For some time I’ve been meaning to test my small collection of PDA/smartphone gadgets to see which of their methods of input was quickest. The iPhone’s software keyboard? The Newton’s handwriting recognition? Palm’s Graffiti? With the possible imminent arrival of a tablet from Apple that will save the world, it seemed a good time to get round to the test.
I have six input methods to compare:
- The Apple Newton MessagePad 2100’s handwriting recognition (1997)
- The Palm Vx’s [Graffiti](http://en.wikipedia.org/wiki/Graffiti_(Palm_OS)) (1999)
- The Palm Treo 650’s hardware QWERTY keyboard (2004)
- The Apple iPhone 3G’s software QWERTY keyboard (2009)
Although an exhaustive test could have included other devices, these four seem to cover the main types of input. For comparison’s sake I also tested:
- Pen and paper
- A full-size QWERTY keyboard
To test the speed of input I was going to use the same piece of text for each one. I also wanted to use some text I could memorise, so I didn’t have to pause typing/writing to look up, and the later tests wouldn’t benefit from my increasing knowledge of the text.
So I used a piece of text I memorised a couple of years ago as it didn’t take much effort to retrieve it from my memory and become fluent again. This is the final paragraph of a speech given by Aneurin Bevan to the British Labour Party in 1959, following their general election defeat:
I have enough faith in my fellow creatures in Great Britain to believe that when they have got over the delirium of the television, when they realize that their new homes that they have been put into are mortgaged to the hilt, when they realize that the moneylender has been elevated to the highest position in the land, when they realize that the refinements for which they should look are not there, that it is a vulgar society of which no decent person could be proud, when they realize all those things, when the years go by and they see the challenge of modern society not being met by the Tories who can consolidate their political powers only on the basis of national mediocrity, who are unable to exploit the resources of their scientists because they are prevented by the greed of their capitalism from doing so, when they realize that the flower of our youth goes abroad today because they are not being given opportunities of using their skill and their knowledge properly at home, when they realize that all the tides of history are flowing in our direction, that we are not beaten, that we represent the future: then, when we say it and mean it, then we shall lead our people to where they deserve to be led!
Armed with this not-entirely-appropriate 221-word passage, I set to work in chronological order.
Pen and paper
I decided to time my fast handwriting, rather than my neatest; so long as I could read it easily, that was good enough. For those who care about such details, I used a Lamy Safari pen filled with Noodler’s ink on some cheap and unlikeable Ryman’s brand A4 lined paper.
It took me 5 min 33 sec to write the passage. I could probably have gone faster at the expense of legibility, but it seems a good balance.
Full-size QWERTY keyboard
For the past few months the only keyboard I’ve been using is my MacBook’s, so I used that for the test. I can touch-type, although I tend to make too many mistakes. As with all the tests, I corrected mistakes as I went, which was much easier and quicker on a computer than on some of the more fiddly devices.
Typing the passage out took 3 min 14 sec.
Apple Newton MessagePad 2100’s handwriting recognition
From here on I performed each test twice and used the best time. I figured that despite practising a little before each test, and being surprised how quickly the knack always came back to me, I’d need more time to get into the flow of each input method.
Contrary to its reputation I didn’t have many problems with the Newton’s handwriting recognition. Just as when I used it (and the earlier 120 model) regularly I relied on the printed character recognition, rather than the cursive; my handwriting is only partly joined-up anyway, so this wasn’t much of a problem and, as John Gruber pointed out recently, the printed recognition is much more reliable.
For those unfamiliar, you write with a stylus anywhere on the MessagePad’s large screen and your handwriting appears there as you write. A moment after you finish a word it disappears and reappears as “typed” text wherever the cursor currently is.
This method was easier than I expected and it felt like it was as quick as handwriting recognition could be — there is some delay in translating my scribble into text, but it’s no slower than the speed of my writing, so the device didn’t get behind. Once I got into it, correcting mistakes was also easy: tapping a word brings up alternative spellings and capitalisation, and one further tap away are two ways to manually correct the characters.
The first test took 11 min 30 sec and the second, with fewer mistakes and quicker corrections, took 9 min 17 sec. This is about twice as long as my standard handwriting test, mainly because of the need to write neatly enough for the MessagePad to recognise my writing.
Palm Vx’s Graffiti
[Graffiti](http://en.wikipedia.org/wiki/Graffiti_(Palm_OS)) is a simplified version of handwriting input. There is a small part of the screen dedicated to accepting input, and you write each specially-stylised letterform in the same place. Unlike the Newton, you don’t see the marks you make with the stylus.
In theory this simpler method is less error-prone than having to interpret normal handwriting, although I remember that even when I was using the Vx every day I found it at least as frustrating. Small mistakes mean letters are mis-interpreted and, then and now, I was forever having my ‘Y’s appear as ‘X’s.
The first time through it took 13 min 24 sec to tap the passage out, improving slightly to 12 min 16 sec on the second pass. Although I could have improved the accuracy a little, it felt like there was a speed limit that wasn’t present with the MessagePad — there’s less “flow” in the input and it feels like a more regular and mechanical means of input, o n e l e t t e r a t a t i m e.
Palm Treo 650’s hardware QWERTY keyboard
I was interested to see how the Treo and the iPhone would compare, given how many people dismiss the iPhone’s software keyboard in favour of a “real” keyboard something like the Treo’s.
I had pleasant memories of using the Treo’s keyboard (and the device as a whole) and after a little practise I wasn’t disappointed. I was using the KeyCaps600 extension which makes a few things easier, although I’m not sure it was much of a factor in this test.
I ended up doing this test three times. The first couple I had times of 6 min 18 sec and 6 min 8 sec but it felt like my thumbnails were getting in the way. They weren’t long but were long enough to make me uncertain whether to type with the nails or the pads of my thumbs. So I cut them shorter and tried again, using only my thumbs themselves: 5 min 23 sec.
Apple iPhone 3G’s software QWERTY keyboard
Finally, the loved and loathed software keyboard. I remember being frustrated with it when I first purchased my iPhone 18 months ago, but after a couple of weeks, and after finding some degree of peace with its auto-correction, I started to like it. I still had no idea how it would compare to the other methods, particularly the Treo’s.
I used the iPhone in the vertical position. I expect the larger horizontal keyboard might be better in general, but I never use it as I get frustrated with the tiny area remaining to view what you’re typing. It’s like using one of those old 1980s “laptops” with a tiny LCD strip display. So I stuck with the orientation with which I’m most familiar.
The first pass at the text took me 6 min 1 sec to tap out, but I must have got into the flow as the second took only 4 min 56 sec.
Conclusion
If it’s not already clear, this isn’t a particularly scientific experiment. If you performed the same tests you’d probably come up with different results depending on your proficiency with the various devices. But here are my results, shorter bars are quicker and better:
I’m surprised by some of this. I thought that typing on a full-size keyboard would be a lot quicker than the iPhone and Treo than it is, although a real touch-typist could have beaten my 68 words per minute.
I’m surprised at how slow Graffiti was for me, particularly when its simplicity was supposed to be a huge benefit over the Newton MessagePad’s ridiculed handwriting recognition.
I had no idea how the iPhone and the Treo’s keyboard would compare but they’re similar; I expect someone who uses a comparable miniature hardware keyboard frequently could beat my iPhone time but I suspect neither is radically, objectively quicker than the other.
Comments
Commenting is disabled on posts once they’re 30 days old.
robinson at 18 Jan 2010, 10:46pm. Permalink
Incredibly wonderful test and write-up! You put in a lot of time, and wrote up the results well... Given that I have used most of these input methods for years (except for the thumb keyboard and Newton), I could relate to what you were describing--you captured well what was involved.
It also made a good historic connection with different eras of computing and entry.
Hmm... I wonder how Dragon's Naturally Speaking voice system would have compared!
Oh, an aside... love the quote you chose. How apropos for today's world, yet how dismaying that Bevan's commentary still applies!
Ian Goss at 19 Jan 2010, 8:20am. Permalink
Interesting post.
How well-trained is your MP2100?
Mine took ages to learn my writing but got very good eventually.
I find I could write more quickly on the Newton than with pen and paper (with recognition turned off) but then of course needed to do the old double-click to turn the scrawl into text.
The past tense is used because the battery finally crapped out last year (not bad for something from 1997).
If a new tablet had trainable scribble recognition, I'd get one in a heartbeat. But since the Knowledge Navigator idea came from Sculley, that may not happen.
Phil Gyford at 19 Jan 2010, 10:06am. Permalink
Ian, it's been years since I used it, but I think the MP2100 was pretty well trained, as I used to use it a lot and it still contained all my old data. I never really used the "scribble it now, recognise later" input method -- I only used the one where it tries to recognise it immediately.
Jon at 20 Jan 2010, 3:25pm. Permalink
I don't know your writing style on the Newton, but I could write as fast as, if not faster on my Newton than with pen and paper.
The key was to not wait for it to finish recognizing, but to write in any and every free space where you could fit a word. On my 130 and 2100, at least, I seldom had problems with words going in the wrong place, and I was typically writing outlines.
I have clocked myself at about 35 wpm for normal writing. I would correct as I went, but since the errors showed up long after I had written them, ;-) I typically waited until a lull in the writing to make corrections, or even do a review afterwards. Another trick is to "squish" the settings, writing your letters and words close together, and lower the wait before recognizing. Once you learn to make sure you write very compact (a trick I learned writing on paper in college). I haven't written hardly a sentence at a time since 2005, but I bet it wouldn't take me two minutes to get my speed back. That's how natural the Newton felt to me. I used mine for 10 years, though. ;-) Like you, I've transitioned to Graffiti, then a Treo 650 and now an iPhone. I'm shocked at your iPhone speeds. I don't think I could touch that. I miss my Newt. sniff. ;-)
(It's actually sitting here behind me, plugged in, because the battery is dead, but I only use it for conversions nowadays. Since this past Jan 5, its dates are messed up due to the end of the world as the Newton knows it--aka 2010 bug--I never bothered installing the fix/hack)
-Jon
Dr. Tablet at 21 Jan 2010, 8:16pm. Permalink
Great idea for a test!
I'm curious w/regards to your Newton test: have you trained the Newton much? Or were you starting from a factory-reset state? I have an old MessagePad 2100 here, and I'm going to try your test now to see how it does (from a factory reset condition). Will post my timings shortly.
T.S. at 21 Jan 2010, 8:31pm. Permalink
How about adding Dragon Naturally Speaking and MacSpeech Dictate to this test?
Max Pinton at 21 Jan 2010, 8:33pm. Permalink
Just for the sheer nerdity of it, I'd love to see someone write this out in Gregg Shorthand. I'd try it, but I'm not anywhere close to competent. It's probably not a fair contest since it's phonetic, but I bet it'd win hands down.
As an aside, a couple years ago someone wrote a METAFONT for Gregg. It's not perfect, but it's amazing anyway:
www.tug.org/TUGboat/Ar…
You can try it out here:
www3.rz.tu-clausthal.d…
Jason at 21 Jan 2010, 8:37pm. Permalink
How the hell do you type on a iphone in vertical format so quickly? I've had every version of the iphone and I can never type even 1 full sentence without a typo that the dictionary doesn't understand. Do you have little girl fingers?
Dr. Tablet at 21 Jan 2010, 8:38pm. Permalink
Just finished the test. I'm not familiar with the passage (Canadian, eh?) so I expected it would take me a little longer: 13 mins 5 seconds.
With a factory-reset Newton and printed characters, I found the recognition to be very good. Most of the mistakes that I had to correct were my own, from trying to write as quickly as possible. On a piece of paper, I would have simply scribbled them out and rewritten them. On the Newton, I erased/corrected.
No question about it - for quick text entry the keyboard is still king. Still, I'm impressed with the out-of-the-1997-box performance of the MP2100's recognizer. Much better than my original Newton messagepad.
---
www.drtablet.com - Tablet news, info, reviews & opinions
Peter da Silva at 21 Jan 2010, 8:40pm. Permalink
When I was using graffiti regularly for taking notes I could graffiti faster than I could write, and type WAY faster than either. The main advantage of graffiti over handwriting is I can do it without looking at the screen, because I don't have to move the input point... I just enter letters one after each other in the same place. It did take a few months of regular use to get fast, it's not something you can just pick up and use effectively without training, and I'm nowhere near that fast any more.
Elijah Nicolas at 21 Jan 2010, 8:41pm. Permalink
Amazing test and write-up. The time that you dedicated to actually conduct the test but the dedication. The paragraph that you chose was also exemplary. Now, people can move on in their lives when comparing a tactile keyboard vs. the iPhone. That was my initial hesitation to converting as well as the feelings of my fiance when we both migrated from the Treos to the iPhone 3G.
I can't wait till you test out the new tablet. <-- now there's a new idea!
John Lascurettes at 21 Jan 2010, 8:44pm. Permalink
I don't know if you ever experience Graffiti 2 on newer Palm OS devices. Palm had lost a patent lawsuit (filed by who I can't remember) and had to ditch their one-stroke for all characters notation. So certain characters that naturally take more than one stroke but in Grafitti used to take one stroke (t, x, ?, etc.) now took two strokes. Palm attempted to market it as an improvement but it was maddening, making text entry far worse and it didn't even work properly with its own default shortcuts.
Dan Dan The Art Man at 21 Jan 2010, 8:46pm. Permalink
Wow, that was really cool. I'm a writer, and I've always wanted a small device that could fit in my pocket that I could use for writing. I've had a flip phone until last year when I got a new 9800 Blackberry. I was really surprised at how well my thumbs became used to the tiny keyboard, when I first bought it I didn't think I could get very fast on it, and was used to T9 while texting on a flip phone. I now love using my Blackberry, and while a full QWERTY keyboard is a lot faster, I can write pretty quick with my Blackberry, and unlike a small paper notepad in my back pocket, it's already digitized, and I can just email to myself. It was really cool to see the speed on these different devices, thanks for taking the time to write all of this up, great job.
Matt K. at 21 Jan 2010, 8:53pm. Permalink
Interesting test (and nice collection of portable gadgets, by the way).
However, not surprisingly, the results of the speed test seem to be skewed towards the inputs methods with which you are most familiar/fluent.
It would be interesting to see the results when comparing input speed of users employing their most used devices so none of them has home-field advantage... of course, that would a little more difficult to pull off, I imagine (and would never get written up like this one for all to appreciate!).
Sofa at 21 Jan 2010, 9:01pm. Permalink
The write up is well done, everything looks great. However one quibble with your chart, you mentioned you did each test twice and took the better time on the electronic ones. The chart has the longer 13m24s (221/13.4 to get wpm) listed as the bar rather than the faster 12m16s time (221/12.266)
The other bars are correct, great series of tests, I personally use the Dvorak layout on anything I can get on.
en.wikipedia.org/wiki/…
-Sofa
Larry P. at 21 Jan 2010, 9:03pm. Permalink
I've owned various Palm products, from the III to a Treo 650, and so was not surprised with how the iPhone input speed compared to those devices. The iPhone was the first of a class of devices that showed how well a virtual keyboard can work, and I don't think it is too risky to predict that these will eventually dominate the mobile market. However, I think character recognition technology, like Graffiti, might make a comeback if refinements gained from newer technology are applied. Graffiti on the Palm was often very frustrating; but when it worked, it felt like the most natural way to jot down a quick note.
Sam at 21 Jan 2010, 9:05pm. Permalink
Nice writeup. I remember in the 90's doing tests between me on my MessagePad 130 vs. a friend of mine who was used to his Palm with Graffiti. The MessagePad was always faster, even against experienced Graffiti-ers. But Graffiti often felt faster.
I miss the MessagePad, but really like that I can pop my iPhone into my jeans pocket and I'm not sure stylus input would work on its small screen.
Chris Bardell at 21 Jan 2010, 9:09pm. Permalink
Great test & write-up. Well done.
I have a hand-me-down Treo 650 myself, and the tiny QWERTY keyboard is great for sending txt messages.
Ted at 21 Jan 2010, 9:10pm. Permalink
I'd be curious how Swype stacks up: www.swypeinc.com/
Henrik at 21 Jan 2010, 9:12pm. Permalink
Great article! I've actually conducted a usability study in 1999 at the University of Gothenburg, studying correctness (in spelling), readability and speed of different methods of taking notes. Ballpoint pen (two modes, fast and readable), computer keyboard and a Palm V, using Graffiti, Jot (a compteting input method) and the soft keyboard (single touch with stylus). We had 54 subjects, and we read them articles from a tabloid (easy to read, and interesting to listen to :) for them to transcribe. We classed the subjects info three classes: Beginner, User and Expert (also measuring how fast subjects learned). We interviewed the subjects on what they thought about the different types of input, before and after the tests.
The results are essentially identical to yours:
Speed: Keyboard, pen (fast), soft keyboard, pen (readable), Jot and Graffiti (the last two waaaay behind the rest)
Correctness: Pen (readable), Pen (fast), keyboard, soft keyboard, Jot and Graffiti (the last two waaaaaay behind the rest)
Readability: Keyboard, soft keyboard, Jot, Graffiti and pen (readable) and (fast). (Fast pen suck at readability, not even the ones that wrote the text could read it without problems, all other methods fared great (thanks to great fonts on the devices!)
We wasn't supposed to draw conclusions, it was for another class to do, but I think it's pretty obvious that stylus based character input methods suck, keyboard is the way to go. Our suggestion to the professor was to extend the study to Newton-like input (where you write whole words or sentences), T9 in mobile phones and speech to text. I have no idea if they did it though. I still suspect however that a keyboard user/expert will beat everything easily.
When Apple introduced iPhone, I wasn't at all surprised that they didn't include any stylus based input methods or character recognition. It will _always_ be slower, less accurate, harder to use and harder to learn than any type of keyboard, while using magnitudes more processing power.
That said.. I do suspect that they will include it in the iPad.
Kendall Helmstetter Gelner at 21 Jan 2010, 9:33pm. Permalink
I tried the iPhone Dragon app, after a few first tries I was able to get a time of 7:29. That was after having to break up the speech into a number of small segments and waiting for processing, and also correcting a few mistakes made in translation - in a few spots Dragon helpfully offered up the correct alternative word from what it had used (flowing instead of following or something else like that). In a few spots I had to re-type a few words that Dragon got horribly wrong. I was on a WiFi connection at the time, it might be somewhat slower on 3G (though in practice I think that's not much slower).
Also of note is that I was about to give up altogether when I just used the phone to speak into, it was only when I switched to using the headphones with mic that I was able to get good results (as long as I very carefully pronounced each word).
I'm still not convinced even good voice dictation is that much faster than typing, or any faster if you are not naturally a very clear speaker.
Phil Gyford at 21 Jan 2010, 9:42pm. Permalink
Sofa: Well-spotted on the error in the chart. I've updated it now to use the correct timing for Palm Graffiti -- it looks slightly faster now, but is still the slowest.
Everyone, thanks for the kind comments. Yes, the results probably are, inevitably, slightly biased towards what I'm most used to using, although I was surprised how quickly it felt like the older gadgets seemed "normal" to me again, after so long. Much like re-learning that 221-word passage.
Thibaut at 21 Jan 2010, 9:49pm. Permalink
Hello Phil,
great test, thank you for the efforts you put into this. It is very informative to see such test - even if it's not really scientific - as it lays some facts on guesses I had about these input methods. Two weeks ago I posted some thoughts about the use of handwriting on a tablet sized device, and this post comes as a good complement.
How about rating the satisfaction you had using these different methods ? It's more subjective than a chrono, but it's important too. In some context, success isn't linked with performance against time.
Harry Plantinga at 21 Jan 2010, 10:00pm. Permalink
Nice article!
I tried it with Dragon Dictation, and on the first try it took 3 minutes and 30 seconds, including time to proofread the result and fix the (two) errors.
Richard Lawler at 21 Jan 2010, 10:04pm. Permalink
Why no Tablet PC? I think you would find it does pretty well.
Tommaso Lanza at 21 Jan 2010, 10:10pm. Permalink
Is that an handwritten typo on the "pen and paper" photo? It says "[...] be lead", whereas it should have been "led!" ;)
Leo at 21 Jan 2010, 10:11pm. Permalink
Great read.
However, I beat all of the tested methods with my own simple input method: copy & paste. Took 5 seconds to input the test text. ;)
Phil Gyford at 21 Jan 2010, 10:13pm. Permalink
Richard: There's no Tablet PC because I don't own one! It wasn't supposed to be an exhaustive test, or else I'd have practised with a Blackberry, T9, learned shorthand, got some kind of Dragon software... It's just an indication of different techniques.
Phil Gyford at 21 Jan 2010, 10:14pm. Permalink
Tommaso: Yes, I spotted that typo after I took the photo -- if you click through to the photo on Flickr I've belatedly pointed it out :)
Iain Farquhar at 21 Jan 2010, 10:15pm. Permalink
Hi,
liked the review. Did you ever use the FITALY keyboard for the Palm, (look for FITALY.COM). It is a keyboard overlay for the Palm and Pocket PC designed for one finger input with most commonly used letters grouped around the centre of the keyboard. Taps produce lower case letters and slides origninating over the letter produce upper case letters. There is a numeric sidebar with taps for punctuation and slides for numbers.
I haven't tried your test but find the keyboard fast enough for taking notes in meetings. Hoping that an overlay will be produced for iPhone/iPod Touch but all has gone quiet.
Cheers
Iain
Jacob at 21 Jan 2010, 10:28pm. Permalink
I was interested that many people seem to think you are an abnormally fast iPhone typer, but I tried it on my iPhone and was able to finish in 7 minutes 3 seconds - and that's without having it memorized, so I had to stop after each phrase to read the next one to know what to type. I really think that I can type faster on the iPhone than I can on a Blackberry (especially when taking corrections into account - those are much harder on a Blackberry).
Rory O'Kane at 21 Jan 2010, 10:31pm. Permalink
What about the Palm keyboard? I'm talking about the keyboard you can bring up on the Palm where you tap the letters you want with the stylus. How does that compare in your experience to the other methods?
Phil Gyford at 21 Jan 2010, 10:38pm. Permalink
Rory: I didn't try the Palm keyboard mainly because I almost never used it when I was using the Palm full-time, and so it seemed a bit unfair -- I didn't have any residual muscle memory associated with it. It also didn't seem fundamentally different to the iPhone keyboard, except I'd only be able to use one "pointer" rather than two thumbs. Anyway, yes, there will always be more things that could be added.
vgbnd at 21 Jan 2010, 10:45pm. Permalink
retyped the letter on my iphone 3g in landscape mode: 08min03sec
akatsuki at 21 Jan 2010, 10:52pm. Permalink
How many of those words weren't in the iPhone dictionary? Put in some less common trade names or an occasional esoteric term, and I'd like to see how it does then.
Jussi at 21 Jan 2010, 11:35pm. Permalink
A great test, Phil. Thanks a lot!
I too was curious. Without memorizing the passage, I typed it in about 8m 50s with the iPhone in portrait mode, which I use daily.
I was forced to hunt for the correct position in the passage at times, memorizing would have helped me a minute or two the least. I also made some mistakes that I had to fix, that part of iPhone text input would need some work.
I must say that it is not only about speed, it is also about how pleasant the text input is. I find the virtual keyboard much more pleasant than a cramped physical keyboard with a bad feel. At times, when the dictionary and adaptive click targets work as they should, it is almost Zen like.
Matt Ginzton at 21 Jan 2010, 11:37pm. Permalink
I just tried this on a computer keyboard and a Palm Pre. I'm not familiar with the passage and didn't spend any time memorizing it, so I was just transcribing it -- that wasn't a problem for the computer input since I could keep my eyes on the source passage and touch type and proofread at the end, but it slowed me down somewhat on the Pre since my eyes had to dart back and forth and I kept getting lost in where I was in the source passage and then having to hunt around and find my place. So if I tried a few more times and knew the passage, the computer time wouldn't get much better, but the Pre time would get somewhat better.
Results: computer 1:59, Pre 6:00.
(I've also owned a Treo 650 and an iPhone; I don't have the iPhone handy but I'd be curious to try it; my subjective impression having used the 3 of these tiny keyboards is that I preferred the Treo to the Pre and the Pre to the iPhone; the iPhone comes in last not so much purely for speed but because it feels like it requires my full concentration, which is less true typing on the physical keyboards.)
Genghis7777 at 21 Jan 2010, 11:44pm. Permalink
Cool write up.
I had a crack at doing something similar last year: myapplenewton.blogspot…
I didn't have the patience for a piece of text as long as yours though. I got a speed of around 20-ish words per minute with a roughly 95% accuracy. This is comparable to your 23-ish words per minute.
Handwriting as a means of producing information is slowly becoming an obsolete practice. Those who do write anything with a pen all print these days. My staff complain that they don't know how to read cursive writing anymore and teachers tell me that they don't teach it anymore, only printing.
I haven't tried it but I suspect that HWR accuracy would go up if the text is printed rather than cursively written but input speed might fall. Hmmm... must give it a try.
Genghis7777 at 21 Jan 2010, 11:50pm. Permalink
Duh, should have read your entry more closely... you already did your test using print writing rather than cursive. Apologies for the wasted bandwidth
alanj at 21 Jan 2010, 11:54pm. Permalink
I gave this a try with my two preferred devices, my MacBook Pro and my Sidekick LX. I took 2:35 on the MBP's standard full-size keyboard. I took 3:52 on the Sidekick's keyboard, aided somewhat by the built-in AutoText software. ("i hv engh faith" automatically expands to "I have enough faith", etc.) So, about 25% faster than your iPhone time, and without the advantage of familiarity with the text.
In both cases I was touch-typing, spending most of my time reading the source text, a little bit checking my output for errors, and almost none watching the keyboard. This is an often-overlooked advantage of tactile keyboards, especially for mobile devices - you can type accurately while walking around and still watch where you're going.
I find it perverse that all the best tactile keyboards on mobile devices are reserved for so-called "messaging phones", cheap phones with poor specs marketed at teens. The Sidekick is the best of the bunch, but is still considered a kid phone, and is now orphaned technology. Even so, the keyboard alone makes it so far superior to more "serious" phones for core functions of sending messages and entering data that I'm reluctant to switch.
Kris at 22 Jan 2010, 12:00am. Permalink
Brilliant article. My hat's off to you for taking the time to run that many tests. And that's a surprising result - I would have expected the iPhone to fair much worse.
John Birmingham at 22 Jan 2010, 12:04am. Permalink
Fascinating post. I'm a writer. I broke my arm last year and had to start using dictation software. I'm on a Mac so the only option was MacSpeech Dictate. I've found it great for generating new copy, not so much editing old text which was previously input with a keyboard. FWIW I just tried the software with your passage. From a standing start, completely unfamiliar with the text, it took 3.24mins to enter and another 1.22 mins to correct.
After some practice I was able to drop that to a clean read in 2.47 mins.
Sean O'Flaherty at 22 Jan 2010, 1:24am. Permalink
Not only have you countered curent claim of the iPhone keyboard being slower then a mobile hardware keyboard. Bot the older claim that a Newton is much slower then a Graffiti based palm.
Robert Fisher at 22 Jan 2010, 1:44am. Permalink
Using the Dragon Dictation app for the iPhone made me realize something. With all the other input methods, I compose as I input. With Dictation, I have to compose a certain amount in my head and then speak it all at once.
The ease of editing is at least as important for me than metrics like speed and correctness since the vast majority of the time I’m not just inputting but composing.
Also, I’m amazed that so many people seem to find the landscape iPhone keyboard easier to use. I find it’s width awkward for thumb typing. The dynamic target sizing—for me—seems to make up for the small keys of the portrait keyboard very well.
Josh Holat at 22 Jan 2010, 4:17am. Permalink
Hey, I was wondering if I could have permission to use one of these images in an Appletell.com post about this article. Your site will be linked to as the source for the content. Let me know!
Jon Whipple at 22 Jan 2010, 4:23am. Permalink
Phil, this is great and I am glad that Gruber Fireballed this. I am impresses that you took the time and just did what is a simple empirical experiment. It's pretty interesting that the Newton performed so well. Testament to the fact that Apple has always been an engineering company no matter how they're percieved. I love this kind of stuff. Thanks again.
elmas at 22 Jan 2010, 4:36am. Permalink
I guess some people have a lot of time on their hands.
Hans Gerwitz at 22 Jan 2010, 5:08am. Permalink
This was a fabulous idea and your write-up is. Serendipitously, I have extensive experience with precisely the same group of equipment (with the exception of a "Pro" on that 13in MacBook and generic ink & American paper with the Lamy). My memory suggests the MessagePad and Graffiti were both nearly as fast as paper, though my natural handwriting is blocky printing.
Phil Gyford at 22 Jan 2010, 7:52am. Permalink
Thanks again for all the nice comments, they're much appreciated. To respond directly to one, akatsuki said:
Although I don't find the iPhone's autocorrect as annoying as when I first started, I rarely use it - I see it suggest words but by the time I've thought "Oh, if I press space now then it will fill in the rest of the word for me" I've finished typing the word and moved on. So I don't, unfortunately, get much benefit from the dictionary.
Joanna at 22 Jan 2010, 7:52am. Permalink
Did you use graffiti I or graffiti II? The original graffiti is amazingly fast, but the second one that they used while fighting that stupid court case had a lot more double strokes in & is very, very slow. you can't have t's and i's taking longer than x's.
Phil Gyford at 22 Jan 2010, 7:55am. Permalink
Joanna: It was Graffiti 1, as far as I know -- they were all single strokes. I was pleased that I remembered a few of the slightly alternate but more successful ways of doing some letters. Like writing the 'u' from left to right, but the 'v' from right to left. Something like that, anyway.
djk at 22 Jan 2010, 7:57am. Permalink
For software keyboards (iPhone etc.) is cruciat to have disctionary ready. If you try it in some unsupported language (and there are still lots of such) it will turn into tedious task. :((
towi at 22 Jan 2010, 9:51am. Permalink
Great idea to do such a test. I like it.
...although (there had to be an although, right?) -- I would like to see a broader experiment.
I.e. I am terribly bad with the keyboard of the iPhone. I have it for over 2 years now. I really can not use it to enter more then the subject of an email. I miss the keys, the proposed words are always wrong and I have to stop typing, look, check, find the tiny 'x' in the suggestion and read where I was, cuntinue. Horrible! Ok, so I am typing german, ok, but can it be that the autocorrection for english is su much better then it is for german? Other applications can do it (Google Wave can't by the way).
Then, in addition, I used a Palm Vx and the excellent Palm clone Sony Clie back then. Ok, typing coule have been faster, but I managed quite well. I was definitly faster then I am today with Mobiles T9.
Actually -- the difference is so big, that I frequently wish there woule be a Graffity implementation on the iPhone! I would immediatly buy it!
Anyway: good comparision. I with I had all those gadgets... Hmm, wait. I still have them! Except for a Newton. I would have to dig out the Clie, and I seriously doubt that the battery is good for swiching on even. hmm...
tschau, towi.
Richard at 22 Jan 2010, 10:28am. Permalink
Nice job phil.
Contrary to all the derision in the tabloid press, and all buffoons who believed their rantings. The Newton text recognition was far better than anyone perceived it to be. (including Steve Jobs - and he was very short sighted on this matter)
If only they would provide Newton text recognition (retailed software?) for modern touchscreen hardware. I'd buy it in a heartbeat.
By killing the technology, Apple has probably set the UI field back by 10 years.
Arnold at 22 Jan 2010, 10:30am. Permalink
Here's another that is outside your equipment, Phyl, I was hoping the folks at TenGo would put up an ipone app. I've been using Tengo Free for a few years and its BRILLIANT. Six keys only and I can type on my old WindowsMobile 6.1 HTC phone with a stylus almost as fast as I can touch type. Only for short diary entries and contacts though. They used to show a video of a girl speed typing but Tengo seem to have lost direction lately. www.tengo.net is a suspended domain today, bankrupt? Who knows? What a shame! I hope somebody picks up that idea of six keys and predictive completion.
David Neale at 22 Jan 2010, 12:35pm. Permalink
Very interesting indeed. The MessagePad's performance is impressive, given its age and what can best be described as underdevelopment. Furthermore, the MP offers not only the option of immediate hand-writing recognition, or deferred, but also that of no recognition, so that what you write appears on the screen just as you write it: just as what you write appears on the paper notepad just as you write it. In this way, the MP provides almost as fast input (somewhat reduced, I suspect, because of the smaller viewing area and the requirement to scroll) and equally legible (or illegible!) output. This is one of those cases where the MP was able to adjust to the situation, also demonstrated by its ability to work in both portrait and landscape modes.
Anyway, I'd like to see the time you'd need to simply write the text on the MP without HWR.
The iPhone's performance is remarkable and not at all what I would have suspected. I'm a relatively new user (iPod Touch) and still have trouble with the keyboard. However, I find its use in landscape mode to be far less prone to error, so I imagine that the result in your test would have been even better had you used it that way.
joh at 22 Jan 2010, 1:50pm. Permalink
Towi: Which version of the iPhone OS are you running? I found that at one point there was a dramatic improvement in the german dictionaries. You should also try and reset the stuff it has learned (it's somewhere in the preferences) since it may have picked up a lot of wrong words from you. If nothing helps you can still disable it altogether.
J at 22 Jan 2010, 1:58pm. Permalink
Richard -
Apple never killed the handwriting recognition in the Newton - it has been in OS X for years now as "Inkwell" and works just as well, if not better than it did so many years ago. Grab a Wacom Intuos Bamboo and have fun!
en.wikipedia.org/wiki/…
Farzal at 22 Jan 2010, 2:03pm. Permalink
very very well written. the touchscreen keyboard performance definitely beats the intuition
A Stevens at 22 Jan 2010, 2:18pm. Permalink
Very interesting article and experiment.
Thank you.
Sumocat at 22 Jan 2010, 2:27pm. Permalink
Great battle royale, Phil. Curious about your typing style on the iPhone though. Two thumbs same as on the Treo? I use one thumb myself. Wind up tripping over myself otherwise.
Zack at 22 Jan 2010, 2:46pm. Permalink
I think a desktop keyboard would easily outperform a laptop keyboard. The best is the Model M:
en.wikipedia.org/wiki/…
Matt at 22 Jan 2010, 2:51pm. Permalink
Interesting article, I too have had most of the same hardware as you, although it was a Palm V and not a Vx, and I never had the Newton, but the iPhone, the Treo and the MacBook Pro I have as well.
Honestly not surprised the iPhone keyboard did better than the Treo, I hated the Treo's keyboard. It was the most awkward thing I have used on a phone. Unfortunately for the Palm V I never got the hang of graffiti.
The iPhone though, too easy. But it is one of those things where sometimes you let go a little and trust it. I never liked the predictive input on the Nokia's either and used to manually input the entire text for about the first 6 months that I had a Nokia. After a while I started to use it more. The girl I knew who could message from her Nokia without looking at the screen blew me away though.
Landscape keyboard is definitely easier, the buttons are not so cramped, so you get a larger surface area to target, although I agree on the text area issue. Still, I've used it to make posts of a couple of paragraphs into online forums, and while I wouldn't use it for extensive text input, that's not what it's for.
Phil Gyford at 22 Jan 2010, 3:08pm. Permalink
Sumocat: I use two thumbs on the iPhone and it works well for me, but I can understand the possibilities of tripping up!
Zack: For a long time I've used an adjustable Goldtouch keyboard which is very good. I haven't tested the two directly, but I don't feel noticeably slower using the MacBook's keyboard, which has smaller amounts of travel.
Karl at 22 Jan 2010, 3:14pm. Permalink
I'd like to see another test - QWERTY vs. Dvorak keyboard layout with an experienced user on each.
Todd at 22 Jan 2010, 3:47pm. Permalink
Just for prosperity's sake, I tried this on a Blackberry Bold (9700) and got it in 5 min, 46 seconds. Then on my iPhone and got 6 min, 53 seconds. Same test and same setup. Also have similar lengths of experience of both devices. Perhaps a more conservative, pro-capitalism paragraph of text would provide different results? :)
DaveM at 22 Jan 2010, 3:58pm. Permalink
You should do your test with ShapeWriter. I found it very comfortable and fast. www.shapewriter.com
Richard at 22 Jan 2010, 4:23pm. Permalink
Re: inkwell
I don't know how I missed that. But I'm not such a heavy OS-X user nowadays. Though I've still got my G4 & OS-X Panther, it looks like I might have to invest in a new Wacom tablet. The one I'm still using (on XP) only has a 9-pin serial connector (Intuous) and finding a keyspan 9-pin serial-USB adapter, for use with a USB-only Mac is very difficult these days. (stupidly gave one away to a teacher for whiteboard use)
DOH!
Russell W at 22 Jan 2010, 4:35pm. Permalink
This is a great idea. I recently discovered the Livescribe Pulse pen and am in love with it for three reasons. I have always printed (when writing) and have kept journals for years. I would suspect I write faster than I type, but moreover I love the feel and thought that goes into writing by hand more than typing. Two, with the pulse pen, all of my writing is automatically preserved. And really, what's a safer back up than paper? And finally, the Optical Character Recognition (OCR) from Vison Objects is excellent (especially with my handwriting I think) so I can pretty much avoid typing up things I've written by hand. Let me tell you, it is a bizarre thing (akin to sorcery) to write an email on a piece of paper. That it records audio has been a fantastic thing for lectures, too!
mbs at 22 Jan 2010, 5:10pm. Permalink
Thought I'd try typing it myself on a regular fullsize keyboard, and came in within a second of OP's time. Neat.
BE at 22 Jan 2010, 5:40pm. Permalink
This is interesting, but the only thing I want to point out (and you did say that your results were definitely not scientific), but you're USED to typing on an iPhone ... if you were as used to typing on a Treo than you were with the iPhone, or with some of the other options, I wonder how this would change the results?
I personally just got a new Droid, but I can't type with the speed or accuracy that I did with my HTC Touch yet, even though my touch was on its last leg...
Interesting either way, though :)
Dankoozy at 22 Jan 2010, 5:50pm. Permalink
You should have tried the Nokia E63 keyboard. It is much better than the iPhone, I used to have an iPhone and typed much faster on the E63.
Of course the best keyboard on a phone has to be that of the Nokia 6800/6810
SLB at 22 Jan 2010, 5:56pm. Permalink
Another Fitaly fan here. I no longer have a working Palm to test it on, but I think it was faster than the physical keyboard on my Nokia N810, which is a fairly large horizontal-format thumb keyboard. I bet Fitaly would have challenged or beaten the iPhone in this test. (It does take a bit of practice to get fast on, though.)
Reagan at 22 Jan 2010, 6:15pm. Permalink
Hey great article. I've been using my iphone 3gs for the last 3-4 months now and I must say that it still causes me a bit of typing grief. I got about 7 minutes 40 after typing out that passage in landscape mode and I figured that because you had the piece memorized and weren't copying it like I was, that it proved to be the decisive factor in speed. I can type quickly with my phone but only when it has my full attention - and not when I'm copying something having to look up at the computer screen and back to check on what I'm typing. It's a real bottle neck. Actually, I wonder if anyone had any thoughts of the best input device for a mobile phone. Tactile, i.e. Blackberry? Speech recognition, i.e. Nexus one? What does the future hold for these mobile devices? Personally, I want a phone that allows me to type on it without giving it my full attention and sadly, a touch screen on my iphone doesn't fit the bill.
Scott Berkun at 22 Jan 2010, 6:26pm. Permalink
This was awesome - thanks for taking the time to document this.
It'd be great to replicate this with more than one person's data :) - anyone up for it? Most folks don't have all these devices, but it might be possible to compile individual trials for individual devices.
Jones at 22 Jan 2010, 6:28pm. Permalink
I would've liked to see Blackberry devices included in this since they're made for typing on the go and have a better layout than that Treo (which was awful).
Also, unless you've been using an iPhone for the past year or more, there's no way you can type that fast without multiple input errors; the cramped layout is difficult to work with, I just assume you've had a lot of practice with it.
Larry Marso at 22 Jan 2010, 7:05pm. Permalink
I taught myself Gregg shorthand in graduate school.
With pen and paper, I can write 250 words per minute.
Read time takes just a bit longer than longhand or computer output, though. And skimming is difficult.
caleb at 22 Jan 2010, 7:17pm. Permalink
I am also curious how FITALY would place in this test.
Jeff at 22 Jan 2010, 7:44pm. Permalink
Here's another vote for Gregg shorthand. I'm waiting for a device that can read it.
Jon Wright at 22 Jan 2010, 7:59pm. Permalink
Brilliant piece. Actually i can now keep up with my daughter on my trusty iPhone using one finger. Before that there was usually a 5 minute delay between replying to her texts.
On the Bevan quote. Kind of ironic given the mess Labour have put us in. Sadly that speech could be attributed to any modern day party.
Genghis7777 at 22 Jan 2010, 8:51pm. Permalink
7 minutes 33 seconds on a HP 200LX: en.wikipedia.org/wiki/…
Andrew Horning at 22 Jan 2010, 8:54pm. Permalink
Thanks! You've reminded me that I miss my Treo. I've come to hate my newer rubbish, so I'll reignite the 650.
But mostly, thank you for Bevan’s commentary. As the whole world bows and sacrifices to the Golden Calf of State, it's good to remember that it wasn't always so, and needn't always be so.
David Neale at 23 Jan 2010, 12:35am. Permalink
The FITALY keyboard layout is (was) available for the MessagePad, too. I have it installed on one of my MP2100s. It was recently the subject of a number of exchanges on the NewtonTalk Mailing List at newtontalk.net/
Richard Fairthorne at 23 Jan 2010, 6:14am. Permalink
Great idea.
While I am a strong proponent of the iPhone keyboard, I was actually surprised to see how high it placed. I have never beat a blackberry in a typing test, but I always come within seconds.. usually surprising the competing blackberry owner.
The key to the iPhone virtual keyboard for me is to ignore the suggestions, and assume it is doing something reasonable even when I make minor typos. With that in mind, I just mash my thumbs, and go back to correct 1 or 2 words when I'm done.
Anton at 23 Jan 2010, 6:40am. Permalink
I did nearly as well on my thumbpad (HTC-G1) as on my Natural keyboard - I can touch type on both so I can keep focus on the text. Using "BetterKeyboard" was significantly worse, mostly due to context switching and looking for home keys.
around 4mins on keyboard, 4:40 on thumbpad, and 6:30 on touch-screen.
I tried Android's voice recognition stuff, but my poor G1 can only deal with a few seconds at a time, and whilst accurate, was tedious.
Some of this is practice, I only use the on-screen keyboard for short text entries, but will frequently break out the thumb-pad for vim sessions on remote servers.
Olaf Kreitz at 23 Jan 2010, 6:44am. Permalink
I am surprised that the Newton can beat a Palm. I remember when I first got my USRobotics Pilot 1000 (I believe that was the model), I spent the summer in Hamburg for an internship with a small branding firm. My boss had a Newton (of course) and claimed his device was way superior. I won every single speed contest we did, the Newton was soo slow and my first gen Pilot way faster. After a while (I owned pretty much every Palm device, maybe skipped a few), I was faster than handwriting and only a little behind a full keyboard. I still believe Graffiti was the best char input method ever. Rumors are that I adopted a few Graffiti characters in my handwriting (A, K). I am very pleased with my iPhone today but Graffiti totally ruled back in the day.
Najam at 23 Jan 2010, 8:18am. Permalink
Well, it depends upon the situation. If one wants to write a blog or essay, full size keyboard will be the best. But if you are in meeting or lecture and wants to take notes, most probably pen and paper will be the best.
Most of the people conduct experiments in ideal environment but we have to look at different situations as well.
Geoff at 23 Jan 2010, 1:06pm. Permalink
This is all very well, but how long would it take you to enter the same text in Morse code? Or semaphore? Or phone TXTING?
Itsuro Yoshimoto at 23 Jan 2010, 1:44pm. Permalink
Phil,
It is always great to do testing by comparing different methods.
Thank you for your information.
I just feel that it might be a good idea to seperate basic input part and word prediction/correction part because input methods are combined with these two elements.
----------------------------------------------------
I have a feeling that the reason why iPhone's performance was good is just because the word correction/prediction part was great.
How fast was it when you type in vertical position instead of horizontal? Do you have an error when you type with your thumbs in vertical position?
~~~~~~~~~~~~~~~~~~~~
Anyway, this is great article.
Please continue to try different input methods.
I want you to check the following video as another option for future.
www.youtube.com/watch?…
I broke a world record of fastest text entry with FIO-KeyBO.
FIO-KeyBO is simply aiming to replace existing multi-tap text entry for 10-key mobile phones, and OEMs can combine any kind of word prediction/correction software.
Please let me know how do you think.
-Best..yi):
Preetham Venkky at 23 Jan 2010, 8:05pm. Permalink
Fantastic work done. Logical step-wise conclusion. I just converted it to an Info-graphic which I'm sure you will like. Check it out at - www.ideasmaverick.com/…
Adit at 23 Jan 2010, 8:12pm. Permalink
Very cool. I I too think that this experiment could benefit from a test using Swype. It's a gesture typing input method that I know is available for Android phones. If there is a iPhone version, it would definitely be worth checking out.
Andrés at 23 Jan 2010, 8:35pm. Permalink
Took me 2 minutes on a keyboard :)
Damien Stolarz at 23 Jan 2010, 9:35pm. Permalink
Could you try out danbricklin.com/log/20… and post the results? I'll reimburse you the $2 if you wish.
Amy at 23 Jan 2010, 9:46pm. Permalink
Itsuro, of course the iPhone's word recognition effects the outcome--that's the point. I think what this experiment was testing wasn't how fast can one person manipulate the tool to get text in, but how quickly can a person AND the tool together get accurate text inputted.
I used to love Palms. I had 65 IIIc's at one point that I used with my 6th grade science students for a while. It was awesome, and some of them got really good at Graffiti. I thought I was pretty good at it too (1, not 2. I was pretty annoyed by Graffiti 2 and didn't know it came about because of legal issues). I'm surprised at how poorly it did in your trials. As I fumble with the typing on my iPod touch, and often long for a stylus and handwriting recognition of some kind.
rlmorel at 23 Jan 2010, 10:05pm. Permalink
I did it with the iPhone Dragon application, and it took me 2 minutes and 45 seconds, with three errors.
My wife was wondering what the heck I was dictating...:)
Collin at 24 Jan 2010, 9:39am. Permalink
I did the test using Swype on an HTC Touch Pro 2 in portrait mode. Did it 3 times:
1) 13:12
2) 9:56
3) 9:55
I've been using Swype for about two weeks and have been very impressed. I can take brief notes while listening to lectures which I never could using the hardware keyboard or the standard onscreen. I'm very slow on my iPod touch onscreen keyboard.
So, I was pretty surprised to see how slow my times were on this test. However, I think my main problems stemmed from having to look back and forth between the two screens (laptop and phone) to read the text and then to swype it in. When I tried swyping blind by keeping my eyes on the computer and just swyping by instinct, i went much faster but accuracy on certain words went way down because they would be dependent on starting position. (you don't have to start exactly on the first letter for many words, but for some, you do.) the large number of commas was also a big speed bump since the comma is on the SYM page, which means one press to switch, one press to type the symbol, another press to go back to ABC. A comma on the main kb would speed things up, but even more so memorizing the text.
a at 24 Jan 2010, 9:42am. Permalink
the graph is wrong, or the time reported for macbook typing is wrong.
You state 3:15 in the article but the graph reads 4:15.
Phil Gyford at 24 Jan 2010, 10:26am. Permalink
A: I think you're mis-reading the graph, which shows 3:15 for the full-size keyboard.
Damien: I'm not sure I have time to learn new techniques and re-do the test, but I'll have a think about it.
Fred at 24 Jan 2010, 11:47am. Permalink
Great test!
Would be even greater if you tried swype or slideit and timed that too!
Also a hw keyboard like the one in the droid would be interesting to know speed on.
Phil Gyford at 24 Jan 2010, 12:42pm. Permalink
Is that anything like the hardware keyboard on the Treo which I tested?
Thomas at 24 Jan 2010, 8:03pm. Permalink
I find it interesting just how modest the advantage is, that hand-held devices' keyboards have over pen and paper.
I suspect that there are two key problems with little keyboards:
1. Severe demands on the thumbs' rate of acceleration.
2. Waste of momentum on keypress (no spring).
I think both problems can be solved, and perhaps they will be with the invention of a new kind of keyboard. I imagine QWERTY keyboards on handheld devices will amount to a stop-gap the way trackballs on laptops did.
Jacob Ras at 24 Jan 2010, 8:52pm. Permalink
Nice test, looks like you can type really fast with iPhone. :) Maybe you should also test a phone with real keyboard, like a Blackberry.
Lars Tong Strömberg at 24 Jan 2010, 8:57pm. Permalink
The comparison as such is interesting, but as you point out, it´s hardly scientific. I think it very much depends on the individual experience as well as technique as for how fast each of the methods go practically.
Hope some researcher digs deeper into this though (someone probably already have?) An area I would find interesting is to see comparisons between different languages. My guess is that Chinese for example is a much faster language than say e.g. English.
dmccall at 24 Jan 2010, 9:47pm. Permalink
Great writeup, however I cannot believe anyone is faster on an iphone than a Treo. The iphone is so error prone and wrong in its auto correction, that it takes me at least twice as long to type on it than the Treo.
Also a fairer test would be to use a laptop with a real keyboard instead of the dreadful one on MacBooks.
Mikkel at 24 Jan 2010, 10:33pm. Permalink
Interesting test. I compared typing the text from this article (not having seen that previously) on two devices. Here are the results.
My own test:
1. Lenovo X61 portable keyboard 4:16
2. Nokia e71 qwerty smartphone keyboard 6:34
The common method for typing on the e71 (and mine too) is using two thumbs, holding the phone with both hands.
I am VERY impressed that the author types faster on the iphone software keypad than on a smartphone qwerty. I have always bought and used qwerty smartphones for their supposedly better notetaking and writing capability. I am sure that is still the case for myself; however I have come to re-evaluate that fact in light of this article. Thank you.
Genghis7777 at 24 Jan 2010, 10:50pm. Permalink
Fitaly keyboard on a Newton MP2100: 14 min 28 sec.
Shawn K. at 25 Jan 2010, 1:08am. Permalink
How about Simplified Chinese and other unique character sets!? :)
Foo at 25 Jan 2010, 3:58am. Permalink
People keep mentioning training on the Newton. There was no training on the Newton engine you used.
The Newton's cursive engine, Calligrapher, created by the Russian firm Paragraph International, required training.
The Newton's later print engine, Rosetta, created in-house by Apple, had no training at all. There were "training" apps, but they're actually to train *you* to write better for Rosetta.
Foo at 25 Jan 2010, 4:02am. Permalink
Oh and...
I get about 2 minutes 15 seconds typing. Yes, fast. :-)
Genghis7777 at 25 Jan 2010, 1:12pm. Permalink
17.1 words per minute using the Newton onscreen keyboard
Genghis7777 at 25 Jan 2010, 1:13pm. Permalink
... which is 12 minutes 55 seconds
Mike Connally at 25 Jan 2010, 5:47pm. Permalink
Well done on conceiving and executing this comparison. It pretty much confirms my expectations. As a Newton-lover since OMP, and still having a brace of MP-2100s ready for action, I was pleased to see confirmation of my firm belief that Rosetta would beat Grafitti. But I must pedantically point out to the several people wondering how much training the Newton had, that the printed recognition engine (Rosetta) did not do 'training'. Only the cursive engine (Paragraph) did. Cursive never did it for me, but Rosetta rocked. For those lucky enought to still have an operational Newton, you must know this Easter Egg. Write the following: Rosetta! Rosetta! Rosetta!
Olavi at 25 Jan 2010, 7:55pm. Permalink
Interesting indeed. Physical keyboard was the reason for me getting Motorola Milestone (definitely worst physical keyboard I have witnessed). 11:23 - not too good. I guess the iPhone significantly outclasses Milestone.
lillejord at 25 Jan 2010, 8:39pm. Permalink
I think some people are really missing the point here. The difference between the iPhone and the Treo are negligible and I think the difference between the Treo and any other QWERTY phone would be equally negligible. I think it shows that Soft-key and Hard-key very similar. Also, for landscape orientation, to be fair, you would have to test a landscape physical keyboard.
What really shocked me was how slow (relatively) your writing was. Especially for how sloppy it is. I wonder if this is due to how little we actually write these days? Maybe your texting thumbs are in better form?
The second thing to come to mind (since I'm currently in the market--for artwork, not writing) was how a modern tablet would compare to the Newton. 13years is a long time to improve handwriting recognition. I know you said it wasn't tested because you didn't have one, it doesn't change my curiosity.
Lastly, you need to redo all these test while riding the bus! :)
Doedina at 25 Jan 2010, 10:16pm. Permalink
I wonder about this every time, I have to write something, I tend to use pen and paper, it always feels easier to edit.
I do wonder if you wrote in shorthand, how much faster pen and paper would be.
Shane at 25 Jan 2010, 11:03pm. Permalink
Tommaso: "an" handwritten typo? you mean "a" handwritten typo? :)
Shane at 25 Jan 2010, 11:06pm. Permalink
Tommaso: "an" handwritten typo? you mean "a" handwritten typo? :)
Mxx at 26 Jan 2010, 5:28am. Permalink
Like others have asked, would be interesting to see your results with, for lack of a better term, "non-tapping" keyboards such as Swype(Windows Mobile, Android) or SlideIT(Windows Mobile) or ShapeWriter(iPhone, Android, Windows Mobile).
Dan at 26 Jan 2010, 5:55am. Permalink
Great test and very interesting results! I see one possible flaw in your method though: You continually typed the same paragraph over and over again, each time becoming more familiar with it. By the time you reached the iPhone, you had typed it 9 times. This may have contributed to your speedy result. Likewise with the Treo. (the Graffiti just sucks, that's why your time was so slow on the Palm.) I'd like to see a few different people try it while writing a different paragraph each time. Just a thought.
Dan at 26 Jan 2010, 8:54am. Permalink
Have you tried the FITALY on-screen keyboard? It works on your Palm V (and I think it's free on that platform, now that it's outdated).
It is extremely fast. In fact, I am faster using FITALY keyboard than when I use a thumb keyboard.
Phil Gyford at 26 Jan 2010, 9:35am. Permalink
Thanks for all the thoughts folks. I won't be repeating the test with any other devices/software though -- I just wanted to try the devices I know and am familiar with, and using the input systems they shipped with. I'm sure there are many clever third-party solutions that are great if you invest the time to learn them, but that's not something I have time for, sorry.
Flavio at 26 Jan 2010, 10:15am. Permalink
It is intersting but not much relevant.
I mean, this test spawns from the recent tablet debate; i don't see the tablet as a runner up in office productivity, i can see it only as a pda, so for eg. in school doesn't really matters _how much_ text you can input, but _just how_ you organize it; 10 words can be sufficent, if are the right words. So the tablet has definitely the potential to be a big hit (i mean a revolution in our digital lives), what is lacking till today is the right annotation software as well as drawing precision/text recognition.
At now i don't see this coming on the PC market, seeing straight ports of Win7; let's see tomorrow what Apple's got, but i'm still skeptic this is still too soon.
PS: i think a ballpen it's faster than a stylo.
Navarro Parker at 26 Jan 2010, 6:44pm. Permalink
Everyone complains about the iPhone keyboard at first use. But once I learned to "use the Force, Luke" and trust the autocorrect, I can blaze through messages. I think new users (or people who want to look at your iPhone) try to t y p e e v e r y w o r d 1 0 0 % p e r f e c t l y .
I can type MUCH faster on iPhone than I can with other phone's "real" keyboard.
Dennis at 28 Jan 2010, 3:36pm. Permalink
Recently have been working with SWYPE (www.swypeinc.com/) and found it to be an interesting alternative that greatly speeds up my text input on cell phones. I'm still not as fast as when I use a regular keyboard, but getting pretty close - definitely about twice as fast as text typing on smaller screens.
Billy Bob at 2 Feb 2010, 2:17pm. Permalink
Thank you for spending so much on time on such a useless comparison. I am dumber now than I was before, and that's saying a lot. I hope you got paid well to waste so much of your time. Wait a minute, I'm wasting both our times' by writing this. Oh, well, I'm glad somebody took the time to say it.
Charlie at 13 Mar 2010, 9:07pm. Permalink
I've been developing a nice elegant phonetic shorthand to replace Gregg - every sound has a unique shape, and they're natural and quick to write. Now I'm considering turning that into a stylus-based text input method. Does anyone have any thoughts on whether there would be a market for this?
Like Graffiti (and unlike Swype), you wouldn't have to look at what you're writing all the time, but it would be quicker then Graffiti because you would only need to input the sounds. Combined consonants also have one shape (e.g. s-t-r, g-l), and common consonants can be added to vowels with a slight flourish (e.g. a-r, a-l). For homonyms, the user might have to choose from the options unless there is context intelligence.
Is this just 19th century, or could it become popular?
Arthur Chandler at 19 Apr 2010, 5:53pm. Permalink
How about an update with iPad?
Phil Gyford at 19 Apr 2010, 5:59pm. Permalink
If you would care to buy me an iPad (and send it to the UK) I'd be more than happy to update this!
Laurent Guyot-Sionnest at 10 Sep 2010, 8:45pm. Permalink
Super post and comments !
Many numbers from various users of various input systems and devices.
But, what may lack here, are some conceptual and analytical background.
McKenzie has provided it and he has built a method to compare any input solution.
The number of fingers is a very important factor (10 >> 1 ! ).
Also, how long they have to travel to input words, spaces, punctuation, numbers and correct errors.
And how much attention you have to focus on typing on the wished key (blind touch typing vs hunt and peck).
From the very various contributions above, i can tell that automatic correction and suggestion software is now the main differentiator.
The problem is that we don't have the same language software on these different input devices.
For instance, the advantage of Swype, Shapewriter and SlideIt are that by maintaining your fingers on the screen
you give additionnal information before reaching the next letter, which help the language software, then you need less attention, which allows to move a little quicker...
Cumulating a stylus on a very small keyboard with these "gliding" solutions and good correction-suggestion you can nearly conpensate from the fact you use only one actuator...
But the future quickest input solution, in fact a brain output solution, would propose a flute-like fingers use on a touch surface
with a good correction-suggestion sotfware providing suggestions where your message is built,
because your fingers would be able to select the good one without moving your hand, as fast as you would type a letter.
With that solution, every user would reach his personal maximum speed allowed by his own fingers druming capability
[ranging from 2 tps (old) to 12-15 tps (virtuoso) ].tps = tap per second.
The real record for input will then be, by a trained virtuoso:
nb of produced letters each second = 12*2*2*2 = 96 l/s = 5760 l/mn = 1152 w/mn
12 tps as an average,
each tap = a syllab (phonetic ) = 2 letters,
suggestion = a full word = 5 letters
2 hands acting alternatively (as on a full qwerty).
For ordinary people like us who provided comments, numbers could be lower:
7*2 = 14 l/s = 840 l/mn = 168 wpm with only one hand and 336 with two hands...
Who want to build that with me ?
tikilgs@gmail.com
Noop at 14 Oct 2010, 4:26pm. Permalink
Tried to type your text on the hardware keyboard of my HTC Universal. Got, after corrections, roughly 8:50 on the first try, which is good, since I'm belarussian and don't have much experience typing long English texts. No autocorrection or spellchecker. Corrections were easy thanks to presence of stylus. I believe, HTC Uni keyboard could easily beat iphone and iPad keyboard I'm currently using in hands of more experienced native English speaker. Also, autocorrection dictionary on iPad drives me mad. English one only makes me type wrong search queries, but Russian dictionary is completely intolerable, making ten time more errors than it corrects. And I'm using latest iOS 4.2 beta
Ants Neo at 14 Nov 2010, 7:50pm. Permalink
it took me about 14 minutes to write it with Grafitti on my HTC HD 1 with winmobile and plugin Block Reqonizer. I'm sure I can write it a lot faster in my language Swedish, like 10 minutes?
Tried then on iPhone and gave up after two minutes having only written 11 words. I totally suck on iPhone. Will tomorrow bye me the new HTC Desire HD, an electric stylus and install then Grafitti for Android.
Sorry Apple, for me the Graffiti is the best solution and the most important for my choice when buying a phone.
The second function I need is Wifi thetering and iPhone does not offer me that without jailbreaking the phone.
Fix this two functions and I'll buy your phone!
change your life at 25 Mar 2011, 6:53pm. Permalink
The difference between the iPhone and the Treo are negligible and I think the difference between the Treo and any other QWERTY phone would be equally negligible.
Fat loss Nutrition and Fitness at 12 Jul 2011, 7:51pm. Permalink
While I am a strong proponent of the iPhone keyboard, I has actually surprised to see how high it placed. I have never beat a blackberry in a typing test, but I always come within seconds