4 ms·
Why Programming Is Like Dancing
- AlexeyMK 15y agoPlato's work has had a lot of support/maintenance over the years, as it has been ported from the spoken word to parchment to book to e-book, and finally translated to English, which is how I assume Dan read it. Programming languages and environments just evolve quicker than human ones. As with books, the best ideas get ported over: you can read Plato, but other than (Socrates/Plato/Aristotle), how many Greek philosophers were around whose ideas failed to survive? VisiCalc was first implemented in seventies for the Apple II, but its ideas have survived and been ported to Excel (and now Google Spreadsheets).
- dshipper 15y agoI really like this. It shifts the importance of the production from the physical e.g. the actual code, the actual book to the idea. Great comment.
- zavulon 15y ago> I guarantee that no software we write today will still be in use after 2,400 years. I don't know about that.. pretty sure COBOL software used by big banks written in 1960's and 70's will still be used.
- dshipper 15y agoHahaha I think universities will still be teaching Java too :)
- sjm 15y ago"if you program and want any longevity to your work, make a game. all else recycles, but people rewrite architectures to keep games alive." - _why
- dshipper 15y agoGreat quote! Most of my experience centers around developing small web/mobile apps where building ten new apps a year is pretty commonplace. I'll add games to enterprise software as the two places where you'll have the most longevity for your code.
- tobiasSoftware 15y agoI've found good longevity for my code, I work in image processing and file organization. I think I would categorize it as tools. I still use a lot of tool like software too, like Notepad, Paint, I use a video processing software called TMPG that I bought about 8 years ago, I occasionally use audio processing software that I bought around the same time, and I use some file processing software that I got a few years ago along with a basic video editor. I think when you find a tool type software you like, you tend to stick with it. I think the key is to make it as abstract from the system and algorithmic as possible. For example, with image processing, you have a small GUI component and you have a small image type to RGB converter, and the majority of the code is algorithmic RGB array to processed RGB array code. Algorithms don't decay much, but GUIs and file types do, so software that is very tool like and algorithmic has much more longevity as long as the small non algorithmic components are updated occasionally.
- kpennell 15y agoI liked this.