https://www.exideas.com/ME/android/down … agEase.apk
I got the news from MessagEase's page on Facebook.
[edit: Don't worry though – the new version 10.0.2 has been released as of 2014-08-01!]
]]>There's a new keyboard I really like the look of called "Nin". It works like Keymonk so you can swipe with both thumbs. It's THE most creatively designed keyboard I've seen yet. The attention to detail that has gone into it is amazing. You also have a lot of settings you can mess with. Right now you can get it as a jailbreak tweak on ios, or in the apps Hipjot/Nintype. Should be coming to Android soon as well.
]]>MessagEase is a wonderful touch screen keyboard for its large friendly keys, fast typing and lots(!) of possibilities! The combo of tap mappings for the 9 most common keys and drags between those keys for the other letters/symbols is very powerful. It's very well optimized for speed and ease too – but a bit of a hassle to learn and remember early on. That may be a stumbling stone for the new user.
I've made a "Colemakoid" remapping for the drag positions of MessagEase. This works because the home row in Colemak and other optimized layouts corresponds largely to the tap mappings on the MessagEase keys so the drag mappings are left to be arranged to a pattern familiar to Colemak users. It won't quite work for users familiar only with QWERTY, obviously, but should still be easier to recognize and remember than the default layout since many of the less frequently used Colemak mappings are similar to QWERTY ones. The only downside is that you won't benefit as much from some of the MessagEase teaching Game levels that are based on the default drag mappings.
The horizontal drags were kept for shifted number keys according to the numbers on the keys you drag from: For example, right-drag from 'A' (which is also '1' on the NumPad) corresponds to Shift+1 on a standard keyboard so it gives you an exclamation mark. The diagonal drags are the top and bottom letter rows laid out largely as in the Colemak layout, very visually recognizable. Diacritics are logically placed at the top, and punctuation at the bottom.
I find the layout much easier to remember and learn this way, and I think it should be about as effective as the default centric layout: ≈70% of the action will be on the tap keys anyway, and the diagonal drags are positioned so that the best drags (up/down from the middle column for instance) actually get the more frequent letters in my setup.
I like to type with two thumbs like the fastest MessagEase typists do (ChengWei at least), but I'm not all that fast myself and my ≈45 WPM is much lower than the speeds of Ghen and Davkol for instance. [Update 2019: I'm at ≈60 WPM now!]† I think Ghen still uses a single finger, so who's to say what's best. It'd be nice if users with different techniques could check out my layout and compare it to the default to get some more opinions on the table.
Should anyone want to try out my Colemakoid MessagEase mappings, they can be imported using the following export code from the app, replete with instructions and all:
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
That's so clever. :)
This is what you should end up with, but not all mappings are shown on the autogenerated layout exchange image (there's [] {} next to the parentheses for instance). Your imported keyboard should now look like mine which is quite transparent and unintrusive ("stealth mode") as I've learnt where everything is now, but you can of course change the colors and transparency in the settings if you like (along with size, finger tracks and even button roundness!).
Feel free to replace ÆØÅÞÐ etc with something more useful to you (or just hide/disable them to avoid misfires), as MessagEase is easily reconfigurable.
Also remember that there are drag-return mappings as well (sadly not configurable at present), drag-circle/drag-return for quick entry of capital letters, press-and-hold for quick single digit entry, and the wonderful [C] button for a quite rich combine table – press the hand icon for help/info/settings where you can learn more about it. Also, you can make hotstrings, control strokes and more (Ghen has administered servers with it!) – etc etc. All in all, there's such a wealth of options and settings and together they make MessagEase an incredibly nice and powerful touch screen typing tool!
————————
†: Should you want to calculate your WPM speed from the MessagEase Game "World Game" where you type the standard 160 character SMS competition text about piranhas, here's how: If your test time in seconds is T then the WPM speed is (160 C * 60 min/s) / (5 C/W * T s) = 1920/T WPM. So, e.g., a test time of 32 s equals 60 WPM.
it seems to me the layout is not the limitation, but one's physical fitness and dexterity. maybe if i was still a teenager, swiping may work for me. but as I get older and lazier, tapping is just simpler and easier. worse, i get confused mixing tapping and swiping, ala Messagease.
]]>All in all, I think it's safe to say that the top few scores, at least, are outliers even in the informal sense of the word. They're heads and tails over just about everyone else.
]]>The map entries require a minimum of 40 WPM, and therefore they are already "selected." The average and standard deviation are: 51.2WPM/12.7WPM, with the population size of ~100.
Since ME itself does not report the speed of our users (no internet permission), there is no way of getting a true speed of all of our users. But the average and standard deviation of the scores our users have posted through MessagEase Game, are: 32.7/12.0, with the population of 10k+.
While the game reporting may be biased toward higher scores (pride reporting), it is also the case that people with low speeds would not bother with MessagEase, and people with high speeds would not use and train with the game, and therefore not report their speed.
This is in part a question of what you call an outlier. In some cases, anything that's ≈3 standard deviations away from the mean may be branded an outlier. That only works if the distribution is gaussian and the deviation fairly well known, but it's a workable technical definition. I was using the word more informally here, but I probably should have used another term such as "freak occurrences". :)
As seen in the quote, the only data we do have is selected so one would expect a distribution with a lower kurtosis than the gaussian; that is, more points far away from the mean on each side.
]]>It's very obvious for MessagEase judging from the number and speeds of fast ME typers (and yes, that's me in Norway) that it's not about outliers in this case
Actually, the data suggests the exact opposite. The top score is already an outlier for the top 50 alone! When you add in the rest of the userbase, who are most likely much slower than these top 50, many more of the top 50 (if not all) probably turn into outliers as well.
Here is a very conservative estimate of how many become outliers. On Google Play, 2900 people rated ME with 5-stars, so I think it's safe to assume that the userbase is at least that large. (In fact, with over 100,000 downloads, it's probably a lot larger.) Let's assume that those 2,900 people are evenly distributed (in reality, they would probably be clumped up lower thanks to the normal distribution) as follows:
58 wpm - 100 people
57 wpm - 100 people
...
30 wpm - 100 people
In this very conservative estimate, we get:
Mean: 44.4
Standard Deviation: 8.9
Outlier if WPM > 70.9
# of Upper Outliers: 12
so yes, the top speeds being measured here are very much outliers.
This distribution can be played with (http://pastebin.com/HNqYqaWD). Maybe someone with stats knowledge can come up with a better estimate, but I think the writing is on the wall.
tldr: You can't use the competitiveness of a top list to deduce that none of them are outliers. Even though the list of fastest typists may appear to be undominated, that doesn't take into account the thousands of other users who score less than all of them. With respect to the total userbase, many (if not all) of those top 50 are outliers.
]]>There have been at least 3–4 people clamoring over the first place in the ME Speed contest for quite a while. Does that make ME a fast keyboard for me as well? Indeed, it's hard to say as you point out but at least I've managed to get pretty fast with it compared to other touchscreen input methods.
]]>Since people manage to type blindingly fast with QWERTY you cannot really say that it's a slow layout per se (although you can say that fast QWERTY typing may rely rather heavily on non-standard typing).
That's actually part of my concern: as jfmcbrayer laments, those records are of outliers rather than any useful measure of speed. So long as some person - no matter how rare - is able to type blazing fast, that's what goes on the world record, regardless of how well the keyboard/layout really performs with everyone else.
In principle, I don't mind the proof-of-concept. In practice, this particular concept lends itself to some very poor interpretations. As soon as a world record is achieved, marketing (and fans, and random people) immediately starts hyping the new recordbreaker as the "fastest keyboard". Never mind that the data can't even distinguish whether the keyboard, or the typist, is responsible for said record. It's very irksome.
Anyway, I don't really count as a long-term user anymore; I've been avoiding mobile wherever possible. The various setups posted here just make the computer much more enjoyable.
]]>Lalop: Those records are a proof of concept. Since people manage to type blindingly fast with QWERTY you cannot really say that it's a slow layout per se (although you can say that fast QWERTY typing may rely rather heavily on non-standard typing). MessagEase and now Fleksy have proven their mettle in this way, and with MessagEase at least it's easily seen that Cheng Wei was no solitary swallow. The current ME record is held by someone else.
]]>AnySoftKeyboard sounds like a gem. Can one configure it to be a MessageEase-like layout?
currently you can only tap a key to type. swiping to type is not implemented. it's in the works, but no ETA. however it does support swiping gestures to perform general actions like change keyboard, copy, paste, move cursor, etc. you can customize these gestures, and tweak many other stuff as well.
]]>