Pocketables have posted a nice detailed piece about DialKeys and the TIP (Touch/Tablet PC input panel.) There are some nice tips in there but what caught my eye were Jenn’s typing speed co-efficient’s. I know it sounds rather boring but its important. Pay attention please because when I list all the figures in order, it makes interesting reading. To some people. Maybe.
Here are all the typing speed co-efficients that I know about at the moment. Its not many but its a start. What I’d like you all to do is to submit your own if you have time. I’ll record the data and create a big list of UMPC-related input mechanisms which will allow potential customers to understand the limits of each input type. I think that’s going to be something useful, interesting and a good reference for keyboard designers, software and hardware alike.
Here’s what we have already from fastest to slowest.
- Full size desktop keyboard – CE 1
- Eleksen Fabric Keyboard – 0.6 (chippy)
- Kohjinsha SA1 ultra mobile PC (keyboard) – 0.55 (chippy)
- PepperPad 3 (split thumbboard) 0.5 (chippy) 0.54 (jenn)
- Medion & Gigabyte 0.42 (chippy)
- Sony UX (keyboard) – 0.37 (jenn)
- OQO Model 01 (keyboard) – 0.34 (jenn)
- i7210 + On screen keyboard (TIP) – CE 0.3 (chippy)
- DialKeys on Samsung Q1 – 0.27 (jenn)
Its certainly not conclusive yet so please contribute. Here’s the easy test method I used.
- Take your normal desktop keyboard and using your fave editor, time yourself typing “The quick brown fox jumped over the lazy dog” 5 times. Its 220 characters long (not including CR/LF) (220/secs*60) = characters per minute (CPS)
- Do exactly the same with a different input method. Calculate CPS.
Result 2 / Result 1 = speed co-efficient of the alternative input method.
Other methods will work as long as you do the same for test 1 and test 2.
Repeat with another device (to fade!) and publish the results or times on your blog. Place the link in the comments for that threaded SEO effect that we all like so so much! I’ll also place your link in the results table when I create the page/database table of results. Alternatively, post the results directly on the comments or via the contact form.
Please state whether you have trained or not on the alternative input method.
What I’d like to see are Blackberries, PDA’s, Speech input and handwriting recognition (that one will vary wildly!) and any other method you can think off. When I have enough results I’ll publish a nice table and graphic.
Get typing, tapping and talking!
Here’s a mobilitybeat and digg link as this post is probably worth promoting to get as much feedback as possible.
You got almost the same result I have been getting. The OQO and Sony keyboard gave you almost the same speed that if you use the TIP. People keep saying that integrated keyboard on those devices are faster when they are really not. They are just about the same.
I’d suggest not using “the quick brown fox…” as your test text. That sentence uses all the letters of the alphabet, which isn’t typical of most English text, where Es and Ts dominate. This is important because some text entry methods put all the commonly used letters in easy-to-reach places, which increases real-world text entry speeds, but doesn’t provide an advantage for “the quick brown fox…” sentence.
Feel free to use whatever you want. With enough responses we’ll get a good average.
Regards
Steve
peejay,
“The quick brown fox jumped over the lazy dog” is dominated by E’s, but is generally more flattened than a typical english distribution. It’s not a perfect distribution, but it’s not horribly off.
Regarding input speed of real/virtual keyboard there are a few determining factors in my experience:
1. Closeness of layout to normal keyboard
2. Whether the input method allows for fingers other than one’s thumbs
The first is important because we are so used to that particular layout that any (subtle even) differences would slow us considerably–Dialkeys and InScribe are examples. The second is true at least for myself–typing with my pointing fingers is for example already much faster than my thumbs, not to mention if I can use multiple fingers.
The above say nothing about non-keyboard input like hand-writing recognition of course. In addition, if you add in numbers and symbols in this test the difference would be a lot more dramatic even–Dialkeys for example is close to useless in such cases.
Hi Ticonderoga.
Good point about the numbers and symbols.
Steve
Hello,
I guess the test will also depend on how fast you can type on a normal keyboard to begin with. I’m not a touch typist – and I got an average of 1519 cpm on my laptop keyboard (full size). With the detachable thumb keyboard on my aging iPAQ 2210, I pounded out 970 cpm, giving a coefficient of 0.64.
so maybe people who don’t touch-type won’t experience as much of a performance hit from using these alternative input methods, since maybe part of the reason they are slower than using a keyboard is because you still have to look at the keys while you type, AS WELL AS using less digits.
Just my two cents – I think its really interesting that you’ve decided to test this.
Also, just tried the ‘transcriber’ handwriting recognition software on the pda: 665cpm = 0.43. This is to be expected since I already type faster than I write. If I can find the driver cd for my wacom tablet, I’ll test out how fast the handwriting recognition is on the Vista version of the TIP.
Cheers.
There are a few variable in the ‘equation’ of course but because its only a relative speed test, it shouldnt matter how fast you type. The speed percentage should be similar for most people.
S
Would be nice to see speed on a wacom tablet. Thanks Pat.
using “The quick brown fox jumped over the lazy dog” 5 times I’m seeing these numbers
30s baseline (MS Comfort curver)
35s Kohjinsha SA1F00V 0.85
note that since the keyboard is japanese typing symbols (!,”,’,etc)would have considerabley slowed me down, but numbers and letters are fine.
DSD