4 ms·

I'm curious as to which difficult concepts become easy to understand when presented visually. To me, the difficulty of programming has never been in its textual

I'm curious as to which difficult concepts become easy to understand when presented visually. To me, the difficulty of programming has never been in its textual representation. Just as in mathematics, the real challenges have always been related to conceptual understanding. Is there a good example of visual programming making difficult concepts easy?

tiborsaas 6y ago
A good example would be Max/Msp. It's used by a lot of musicians creative coders. Visual representation abstracts the boring pars of programming and enables experimentation much more rapidly. Quickly rearranging nodes is more intuitive than changing how you pass around objects.
slightwinder 6y ago
Showing the "big picture", the data-flows, the relations, etc. are all prime-candidates where visual presentation can sell better than text. A well made diagramm can be more informativ than any amount of text. On top, having your code in a structued source, you can easily manpulate the structure itself and get alternatice views. Ther eis literate programming which kinda goes into this direction, there you detache the sourcecode-organisation from it's technical organisation. Thouhg, this of course is not directly linked to "visual" programming, but work on the same conditions.
dkersten 6y ago
For me, personally, I think in boxes and lines and when I program, I spend a lot of times scribbling on paper to map relationships between data, to show the data flow between modules or concepts, to show the pipeline stages, that kind of thing. I do it mainly as a way to organise or solidify my thoughts, less to document them. When I used Max/MSP some years ago, I had the interesting experience that I found I didn't need to do this on paper because doing it in Max code was enough. I've wished for a Max-like language more geared towards general purpose programming[1] ever since, but alas, I've yet to find one. [1] Richer data structures (I think it may have got these since actually), tools for unit testing, stuff like that.
amatic 6y ago
> Is there a good example of visual programming making difficult concepts easy? Analog computers with patches and plugs, ancestors of Max/Msp and similar software, were successful because they familiarized users with differential equations much more intimately than other tools, like symbol manipulation. Also they were often more powerful, as analytic methods were limited to, say solving linear equations. After the fall of electric analog computers and rise of digitals, there were simulators of analog computers, but in text form. So you would trick the digital computer to perform continuous calculations, but you still had to write code. This code would represent a graph where nodes were analog functions, and edges were 1 dimensional values. (tensorflow is somewhat similar). In regular modern programming, you don't usually try to make a graph of value transformations, so fitting different concepts to visual space does not make them more understandable. That's my guess for why visual programming did not succeed in wide areas. If there was anything that would fit visual programming, my bet is analog computing and control theory, as they were traditionally very connected. I've been working on a little toy graph editor for a while, event handling gets complicated in javascript very quickly, and there is svg issues, etc.
fossuser 6y ago
Instead of having to compute the code in your head and stepping through the changes mentally it’d be easier to see it visually. This doesn’t mean replacing the code since text is an efficient way to write. I’d want the code to generate a real time representation of its flow. Imagine writing a graph and running Dijkstra’s algorithm and seeing the nodes appear as you write them, seeing edges appear and seeing the traversal next to the code. We do this anyway in our heads or poorly with debuggers. We do this on whiteboards when trying to understand the problem. If this was built into our tools we could offload the computing of our code itself to the computer and get immediate feedback. It’d be a lot easier to reason about. Tree structures, recursion, etc. In the end the intuitive understanding of code good programmers have would come, but it could come a lot faster and easier. As a bonus you could see non obvious patterns you might otherwise miss.

tiborsaas 6y ago

A good example would be Max/Msp. It's used by a lot of musicians creative coders. Visual representation abstracts the boring pars of programming and enables experimentation much more rapidly. Quickly rearranging nodes is more intuitive than changing how you pass around objects.

slightwinder 6y ago

Showing the "big picture", the data-flows, the relations, etc. are all prime-candidates where visual presentation can sell better than text. A well made diagramm can be more informativ than any amount of text. On top, having your code in a structued source, you can easily manpulate the structure itself and get alternatice views. Ther eis literate programming which kinda goes into this direction, there you detache the sourcecode-organisation from it's technical organisation. Thouhg, this of course is not directly linked to "visual" programming, but work on the same conditions.

dkersten 6y ago

For me, personally, I think in boxes and lines and when I program, I spend a lot of times scribbling on paper to map relationships between data, to show the data flow between modules or concepts, to show the pipeline stages, that kind of thing. I do it mainly as a way to organise or solidify my thoughts, less to document them. When I used Max/MSP some years ago, I had the interesting experience that I found I didn't need to do this on paper because doing it in Max code was enough. I've wished for a Max-like language more geared towards general purpose programming[1] ever since, but alas, I've yet to find one. [1] Richer data structures (I think it may have got these since actually), tools for unit testing, stuff like that.

amatic 6y ago

> Is there a good example of visual programming making difficult concepts easy? Analog computers with patches and plugs, ancestors of Max/Msp and similar software, were successful because they familiarized users with differential equations much more intimately than other tools, like symbol manipulation. Also they were often more powerful, as analytic methods were limited to, say solving linear equations. After the fall of electric analog computers and rise of digitals, there were simulators of analog computers, but in text form. So you would trick the digital computer to perform continuous calculations, but you still had to write code. This code would represent a graph where nodes were analog functions, and edges were 1 dimensional values. (tensorflow is somewhat similar). In regular modern programming, you don't usually try to make a graph of value transformations, so fitting different concepts to visual space does not make them more understandable. That's my guess for why visual programming did not succeed in wide areas. If there was anything that would fit visual programming, my bet is analog computing and control theory, as they were traditionally very connected. I've been working on a little toy graph editor for a while, event handling gets complicated in javascript very quickly, and there is svg issues, etc.

fossuser 6y ago

Instead of having to compute the code in your head and stepping through the changes mentally it’d be easier to see it visually. This doesn’t mean replacing the code since text is an efficient way to write. I’d want the code to generate a real time representation of its flow. Imagine writing a graph and running Dijkstra’s algorithm and seeing the nodes appear as you write them, seeing edges appear and seeing the traversal next to the code. We do this anyway in our heads or poorly with debuggers. We do this on whiteboards when trying to understand the problem. If this was built into our tools we could offload the computing of our code itself to the computer and get immediate feedback. It’d be a lot easier to reason about. Tree structures, recursion, etc. In the end the intuitive understanding of code good programmers have would come, but it could come a lot faster and easier. As a bonus you could see non obvious patterns you might otherwise miss.