3 ms·
Ah, that's Dijkstra - programming is hard and programmers should suffer. He's right, though. Natural language is for communicating, not describing. When we rea
by Animats 6y ago
Ah, that's Dijkstra - programming is hard and programmers should suffer.
He's right, though. Natural language is for communicating, not describing. When we really need to describe something detailed, we need either visuals or formalism.
- pvg 6y agoHe's got an anti-visual-programming rant in the archive as well, although I'm having a hard time remembering the keywords to find it with, perhaps because all who seek Dijkstra's wisdom must also suffer.
- User23 6y agoI doubt Dijkstra would have objected to a visual formalism that was, well, formal. Something like Peirce's existential graphs[1] for example. Dijkstra despised sloppiness, probably as a consequence of growing up in a war-ravaged country where his group could only manage to beg an hour a week of computer time from the Americans. As you imply, the EWD is a treasure. I've read through the vast majority of the English language ones and it has deeply improved my understanding of "Computing Science." I personally enjoy his non-technical writing too, abrasiveness and all, but I know that's not to everyone's taste. If it isn't, it's easily skipped. I do wish I had some Dutch though to get the bits I missed. Edit: > He's got an anti-visual-programming rant in the archive as well It just occurred to me to ask, is this the one where he argues that any visual description by its nature overspecifies? That's one of the recurring themes in his work, specify everything that needs to be specified, but no more. Of course even a vector of glyphs is still in the end a visual representation, so things are always overspecified. For example a + b and b + a are, in some languages, exactly equivalent and it would be nice to be able to directly represent that rather than having it require additional knowledge of the semantics. [1] https://en.wikipedia.org/wiki/Existential_graph https://en.wikipedia.org/wiki/Existential_graph
- Koshkin 6y agoBut there are things that can only be fully described using natural language. Emotions, for example. There aren't any visuals or formalism for those. (And no, emoticons don't count.) On the other hand, description is a form of communication anyway, so I fail to see the distinction that you are trying to make here.
- cibritzio 6y agoJust curious why you suggest emoticons are glyphs insufficient for the communication of intelligence re emotions. They seem an almost incredibly insightful development in orthography. What earlier alphabetic cultures expressed so clearly via visuals interspersed among lithographs the essence of ineffable universal emotions?
- Koshkin 6y agoBecause a smile is not an emotion.
- cibritzio 6y agoBut are not emoticons smilies?
- DonHopkins 6y agoEmoticons are similes for smilies.
- lloeki 6y agoIf I say the word ‘ellipse’ (in French) you’d be hard pressed to know if I mean an oval or an omission. You can only know that from context. Body language works exactly the same, the exact meaning of a smile by itself is ambiguous, as it’s built from context. Adding back information previously lost via traditional sentence serialisation in text medium through text emoticons, graphic emojis, and other annotations (like /s) conveys this information which might entirely change the meaning of the sentence; emotion is completely tied to the semantics and critical to understanding. To me it looks a lot like didascalies (stage direction notes) in theatre play texts.
- andromeduck 6y agoAnd words are?
- wizzwizz4 6y agoLojban attitudinals do a fair job at conveying emotion, without really being natural language (specifically, mere words). https://mw.lojban.org/papri/ELG._Attitudinals https://mw.lojban.org/papri/ELG._Attitudinals
- toast0 6y agoI've never suffered so much as when programming in Applescript, which was supposed to be natural language programming. (Or for that matter, when trying to communicate with voice assistants, which are supposed to be natural language processing, but would be much easier to work with if they came with a list of keywords they understand)
- hutzlibu 6y ago"Natural language is for communicating, not describing. " But that communication can very well be clearly describing something precisely. Or ordering someone else to do something in clear, well defined terms. Do this. Do that. Do known method y, x times.