6 ms·

This is probably my all-time favorite article on the internet. I've brought it up in conversation dozens of times when referring to ways to control computers. I

This is probably my all-time favorite article on the internet. I've brought it up in conversation dozens of times when referring to ways to control computers. It's why I don't really use phones for anything beyond answering phone calls and handling urgent tasks when I'm not at home/office. I just feel like I have very little control over these devices, unlike the massive control with a keyboard or mouse, or even better, a fork and spoon when eating dinner (although despite its high degree of control, you can't control a computer with them). Tech innovators and its users/consumers need to think outside of the box more often for ways to control computers. Given a physical UI device, you can place it on a spectrum of "computer-friendly to human-friendly". Touch screens are computer-friendly because it's trivial for computers to understand and developers to interact with. The downside is that users feel distanced from their control, so it is difficult to use efficiently. But developers are smarter than users at UI---the difficulty should be put on them instead, to control their software/OS. I've imagined things like stress balls with "electric muscles" for feedback to control various applications, a cube with the ability to change its texture with suction/servos, and things like https://roli.com/products/seaboard https://roli.com/products/seaboard or https://www.expressivee.com/buy-touche https://www.expressivee.com/buy-touche except for general computing applications rather than music performance. Who knows what could be invented if customers could be convinced their computing experience can be improved with better physical interfaces.

phaedrus 7y ago
As an engineer I finally got to take the training course our techs go through and experience how a tech uses the software which we (engineers) developed and maintain. It was eye-opening. One particular example relevant to this discussion is the old version of the equipment had physical potentiometers which the tech would adjust to get a desired meter reading. The new equipment has you just type a number into a textbox in a Windows program. BUT - it doesn't even provide a slider or up/down arrows, you have to play The Price Is Right inputting guessed over/under numbers until a measured output in a second text box matches a desired value. It's clear that when the transition was made from physical controls to software based controls, not enough thought was given to ergonomics and how it would be used by an actual human.
- Misdicorl 7y ago
  Maybe a decent place to implement a computerized linear search algo for your techs? Sounds like they're setting params for a PID/control loop in the first place and second order PID/control loops are notoriously finicky so I'd avoid that. But a simple/static linear search for params where it just binary searches while the tech chooses the best value amongst the 3 options [min, mid, max] probably gets them there pretty fast. Though probably still not faster than a manual dial.
cortesoft 7y ago
The fact that millions of people do a majority of their computing on their phone contradicts some of your argument. I also think it highlights one of the things this article misses; that one of the important design decisions is amount of effort the action takes. Any interface that requires full body movement is going to be too draining to use all day. It is a lot easier to touch a screen than move your whole body to manipulate a thing, even if that would get you more fidelity. For all the downsides of a touchscreen phone, people are able to use them one handed for hours on end.
- vortico 7y ago
  >Any interface that requires full body movement is going to be too draining to use all day I didn't say the ideal solution should take more effort than a touch screen. In fact, it will take less. Touching and dragging glass takes enormous effort compared to holding a stress ball and twitching your fingers in certain ways to control a computer. The ideal physical interface converts muscle power to "information control" with maximum efficiency. >people are able to use them one handed for hours on end Yes, to produce work they could have done in 10 minutes with a keyboard/mouse.
  - cortesoft 7y ago
    Most people aren't 'producing work' when they are on their phone... they are consuming work.
    - vortico 7y ago
      In that case, I'm not interested in that use case. I'm interested in increasing efficiency and control of "producing work" on a computing device.
      - cortesoft 7y ago
        Most of the time spent with computer interfaces is consuming rather than producing, though... doesn't it make sense to optimize for that use case for general purpose electronics?
pfranz 7y ago
I think one part is finding the right domain for innovative UI. I often see a weird UI on a tool I use once a year for a couple minutes. I would very much prefer it use the OS default, clunky UI everything else has even if it's less efficient. Going off the trodden path usually means things like shortcut-keys or scripting breaks. We give a lot more latitude for "professional" apps (where we spend hours every day using). Often, those UIs are built up historically. I was watching an iPad thinking how much of a mess it would be to show them all of this on day 1. A car infotainment system would be an interesting place to innovate. I don't know if it's because car companies have squandered goodwill or if I rent/borrow too many cars, but now I just want to use my phone to plug in and prefer knobs over touch screens.
- vortico 7y ago
  >car infotainment system That's a great area for improvement. Controlling a car absolutely needs a good physical controller that is very close (conceptually) to our human muscle structure. But what do car companies use for their new systems? Touch screens, the absolute worst choice for keeping your eyes on the road.
  - atoav 7y ago
    They want to sell cars. And shiny smudgescreens are still something fancy that sells.
    - TeMPOraL 7y ago
      It also streamlines their work. Physical controls interfere with the design of the car, making changes and iterations more expensive. With touchscreens, the design process boils down to placing a screen somewhere, and they can pawn off UI design and software to another department or to a subcontractor.
atoav 7y ago
As a fan/user/creator of modular synthesizer stuff and design observer in general I can usure you that you are quite definitly not the only person that feels like this. Vinyl record sales have surpased CD for the first time again not long ago, people crave for physical interfaces, modular synths and everything hardware is getting more popular despite the ton of plugins you can get for your music software. People are going back and build physical interfaces for menu-diving-synthesizers from the 80s like Yamaha’s infamously unprogrammable but amazing sounding DX7. People are rediscovering mechanical keyboards etc. Modular synthesizers are a wonderful playground for physical interface design. Because of course you want to expose a lot of functionality of your module, but on the other hand your panel will get big and your module will end up taking up a lot of space. Of course you can get away with using a screen and an encoder but hoe wrll can you use it live while making muic then? Of course you can add 20 dials, but maybe 3 clever ones that abstract over those 10 in just the right ways are better? Also: people might not use your module for a month, how can the design remain in their head just enough to kick a spark? Physical interfaces. Use them blindly, shile wearing gloves, in your pocket.
- vortico 7y ago
  A have a few thousand HP of Eurorack and definitely agree that modular synths are easier, more fun, and more interactive to control than most other interfaces. But even then, they're clunky for some things, and I feel that there are further designs that can achieve control even more efficiently.
  - atoav 7y ago
    I agree. There is even more possible and not all modules are really well designed (e.g. why did Dieter Doepfer feel the need to swap around the otherwise identical inputs between his SEM- and Wasp-Filter?). Other modules just don’t go far enough. many lack attunoators/attenuverters (for economical and space reasons). You have to really fight to get a decent physical interface, to get a good one takes even more. I believe the future of modular interface design lies especially in more visual feedback which makes usage easier and more intuitive and systems more reliable during live usage. There we can learn from modular and semimodular synths from the software side.
azhenley 7y ago
If you enjoyed reading this then I would definitely recommend The Design of Everyday Things (https://en.wikipedia.org/wiki/The_Design_of_Everyday_Things https://en.wikipedia.org/wiki/The_Design_of_Everyday_Things).
ryan93 7y ago
Why do you need more control when scrolling on safari or twitter.

phaedrus 7y ago

As an engineer I finally got to take the training course our techs go through and experience how a tech uses the software which we (engineers) developed and maintain. It was eye-opening. One particular example relevant to this discussion is the old version of the equipment had physical potentiometers which the tech would adjust to get a desired meter reading. The new equipment has you just type a number into a textbox in a Windows program. BUT - it doesn't even provide a slider or up/down arrows, you have to play The Price Is Right inputting guessed over/under numbers until a measured output in a second text box matches a desired value. It's clear that when the transition was made from physical controls to software based controls, not enough thought was given to ergonomics and how it would be used by an actual human.

Misdicorl 7y ago
Maybe a decent place to implement a computerized linear search algo for your techs? Sounds like they're setting params for a PID/control loop in the first place and second order PID/control loops are notoriously finicky so I'd avoid that. But a simple/static linear search for params where it just binary searches while the tech chooses the best value amongst the 3 options [min, mid, max] probably gets them there pretty fast. Though probably still not faster than a manual dial.

cortesoft 7y ago

The fact that millions of people do a majority of their computing on their phone contradicts some of your argument. I also think it highlights one of the things this article misses; that one of the important design decisions is amount of effort the action takes. Any interface that requires full body movement is going to be too draining to use all day. It is a lot easier to touch a screen than move your whole body to manipulate a thing, even if that would get you more fidelity. For all the downsides of a touchscreen phone, people are able to use them one handed for hours on end.

vortico 7y ago
>Any interface that requires full body movement is going to be too draining to use all day I didn't say the ideal solution should take more effort than a touch screen. In fact, it will take less. Touching and dragging glass takes enormous effort compared to holding a stress ball and twitching your fingers in certain ways to control a computer. The ideal physical interface converts muscle power to "information control" with maximum efficiency. >people are able to use them one handed for hours on end Yes, to produce work they could have done in 10 minutes with a keyboard/mouse.
- cortesoft 7y ago
  Most people aren't 'producing work' when they are on their phone... they are consuming work.
  - vortico 7y ago
    In that case, I'm not interested in that use case. I'm interested in increasing efficiency and control of "producing work" on a computing device.
    - cortesoft 7y ago
      Most of the time spent with computer interfaces is consuming rather than producing, though... doesn't it make sense to optimize for that use case for general purpose electronics?

pfranz 7y ago

I think one part is finding the right domain for innovative UI. I often see a weird UI on a tool I use once a year for a couple minutes. I would very much prefer it use the OS default, clunky UI everything else has even if it's less efficient. Going off the trodden path usually means things like shortcut-keys or scripting breaks. We give a lot more latitude for "professional" apps (where we spend hours every day using). Often, those UIs are built up historically. I was watching an iPad thinking how much of a mess it would be to show them all of this on day 1. A car infotainment system would be an interesting place to innovate. I don't know if it's because car companies have squandered goodwill or if I rent/borrow too many cars, but now I just want to use my phone to plug in and prefer knobs over touch screens.

vortico 7y ago
>car infotainment system That's a great area for improvement. Controlling a car absolutely needs a good physical controller that is very close (conceptually) to our human muscle structure. But what do car companies use for their new systems? Touch screens, the absolute worst choice for keeping your eyes on the road.
- atoav 7y ago
  They want to sell cars. And shiny smudgescreens are still something fancy that sells.
  - TeMPOraL 7y ago
    It also streamlines their work. Physical controls interfere with the design of the car, making changes and iterations more expensive. With touchscreens, the design process boils down to placing a screen somewhere, and they can pawn off UI design and software to another department or to a subcontractor.

atoav 7y ago

As a fan/user/creator of modular synthesizer stuff and design observer in general I can usure you that you are quite definitly not the only person that feels like this. Vinyl record sales have surpased CD for the first time again not long ago, people crave for physical interfaces, modular synths and everything hardware is getting more popular despite the ton of plugins you can get for your music software. People are going back and build physical interfaces for menu-diving-synthesizers from the 80s like Yamaha’s infamously unprogrammable but amazing sounding DX7. People are rediscovering mechanical keyboards etc. Modular synthesizers are a wonderful playground for physical interface design. Because of course you want to expose a lot of functionality of your module, but on the other hand your panel will get big and your module will end up taking up a lot of space. Of course you can get away with using a screen and an encoder but hoe wrll can you use it live while making muic then? Of course you can add 20 dials, but maybe 3 clever ones that abstract over those 10 in just the right ways are better? Also: people might not use your module for a month, how can the design remain in their head just enough to kick a spark? Physical interfaces. Use them blindly, shile wearing gloves, in your pocket.

vortico 7y ago
A have a few thousand HP of Eurorack and definitely agree that modular synths are easier, more fun, and more interactive to control than most other interfaces. But even then, they're clunky for some things, and I feel that there are further designs that can achieve control even more efficiently.
- atoav 7y ago
  I agree. There is even more possible and not all modules are really well designed (e.g. why did Dieter Doepfer feel the need to swap around the otherwise identical inputs between his SEM- and Wasp-Filter?). Other modules just don’t go far enough. many lack attunoators/attenuverters (for economical and space reasons). You have to really fight to get a decent physical interface, to get a good one takes even more. I believe the future of modular interface design lies especially in more visual feedback which makes usage easier and more intuitive and systems more reliable during live usage. There we can learn from modular and semimodular synths from the software side.

azhenley 7y ago

If you enjoyed reading this then I would definitely recommend The Design of Everyday Things (https://en.wikipedia.org/wiki/The_Design_of_Everyday_Things https://en.wikipedia.org/wiki/The_Design_of_Everyday_Things).

ryan93 7y ago

Why do you need more control when scrolling on safari or twitter.