger-scale software.
Out of curiosity, does it seem concrete to you what ZigZag is? I’m fairly interested in data structures, especially ones that are pitched as foundational. I don’t see how it is that, since it relies heavily on bidirectional links, which presumably need to be implemented as pairs of pointers, and you can’t build arrays out of it. But the paper doesn’t formalize what observable properties its three primitive types are supposed to have.
I’m trying to figure out whether there’s something useful here that’s simply being communicated badly - there’s a great deal of rhetoric on the site and in the paper about how it’s normal to not understand what’s being said, so I take it as undisputed that it’s bad communication, and just want to know what it’s trying to say. :)
I think his main interest is in conceptual data structures, i.e. “the structuring of data” in the knowledge-representation sense, rather than in data structures like red-black trees or hash tables and such. As far as I can tell, ZigZag comes mostly out of his dissatisfaction with other structures for data, such as spreadsheets, relational databases, and (especially) tree-structured hierarchies. Especially given the initial tool, which lets you build up these linked data structures and then pivot on different fields to give views that look array-like or other-things-like, that’s why I see it as more of a competitor to mind maps or org-mode, than to data structures in general.
It’s true he does also position it as a foundational data-structure in the CS sense, too, and I think he also believes that. My read of that is that it’s in the tradition of classic Lisp (which he’s influenced by), which positioned the cons cell as the foundational data structure that you can build anything from, for certain senses of “anything”. He thinks this is more complex than a cons cell, but still fairly simple. He has a kind of political/aesthetic opposition to anything particularly complex that can’t be understood by normal people. His 1974 classic book Computer Lib / Dream Machines is more or less a polemic against the idea of professionalized technologist, and a call for people to take back computers for themselves.
His style of writing definitely is odd. I think it works well in some cases and not others. It’s a sort of west-coast counterculture anti-authoritarian style, which is very well suited to a book like Computer Lib. I think his article about ZigZag would benefit from a somewhat different style, but that’s how he writes.
That’s all pretty fair. I appreciate the perspective check. :) The comparison to cons cells explains where it’s coming from, and especially why he wanted to talk about foundational structures. From a knowledge-representation standpoint, I really like this, but for the kinds of applications that come to mind for me, it just wouldn’t scale. It would work as an abstraction provided by an API, and possibly as the subject of a relational-ish query language, but not as a low-level encoding.
I can get past the style, it’s refreshing to be exposed to people who think differently. But at some point I do want to map it to the way I think!