• You are not logged in.

Ryan Heise: biased typing test

  • Started by SpeedMorph
  • 35 Replies:
  • Reputation: 214
  • From: Viken, Norway
  • Registered: 13-Dec-2006
  • Posts: 5,362

Ryan, might I make a wild suggestion?

The Key Hero game has been unplayable to me for a while for obscure technical reasons, and I like your longer stretches of quotes better overall. But it has some nifty features that might be considered!

Would it be possible for you to implement a time series of your scores in the different typing games? I'm thinking that on the replay screen there could be a curve of your previous results for that test. It might also compute a mean result for the last n games (n could be set in preferences?) and preferably also a standard deviation!

That way your game would not only be occupied with peak speeds, but show improvement over time as well as a more interesting estimate of your overall typing speed. After all, what good is it that I can type at, say, 80 WPM once in a blue moon if my average is, say, 55 plus minus 10 during actual typing?

I remember you making similar curves of your speed progress in QWERTY versus Colemak when you started out. How cool would it be if your games made such curves automagically?

*** Learn Colemak in 2–5 steps with Tarmak! ***
*** Check out my Big Bag of Keyboard Tricks for Win/Linux/TMK... ***

Offline
  • 0
  • Reputation: 0
  • Registered: 19-Jul-2009
  • Posts: 4

A further note regarding the typing test being biased.  I replayed the top 10 scores for 30 sec, 1 min, and 2 min and found very few difficult sequences in any of the tests.  For example, of the top 10 30 Second scores, here are the following uncommon words/sequences:

300
WONDERFUL
--
Tuxedo
nonzero
orthonormal
protocol

For the 10 Ten 1 Minute scores:

-
Toledo Ohio
?
un-American
--
!
Newark
300
!
?
Kentucky
--
?
?
wombats
tophats
?

Of course, these are spread out among the 10 different test scores, so we are talking perhaps 1-1.5 per test.  I imagine that similar numbers would hold true for other users high scores, in that it is unlikely that a user's high score will have the following text in it:

Humor in the Court:  Q:   .......?  A:

or

I/O, I/O, it's off to work we go

or

Tecumseh, (Shawnee) to his nephew Spemica Lawba 1790

or

Quark! Quark! Beware of the quantum duck!

or

It's is not, it isn't ain't, and it's it's, not it's, if you mean it is.  If you don't, it's its.  Then too, it's hers. It isn't her's. It isn't our's either. It's ours.......

The last one I have stumbled upon more than once.  Whatever lead the writing of that one is beyond me, but it is shear madness.  Thus, there are so many other things besides the heavy quotations (which seem to have come down) that tend to throw up stumbling blocks.  The worst thing is when your on a run to come upon one of them such as some sentence ending with -Samuel Goldwyn.  Not that that is particularly hard, but that when you see it coming, it acts to mentally through one off.  Other stumbling blocks I have come across are:

High level of -- and ...  (in fact it seems that almost every test I take includes at least one of either.  when making my own layout and testing on your test, I was forced to assign the - to a better location because you use it soooo much.

weird contractions such as Heav'n

Non-words such as Whaddya, and tophats(?)

Linus:  ......text here..... Snoopy: , or any other names replacing them such as Dustin Farnum: and Oliver Herford:

I/O, I/O its off to work we go.  (I see your attempts at humor, and some of the normal sentences are funny, but this one is annoying

Fortune finishes the great quotations, #12.....  (when I see this one coming up, it is disheartening and causes me to panic)

and here is a great one

What do you expect from a culture that *drives* on *parkways* and *parks* on *driveways*?  Actually though, the Asterix is pretty easy to hit, but so what, seeing it coming up is a mental derailment waiting to happen.

In summary, I really do like the test and use it almost exclusively, but my question is, if all the high scores tend to be texts that are well balanced with lots of common words and very few if any odd punctuation or letter sequences, why have so many stumbling blocks like the examples above in the test?  I wonder if I am the only one that clicks and clicks begin until the starting text that I can see looks reasonable, and also if I am the only one that stops many upon many tests midway through because up pops an absurd stumbling block that throws off my rhythm.  I would much rather type texts with say, the first 5,000 to 10,000 most common words in English, arranged in sentences that make sense, and so be assured that everytime I click begin to start a new test, I am not going to come upon the dreaded,

It's is not, it isn't ain't, and it's it's, not it's, if you mean it is.....

string of text, or something else that serves to deflate my efforts.

Otherwise, great great typing test really.  Love the new lengths, and the ability to set preferences.  How about 3 min?

Offline
  • 0
  • Reputation: 0
  • From: Australia
  • Registered: 22-Oct-2007
  • Posts: 47
kameloc said:

A further note regarding the typing test being biased.  I replayed the top 10 scores for 30 sec, 1 min, and 2 min and found very few difficult sequences in any of the tests.  For example, of the top 10 30 Second scores, here are the following uncommon words/sequences:

As you say, the words are quite rare, and this is all statistically expected (it's just the nature of the game).

"Difficulty", on the other hand, is a highly subjective matter depending on your layout and the words you are already familiar with (at your current skill level). Other people with different layouts and/or different skill levels and/or a different training program may find the same words that you find difficult to be easy, regardless of whether they are rare. It would therefore not be reasonable to delete a sentence that one person finds difficult and another person finds easy.

So, in the rest of this email, I will try to give you some ideas for how to type these words more easily in Colemak.

300
WONDERFUL
--
Tuxedo
nonzero
orthonormal
protocol

These wards are indeed rare (each occuring once in every 100,000 words), but in terms of typing difficulty in the Colemak layout, they are all about average. In other words, they are not a problem for the hands, it is only a matter of training the mind. I suggest breaking words into chunks. e.g. "Tux" "edo", and master the ability to type any chunk with ease. I love long (quasi) compound words like "orthonormal" and "wonderful" because they are usually made up of familiar components. It only takes two clicks of mental processing to prepare for a word like "orthonormal", because I view this as two chunks. I have no idea why a word like "protocol" might be difficult. But I will suggest this. Try typing that word over and over and over and over... again until you're fingers become familiar with the pattern, and you will eventually see that it presents no problem for the fingers - only for the unfamiliar mind.

For the 10 Ten 1 Minute scores:

-
Toledo Ohio
?
un-American
--
!
Newark
300
!
?
Kentucky
--
?
?
wombats
tophats
?

Once again, I personally don't find any of these words difficult to type in Colemak. Although, these are interesting examples in that names of foreign places, animals and random numbers are likely to be new words that you haven't encountered before. To find these words easy to type you do need to develop a skill in being able to type unfamiliar words. Again, I would suggest "chunking".

Of course, these are spread out among the 10 different test scores, so we are talking perhaps 1-1.5 per test.  I imagine that similar numbers would hold true for other users high scores, in that it is unlikely that a user's high score will have the following text in it:

Humor in the Court:  Q:   .......?  A:

I can imagine how difficult this might be for anyone who has not perfected their use of "shift". The way to type it is this: type "Humor in the Court" hold SHIFT ": Q:" It is even likely that the next letter will also fall under the shift meaning that the cost is only a single shift.

or

I/O, I/O, it's off to work we go

Ok, the difficulty of this is relative to your keyboard layout. I personally never accepted the idea of removing the CAPSLOCK key. If you still have one, then you can do this very easily: CAPSON "I/O, I/O, " CAPSOFF "it's off to work we go".

or

Tecumseh, (Shawnee) to his nephew Spemica Lawba 1790

The words may be unfamiliar, but that can be solved using chunking: "Te" "cum" "seh" "shaw" "nee". Hence, even if you have never seen these words before, you should eventually be able to type such unfamiliar words without stumbling.

or

Quark! Quark! Beware of the quantum duck!

This sentence has some difficult SHIFTing, so you will likely need to pause around the Q and ! while performing the shift. The only thing I can say here is that such difficulties sometimes happen. It helps to recognise what is difficult and what is not, so that you can plan to slow down and speed up (as you would around a speed hump).

or

It's is not, it isn't ain't, and it's it's, not it's, if you mean it is.  If you don't, it's its.  Then too, it's hers. It isn't her's. It isn't our's either. It's ours.......

Note that all of the words in this example are extremely frequent words. Whatever is tripping you, it would probably a good thing to practice this very example until it no longer trips you. I find this example to be a very fast one.

The last one I have stumbled upon more than once.  Whatever lead the writing of that one is beyond me, but it is shear madness.  Thus, there are so many other things besides the heavy quotations (which seem to have come down) that tend to throw up stumbling blocks.  The worst thing is when your on a run to come upon one of them such as some sentence ending with -Samuel Goldwyn.  Not that that is particularly hard, but that when you see it coming, it acts to mentally through one off.  Other stumbling blocks I have come across are:

High level of -- and ...  (in fact it seems that almost every test I take includes at least one of either.  when making my own layout and testing on your test, I was forced to assign the - to a better location because you use it soooo much.

I actually follow a non-standard practice here and type '-' with whatever finger feels best prepared at the time, sometimes my middle, ring or pinky finger. For a double '-' I use a stronger finger such as my ring or middle finger. Anyway, using that technique, I find it quite fast to type --.

weird contractions such as Heav'n

It's not that bad on Colemak. Think about it this way: it's almost like typing "Heavon" but with your pinky shifted a little bit to the right.

Non-words such as Whaddya, and tophats(?)

Chunking: "Whad" "top" "hats"

Linus:  ......text here..... Snoopy: , or any other names replacing them such as Dustin Farnum: and Oliver Herford:

"Dust" "in" (or "Dus" "tin"),  "Far" "num", etc.

Fortune finishes the great quotations, #12.....  (when I see this one coming up, it is disheartening and causes me to panic)

# leaves the left hand in the right proximity to type 12.

I am not going to come upon the dreaded,

It's is not, it isn't ain't, and it's it's, not it's, if you mean it is.....

Practice the above and you will find it is not difficult for the fingers, only the unfamiliar mind, and that can be solved with practice.

Love the new lengths, and the ability to set preferences.  How about 3 min?

Well, 3 minutes is only 50% different from 2 minutes, whereas all of the other tests are at least 100% different from each other.

Offline
  • 0
  • Reputation: 214
  • From: Viken, Norway
  • Registered: 13-Dec-2006
  • Posts: 5,362

You know what kameloc? You're taking it too seriously!

I love getting a little extra training every now and then. Because it's not every time I take the tests - just now and then. If there were no sentences with an extra helping of quotes or hyphens or whatnot, I wouldn't get the extra training I need for exactly those things. In fact, rather than an unbiasing or weakness of the test I consider it one of the strengths of higames.com because it adds variation and challenge!

Sure, if you're cruising in for an all-time personal best and towards the end hit something hard you'll be annoyed. But deal with it. It's overall for your own good, and don't get obsessed with your high score because that's not good for you. As mentioned above I'd like to see a stat or two for your more representative scores as I think that's far more interesting anyway!

*** Learn Colemak in 2–5 steps with Tarmak! ***
*** Check out my Big Bag of Keyboard Tricks for Win/Linux/TMK... ***

Offline
  • 0
  • Reputation: 0
  • From: Australia
  • Registered: 22-Oct-2007
  • Posts: 47
DreymaR said:

I remember you making similar curves of your speed progress in QWERTY versus Colemak when you started out. How cool would it be if your games made such curves automagically?

This is a great feature idea and I'd really like to implement it, but it will be a while before I have time to, I think. First, I have some other work that needs to be done over the next few months. After that, people really want me to add "multiplayer" as the next new feature. I've decided I'll start development on this as soon as the membership reaches 3000 (which will hopefully give me enough time to complete my current work).

Offline
  • 0
  • Reputation: 214
  • From: Viken, Norway
  • Registered: 13-Dec-2006
  • Posts: 5,362

It wouldn't have to be curves, although naturally that's the most 'sexy' option. If you make a little calculation or five, you could add a few lines to each player's high score page. I'd suggest one line with the player's average and standard deviation of the last 10 scores for that test, as well as the same for the last 100 or so scores. Sort of a 'current' versus 'long-term' average, if you wish. Later on you could have the player set preferential values for those two but it isn't all that important for now.

I guess you'd have to store the last n scores for each player. Only the last 100 or so score values would have to be stored (if you were to allow the 'long-term' value a max of 100), so it shouldn't take up noticeable space on your server.

*** Learn Colemak in 2–5 steps with Tarmak! ***
*** Check out my Big Bag of Keyboard Tricks for Win/Linux/TMK... ***

Offline
  • 0
  • Reputation: 214
  • From: Viken, Norway
  • Registered: 13-Dec-2006
  • Posts: 5,362

Another suggestion for the highscore list would be to show a list of recent highscores in addition to the all-time bests. This, too, along with player averages and listing the vicinity of your own best, would encourage the less-than-elite players to compete. I can never touch the top 30 at the 2-minute test; in fact, as you know that test is now so old that all the top 30 scores are in the high elite range. But I might make the day's best list - at least, until the whiz kids log on, heh! - and that'd encourage daily play. This is another thing the Key Hero test uses, and with good effect I feel.

I know that you post new personal bests on the chat and that's cool, but not the same.

*** Learn Colemak in 2–5 steps with Tarmak! ***
*** Check out my Big Bag of Keyboard Tricks for Win/Linux/TMK... ***

Offline
  • 0
  • Reputation: 0
  • Registered: 01-May-2009
  • Posts: 68

I agree with DreymaR.  You could encourage more activity by making people's achievements seem more tangible, by showing them more relative rankings.

One way to do this would be to divide up percentile ranges into ranks, A B C D and F.  You could then tell people what their rank is within their grade, and people would strive to move up.  Getting the ranks ideal would jsut take some trial and error imo, so diving in would be better than debating what the ideal ranges would be.

Offline
  • 0
  • Reputation: 214
  • From: Viken, Norway
  • Registered: 13-Dec-2006
  • Posts: 5,362

Good idea! Moving from 'C+' to 'B-' would feel like a more tangible achievement. Reminds me of Typing Of The Dead, where you had school-type marks based on various aspects of your typing.

*** Learn Colemak in 2–5 steps with Tarmak! ***
*** Check out my Big Bag of Keyboard Tricks for Win/Linux/TMK... ***

Offline
  • 0
  • Reputation: 0
  • Registered: 08-Mar-2008
  • Posts: 303
DreymaR said:

Another suggestion for the highscore list would be to show a list of recent highscores in addition to the all-time bests. This, too, along with player averages and listing the vicinity of your own best, would encourage the less-than-elite players to compete. I can never touch the top 30 at the 2-minute test; in fact, as you know that test is now so old that all the top 30 scores are in the high elite range. But I might make the day's best list - at least, until the whiz kids log on, heh! - and that'd encourage daily play. This is another thing the Key Hero test uses, and with good effect I feel.

I know that you post new personal bests on the chat and that's cool, but not the same.

Actually, if you look at some of the more popular typing tests like this one (http://freetypinggame.net/free-typing-test.asp), if you look at the high scores for just the past week (http://freetypinggame.net/score.asp?pag … n=-&date=7), the highest ones are above 150 WPM and one is even at 178. And that's just for one week.

Offline
  • 0
  • Reputation: 214
  • From: Viken, Norway
  • Registered: 13-Dec-2006
  • Posts: 5,362

You're absolutely right and I even thought of that myself. So, maybe local list in addition to the global one is the better option. (I've been playing on the PS3 network lately and was excited even to move up from position ~600,000 to ~500,000 in a Little Big Planet level!) And/or a 'school marks' system based on your score's percentile of the score population distribution. Both are fairly flashy and tangible, and wouldn't be too hard to implement I'd think.

'Congratulations - your score of ### WPM merits a B- ranking! The grand result for the last ## tests is ### WPM and a C+ ranking.'  :)

*** Learn Colemak in 2–5 steps with Tarmak! ***
*** Check out my Big Bag of Keyboard Tricks for Win/Linux/TMK... ***

Offline
  • 0