If you’re like me, you may have read about the usability problems with the Diebold touchscreen voting machines in the States. Before I saw this video, I didn’t realise quite how serious the problems are. It’s not a question of a subtle usability issue or two – as the video demonstrates (at 1m 15s), even when the touchscreens have just been calibrated, they are innacurate, making it very easy to select the wrong candidate without realising.
It’s been argued that the problems stemmed from a lack of budget for usabity testing. This is nonsense – ‘expensive’ usability testing isn’t needed to uncover this problem. A simple QA process would have shown immediately that the device isn’t fit for purpose.
I can’t understand why didn’t they just use hard buttons rather than touchscreens. It would have been cheaper and would have avoided the need for calibration entirely… The mind boggles.
Touchscreen simplifies development significantly. Suddenly it’s supposedly just like having a mouse with only one button. Your developers can use their own computers for development and testing without need for special emulators or hardware. The same goes for QA. The touch screens were likely sourced separately and installed into the machine just like a regular mouse would – with minimal testing because “it should just work”.
ATMs, vending machines (redbox), visitor info boxes, etc – everyone is moving to touch screen and it sucks. Like you mentioned, accuracy frequently degrades significantly and at best depends on the magical angle and way you touch the screen.
Of course, if they’d gone physical buttons we’d have the age old ATM problem of buttons not lining up with options so you’re not really sure which ones apply…
You’re right that the calibration problem is in large part a QA issue.
It’s also a technology issue: It is obviously possible to make touch screens that don’t require calibration.
It’s a safety issue, too: The touchscreens immediately convert your vote to bits, whereupon it can be altered by erroneous or malicious software.
It’s a confidence issue, too. See above.
But it’s also a usability issue, to me. Touchscreens are prone to error from inexperienced users, people with physical problems. Confirming your vote, changing your mind mid-process, requires navigating a UI. Only so much info can fit on a screen at once, leading to poor information design. The best solution is a paper ballot that you fill in with a pen or pencil, and that is optically scanned for an initial count, and ideally hand counted later for certainty or to settle contested races.
Everyone knows how to fill out a piece of paper. Mistakes are easy to spot and correct or have the ballot replaced.
It’s also better on all the other fronts. It provides confidence. It’s easy to update (print new ballots). It provides a permanent record, so even if the counting machines are hacked or broken the ballots themselves still exist and can be counted again, even without machines.
There’s not a single way in which touchscreens are superior to paper except (perhaps) for people who have special needs.
Pingback: The Weekly a-List | Artefact | Blog
Pingback: Intelligent Experience Design
I’ve been both UI designer and programmer on a handful of touchscreen products (Get it? Handful? Touchscreen?) and there’s many, many design issues going wrong in this video.
Confirmation: They use a a green checkmark to show confirmation but don’t show the ENTIRE BUTTON changing. This is an issue. We’ve noticed in UI studies that touchscreen users focus more on the dead-center of their fingers and surrounding area when interacting than they will with a mouse pointer. Better to have the entire button change color.
Self-detection of miscalibration: On systems where you’re worried about calibration issues, it’s easy to design a self-detection screen. At the start of the pattern present a single “Touch here to begin” screen with no other options. Make the hitbox small. It’s easy to right code to detect mis-clicks on this screen (touches falling outside of the only button on the screen) and automatically shut the machine down for inspection.
Confirmation: On systems where a correct choice is critical (like voting machine, but also on a system where a user is spending money, like a jukebox), a confirmation option can greatly reduce mis-clicks. It doesn’t have to be a separate screen (although it can be), I’ve used “OK” or “confirm” style buttons — but they MUST confirm the choice by displaying the text of the choice they made.
Ugh. There’s lots of other things I could get into too, like overlap (when the user’s hand covers a part of the screen with important information) ….
Hanford – these are really good points that I haven’t read elsewhere, you should write a post on this! Also I’d be interested to know what you think of Pentagram’s redesign of the Wells Fargo ATM (Posted back in June ’08).
Pingback: The Weekly A-List | Artefact