Comparison matrix for keyboard layout proposals


Even if we assume your argument to be true, if you deduct "most" from the "normal non-nerd people" they outnumber the "not normal nerdy people" by orders of magnitude. Selling devices in the order of thousands becomes a lot easier if there arent hurdles to overcome. Readability is an issue fixed with higher pixel density and bigger screen, that is already much better.

When you put Select and Ctrl on the same 3x7mm button, you have transferred that problem to the keyboard. There is no neccecity in that advanced users have good eye-sight.

I dont know where im placed, I have excellent vision, but you already lost me with the complexity. Please consult a bigger usergroup, and you will get the same response as I and Eight-bit have gotten. From my point of view, needless complexity is universially bad. It is no excuse that advanced people are able to use it, and it is no point in arguing that normal people wont buy the device.

How do you know that there arent any normal people who bought the pandora? Im fairly confident by making it more complex you are scaring away some of the current userbase. I am sure that this prospect is a good thing to avoid.
Business realities allowing, I'm against pandering to the lowest common denominator. Pick a focus group or few and optimize for them. Who do you think will be the main audience for pyra? I don't think it has much intersection with the group of people that are scared by different looking keyboards.
Making something good for everyone is great for no one.
 
Last edited by a moderator:
Even if we assume your argument to be true, if you deduct "most" from the "normal non-nerd people" they outnumber the "not normal nerdy people" by orders of magnitude. Selling devices in the order of thousands becomes a lot easier if there arent hurdles to overcome.  Readability is an issue fixed with higher pixel density and bigger screen, that is already much better.

When you put Select and Ctrl on the same 3x7mm button, you have transferred that problem to the keyboard. There is no neccecity in that advanced users have good eye-sight.

I dont know where im placed, I have excellent vision, but you already lost me with the complexity. Please consult a bigger usergroup, and you will get the same response as I and Eight-bit have gotten. From my point of view, needless complexity is universially bad. It is no excuse that advanced people are able to use it, and it is no point in arguing that normal people wont buy the device.

How do you know that there arent any normal people who bought the pandora? Im fairly confident by making it more complex you are scaring away some of the current userbase. I am sure that this prospect is a good thing to avoid.
So what is your proposed solution? This?

post-2326-0-49684100-1420989715.png


Sure, you don't have "Select and Ctrl" on the same small button. Instead you have some kind of obscure hieroglyphs that seem to say something along the lines of "better run away, F-ING newbie!" ;)
 
I changed it just now, sorry for not updating. I keep things up for consideration a while before i commit.

Edit: B-zar: Regular people can use a regular keyboard too, what is your actual point?
 
Last edited by a moderator:
I'm not against international support, but I don't like the way the designed-by-commitee layout is littering arbitrary labels all over the keys (IMO worse than on the Pandora). Yes, it supports a lot of languages (if the user stares long enough at the labels to find the key) and for every individual placement there is a perfectly valid reason. But in the end there is no theme beyond the basic US-QWERTY layout (unlike in Grench's layout with the embedded numeric keypad and other goodies where additional labeling makes perfect sense), therefore the labels are just confusing (as confirmed by several people in these threads) and to date there is zero evidence that these labels will actually enable more fluent entry of non English text (compared to a language specific layout that uses unlabeled mappings). In the end I don't care that much about the actual placement and mapping of keys (though I prefer the P+m/L+n variant with as few modifiers as possible on the keyboard itself), but if it is basically just a QWERTY layout than it simply shouldn't have labels for things that aren't on a US keyboard. My weights in the matrix try to reflect that.


edit: I think instead of taking one arbitrary variant of a US keyboard adaptated to the limited Pyra keyboard and adding to it, we should first explore the possible variants of e.g. English, French, German, Spanish layouts adaptations separately (as if we could get separate keymats) and only then try to find the best way to combine them; and if it isn't possible to create a merged one where we can reasonably assume that it actually is both an improvement for non English input and not a detriment to English input, we have at least created a solid English layout that is significantly improved beyond the Pandora's; The language poll is a good indication that the best possible English support is most important; I'd say that adding German is already just a bonus (even though it is the second largest group of native users by a large margin and has been requested by ED)
 
Last edited by a moderator:
I'm not against international support, but I don't like the way the designed-by-commitee layout is littering arbitrary labels all over the keys (IMO worse than on the Pandora). Yes, it supports a lot of languages (if the user stares long enough at the labels to find the key) and for every individual placement there is a perfectly valid reason. But in the end there is no theme beyond the basic US-QWERTY layout (unlike in Grench's layout with the embedded numeric keypad and other goodies where additional labeling makes perfect sense), therefore the labels are just confusing (as confirmed by several people in these threads) and to date there is zero evidence that these labels will actually enable more fluent entry of non English text (compared to a language specific layout that uses unlabeled mappings). In the end I don't care that much about the actual placement and mapping of keys (though I prefer the P+m/L+n variant with as few modifiers as possible on the keyboard itself), but if it is basically just a QWERTY layout than it simply shouldn't have labels for things that aren't on a US keyboard. My weights in the matrix try to reflect that.


edit: I think instead of taking one arbitrary variant of a US keyboard adaptated to the limited Pyra keyboard and adding to it, we should first explore the possible variants of e.g. English, French, German, Spanish layouts adaptations separately (as if we could get separate keymats) and only then try to find the best way to combine them; and if it isn't possible to create a merged one where we can reasonably assume that it actually is both an improvement for non English input and not a detriment to English input, we have at least created a solid English layout that is significantly improved beyond the Pandora's; The language poll is a good indication that the best possible English support is most important; I'd say that adding German is already just a bonus (even though it is the second largest group of native users by a large margin and has been requested by ED)
The Pandora's non-US-QWERTY symbols are indeed very arbitrary: it has ¥ € £ §  ´

The only extra language you can type with that is Irish (thanks to the acute accent, though I don't think it even works).

Currency symbols and the section sign, there's not much you can do with that.

By contrast, something like this:

http://www.keyboard-layout-editor.com/#/layouts/d7e5a0dab3f707566068e641e29c84d9

does not have "arbitrary labels".  There is nothing arbitrary about this:

  • it is based on a standard US-QWERTY keyboard,
  • with the addition of all the letters and symbols you find on a standard German QWERTZ keyboard: ÜÖÄ ß § ° € circumflex acute grave.
  • Then the essential missing keys (for typing French) from a standard French AZERTY keyboard are added: ÉÈÀÇ and diaeresis  ¨ (Ù can be safely omitted because grave is already there and ù is only used in one French word) -- circumflex is already there.
  • Then the essential missing key (for typing Spanish) from a standard Spanish keyboard is added: ñ -- acute is already there.
And that's it, really. Italian, Portuguese, Dutch and Finnish you get for free if you have the above.

So there's nothing "arbitrary" about this. It's just a matter of priorities. First English, then German, then French, then Spanish, then all the rest.

Sure, we could make separate keymats for QWERTY, QWERTZ, AZERTY etc, and that would be even better for some people. But a nice property of a multilingual keyboard is that it can be used for several languages at the same time, which means it is strictly better than single-language layouts for people who speak multiple languages, e.g. in Belgium or Switzerland. Also in terms of logistics, shipping and ordering complexity, and just overall price tag, one multilingual layout is better than several language-specific keymats.
 
B-zar: Regular people can use a regular keyboard too, what is your actual point?
They can use it because of they're used to it. They had to first invest time and effort to learn it. No person knows from birth what a shift key is and why it makes letters big but turns numbers into entirely other symbols. We cannot have the keyboard they're used to because of physical limitations. There is going to be a new learning curve no matter what. The point is do we try to mimic the keyboard we can't have to reap at least some partial benefit from familiarity or look at what we think the keyboard will be used for and modify that keyboard to better suit those purposes. One option makes for a lower learning curve that may make the device more accessible, while the other provides a better experience with prolonged use.

I'm saying I value early accessibility less than eventual performance. Maybe that's why I use and vi-mode in editors.

This is all, of course, orthogonal to other aspects such as software compatibility.
 
One option makes for a lower learning curve that may make the device more accessible, while the other provides a better experience with prolonged use.
The question is: which is the one with the lower learning curve for a German?

  • A keyboard that has [ < Shift in the spots where a German would expect to find Ü Ö Ä, where you can remap [ < Shift to an unlabeled Ü Ö Ä, live without keyboard Shift and map the actual [ < symbols to AltGr+Ü Ö, in order to get something that somewhat approximates a normal German keyboard except Z and Y are swapped and the number row symbols are messed up. ß could perhaps be mapped to AltGr+Backspace, or maybe to ].
  • A keyboard that has Ü Ö Ä ß labeled as Meta-keys.
Comradekingu will claim that the first option has a lower learning curve, but I don't really think so.
 
The second option has a higher learning curve, and is _far less efficient_.

I am a proponent of making keymats for the big regions, which would reap the benefit of familiarity, function, and habit. When there is one default, i'd rather it worked for advanced users.
Because those are the users we got. As for the users we can get, making it look normal is key.
 

B-zar: Regular people can use a regular keyboard too, what is your actual point?
They can use it because of they're used to it. They had to first invest time and effort to learn it. No person knows from birth what a shift key is and why it makes letters big but turns numbers into entirely other symbols. We cannot have the keyboard they're used to because of physical limitations. There is going to be a new learning curve no matter what. The point is do we try to mimic the keyboard we can't have to reap at least some partial benefit from familiarity or look at what we think the keyboard will be used for and modify that keyboard to better suit those purposes. One option makes for a lower learning curve that may make the device more accessible, while the other provides a better experience with prolonged use.
 
I'm saying I value early accessibility less than eventual performance. Maybe that's why I use and vi-mode in editors.
 
This is all, of course, orthogonal to other aspects such as software compatibility.
I wasnt talking about people who have never seen a keyboard before, im talking about the ones who have. Even the ones who havent will have the benefit of learning something that exists other places, rather than something else.
 
That there is a learning curve isnt a reason to make it a steep one. We should isntead acknowledge it and try to make the treshold of entry as low as possible. ED seems to hold this position too
 

Okay, the plastic inlay for one key is 3x7mm big.
 
What I don't want is hundreds of keyboard labels on one keyboard that 99% of the users won't understand.
It easily scares away new, interested customers.
 
I'd rather do additional layouts if demand is there (demand = at least 500 users who demand that one keyboard layout).
 
The maximum labels per key is 3: One master, one with shift, one with Meta.
No keyboard/typewriter has ever done away with any reapable efficiency, since the 1870s this has been true. Yet it would have been far easier to shoehorn everyone into the same layout, the reason why this doesnt happen, is because this is not what people want.
 
Last edited by a moderator:
I think that conflates orthogonal issues somewhat. The learning curve steepness/length is a separate issue from the multilingual one. ED presented an opinion in that quote about what labels should be on the keyboard. It does not address the symbols available in the layout nor how the learning curve should be weighted in relation to other qualities. The bit about new customers is, as I read it, more about the keyboard not looking too cluttered at first sight, not its use. I have no strong feelings about symbol clutter, but it is not the same as having those symbols in the layout, albeit unprinted.

A steeper learning curve may be balanced by other qualities. Some of which I personally hold to a greater value. But this is of course personal preference.
 
The question is: which is the one with the lower learning curve for a German?

  • A keyboard that has [ < Shift in the spots where a German would expect to find Ü Ö Ä, where you can remap [ < Shift to an unlabeled Ü Ö Ä, live without keyboard Shift and map the actual [ < symbols to AltGr+Ü Ö, in order to get something that somewhat approximates a normal German keyboard except Z and Y are swapped and the number row symbols are messed up. ß could perhaps be mapped to AltGr+Backspace, or maybe to ].
  • A keyboard that has Ü Ö Ä ß labeled as Meta-keys.
Comradekingu will claim that the first option has a lower learning curve, but I don't really think so.
The second option has a higher learning curve, and is _far less efficient_.
Seriously, comradekingu?

You really think it is easier to learn that the keys labeled [ < Shift ] are actually üöäß, than it is to learn that the keys labeled üöäß are üöäß ?

About efficiency: yes, dedicated primary keys are a bit more efficient than Meta-keys. The frequencies of üäßö in German are 0.65%, 0.54%, 0.37% and 0.30% respectively (source). That means they're about an order of magnitude less common than most of the other letters (ENIRSTAHDULCG each have a frequency of over 3% in German).

Without counting space, punctuation symbols and other keypresses, that means you are talking about the efficiency of less than 2% of the letters, while you sacrifice keyboard Shift and keyboard Meta/AltGr/Fn, you have Enter and Tab in a weird position and have dedicated [ ] < (instead of more useful punctuation symbols like ' - /) just to make room for the dedicated unlabeled üöäß.
 
About efficiency: yes, dedicated primary keys are a bit more efficient than Meta-keys. The frequencies of üäßö in German are 0.65%, 0.54%, 0.37% and 0.30% respectively (source). That means they're about an order of magnitude less common than most of the other letters (ENIRSTAHDULCG each have a frequency of over 3% in German).

Without counting space, punctuation symbols and other keypresses, that means you are talking about the efficiency of less than 2% of the letters, while you sacrifice keyboard Shift and keyboard Meta/AltGr/Fn, you have Enter and Tab in a weird position and have dedicated [ ] < (instead of more useful punctuation symbols like ' - /) just to make room for the dedicated unlabeled üöäß.
To be fair, that may hold in german but for example in finnish "ä" is quite a common letter.
That said, I personally expect to write finnish on my pyra only maybe about 10% of the time (mostly for irc), so it's not that big a deal for me to have it under meta, even unlabeled.

I do understand others may have differing use cases and preferences, but those are mine.
 
Last edited by a moderator:
A steeper learning curve may be balanced by other qualities. Some of which I personally hold to a greater value. But this is of course personal preference.
There are also things you can do to alleviate a steep learning curve.  e.g. You could provide documentation on the device highlighting where to find the symbols most commonly used for given languages. 

Seriously, comradekingu?

You really think it is easier to learn that the keys labeled [ < Shift ] are actually üöäß, than it is to learn that the keys labeled üöäß are üöäß ?
Don't forget about the learning curve involved in learning how to remap those keys in the first place.  That would be a factor there too. :)

- Neelix
 
Last edited by a moderator:
About efficiency: yes, dedicated primary keys are a bit more efficient than Meta-keys. The frequencies of üäßö in German are 0.65%, 0.54%, 0.37% and 0.30% respectively (source). That means they're about an order of magnitude less common than most of the other letters (ENIRSTAHDULCG each have a frequency of over 3% in German).


Without counting space, punctuation symbols and other keypresses, that means you are talking about the efficiency of less than 2% of the letters, while you sacrifice keyboard Shift and keyboard Meta/AltGr/Fn, you have Enter and Tab in a weird position and have dedicated [ ] < (instead of more useful punctuation symbols like ' - /) just to make room for the dedicated unlabeled üöäß.
To be fair, that may hold in german but for example in finnish "ä" is quite a common letter.

That said, I personally expect to write finnish on my pyra only maybe about 10% of the time (mostly for irc), so it's not that big a deal for me to have it under meta, even unlabeled.


I do understand others may have differing use cases and preferences, but those are mine.
Yes, in Finnish the ä has a frequency of 5.21%, so it is a very important letter.

In French, é has a frequency of 2.13% so it is pretty important too.

I think it makes sense to label üöäß, but I don't think it's necessary to have them as dedicated primary keys.
 
The question is: which is the one with the lower learning curve for a German?

  • A keyboard that has [ < Shift in the spots where a German would expect to find Ü Ö Ä, where you can remap [ < Shift to an unlabeled Ü Ö Ä, live without keyboard Shift and map the actual [ < symbols to AltGr+Ü Ö, in order to get something that somewhat approximates a normal German keyboard except Z and Y are swapped and the number row symbols are messed up. ß could perhaps be mapped to AltGr+Backspace, or maybe to ].
  • A keyboard that has Ü Ö Ä ß labeled as Meta-keys.
Comradekingu will claim that the first option has a lower learning curve, but I don't really think so.
The second option has a higher learning curve, and is _far less efficient_.
Seriously, comradekingu?

You really think it is easier to learn that the keys labeled [ < Shift ] are actually üöäß, than it is to learn that the keys labeled üöäß are üöäß ?

About efficiency: yes, dedicated primary keys are a bit more efficient than Meta-keys. The frequencies of üäßö in German are 0.65%, 0.54%, 0.37% and 0.30% respectively (source). That means they're about an order of magnitude less common than most of the other letters (ENIRSTAHDULCG each have a frequency of over 3% in German).

Without counting space, punctuation symbols and other keypresses, that means you are talking about the efficiency of less than 2% of the letters, while you sacrifice keyboard Shift and keyboard Meta/AltGr/Fn, you have Enter and Tab in a weird position and have dedicated [ ] < (instead of more useful punctuation symbols like ' - /) just to make room for the dedicated unlabeled üöäß.
Yes, because one of them starts with "I didnt buy a device". Unless you differentiate between advanced users, and novice users, this deals with the users we got, and the minimum amount of keymat default that can be employed.

You dont learn where üöäß are, you know it already. That is an immediate benefit. Nobody forgets about it, so we need the easiest way to facilitate it.

Im not inventing anything new, just acknowledging what is.

Dedicated keys are a alot more efficient. Stopping to learn a new way to type 1. isnt going to happen, and 2. is a lot worse than the frequency dicatates.

And that is the compromise, if you dont want it, make the dedicated buttons into primarily ' - and /  and then use shift and AltGr to type what was dedicated before, not any harder than that. How much less of a compromise can it be and still be a compromise?
 
Last edited by a moderator:
The question is: which is the one with the lower learning curve for a German?

  • A keyboard that has [ < Shift in the spots where a German would expect to find Ü Ö Ä, where you can remap [ < Shift to an unlabeled Ü Ö Ä, live without keyboard Shift and map the actual [ < symbols to AltGr+Ü Ö, in order to get something that somewhat approximates a normal German keyboard except Z and Y are swapped and the number row symbols are messed up. ß could perhaps be mapped to AltGr+Backspace, or maybe to ].
  • A keyboard that has Ü Ö Ä ß labeled as Meta-keys.
Comradekingu will claim that the first option has a lower learning curve, but I don't really think so.
The second option has a higher learning curve, and is _far less efficient_.
Seriously, comradekingu?

You really think it is easier to learn that the keys labeled [ < Shift ] are actually üöäß, than it is to learn that the keys labeled üöäß are üöäß ?

About efficiency: yes, dedicated primary keys are a bit more efficient than Meta-keys. The frequencies of üäßö in German are 0.65%, 0.54%, 0.37% and 0.30% respectively (source). That means they're about an order of magnitude less common than most of the other letters (ENIRSTAHDULCG each have a frequency of over 3% in German).

Without counting space, punctuation symbols and other keypresses, that means you are talking about the efficiency of less than 2% of the letters, while you sacrifice keyboard Shift and keyboard Meta/AltGr/Fn, you have Enter and Tab in a weird position and have dedicated [ ] < (instead of more useful punctuation symbols like ' - /) just to make room for the dedicated unlabeled üöäß.
Yes, because one of them starts with "I didnt buy a device". You dont learn where üöäß are, you know it already. Im not inventing anything new, just acknowledging what is.

Dedicated keys are a alot more efficient. Stopping to learn a new way to type 1. isnt going to happen, and 2. is a lot worse than the frequency dicatates.

And that is the compromise, if you dont want it, make the dedicated buttons into primarily ' - and /  and then use shift and AltGr to type what was dedicated before, not any harder than that. How much less of a compromise can it be and still be a compromise?
If keys that do different things than what their label says is no issue (like you imply), then you can always just completely remap my layout proposal to your favorite layout and be happy :)

I am not a potential German customer who just learned about the Pyra and who is considering whether or not to buy it. Neither are you. You seem to know exactly what effect keyboard layout proposals have on potential German customers. You think all Germans will look at your layout and say: "hmm, yes, maybe P+2, L+2 could be remapped to üöäß, so I don't have to learn anything new, this device is perfect for me!"

I think potential German customers might also think "this is a QWERTY layout, I can't type üöäß on it so I'll have to live with that and type ue oe ae ss or use Compose sequences, that is a bit annoying".

By contrast, when they look at a layout which has explicit labels for üöäßéèàçñ, some will be intimidated a bit and maybe think "oh no, it has French stuff, bah! ugly!", others will think "this seems to be a multilingual keyboard, let's see if I could type German on it... [pauses to look for those letters]  ... ah yes, nice, this thing was designed with the German language in mind!", and others will think yet something else, perhaps "this is horribly inefficient, it does not have dedicated üöäß keys so I will type 1.5% slower than normal on this!". I don't know how an "average" German would react.
 
I really want to avoid sounding elitist but we seriously need to consider: if a few extra labels are enough to scare someone away from not purchasing it, is that really the kind of person that would be happy with a Pyra? I've been saying since its inception that the Pandora isn't for everybody, a lot of people complain about a lot of things about it and can't understand why anyone would want one, but for those that do own one they generally know why they wanted it, and none of us were turned off by the weird keyboard layout: the fact that it had a keyboard at all was a selling point. The Pandora isn't for everybody, the Pyra is not going to be for everybody.

If the person is so afraid of a few extra symbols, how will they handle a different interface? If they were considering purchasing but decided against it because of these extra symbols then what did they really want to buy it for? If a few extra symbols is enough to confuse someone because they aren't normally found on "their" keyboard then why doesn't the fact that it has a necessary Fnaltgrmetamod key equally confusing?
 
Any amount of noisy labels is going to scare people away, because it gives a bad impression of the device. And its not irrelevant to fairly advanced people.

The people speaking those languages, and i do, dont think you had them in mind, they think its a toy keyboard, made with LSD in mind. Fairly reminissent of scientific calculator.

You cant seperate language from keyboard. So says the current pandora users. Thats _with_ the bias of already excluding everyone who isnt a monolingual english-speaker.

The pandora concerns were all legitimate. Being a nieche can happen for the right and for the wrong reasons.

The screen is fixed, the case is fixed, the OS is fixed, all it needs is a not-broken keyboard for half the audience.

People are very ok with new interfaces, the most used OS in the world is Android. People are however never ok with gimped keyboards/typewriters. That never changes.

International users consider devices that have their full alphabet, and a way to type their language. The pyra is one of those devices. Yes it will sell a lot of devices because its special, but lets not ride on that.

Lets say a keymat costs 10 €, with 500 takers thats 5k €, ill gladly pay that to be able to not have gimped keyboard.
 
Last edited by a moderator:
People are very ok with new interfaces, the most used OS in the world is Android. People are however never ok with gimped keyboards/typewriters. That never changes.
It's funny you say people are "very ok with new interfaces" beause a keyboard is an interface.  Also you mention the most used OS in the world is Android, that keyboard is not even close to any real keyboard I have ever seen, and yet people seem perfectly capable and willing to learn how to type with it.
 
An Android touch screen keyboard in Swedish has åöä, but a German one has üöäß only at long-press uoas.

The French one has éèàçùâêîôûœäëïöü only behind long-press (àçéîôù are the default choices behind aceiou).

The Italian one has à3897 as defaults behind aeiou, so the Android people don't consider the graves as very important in Italian.

The Spanish one has a direct ñ, and á3897 behind long-press aeiou, with éíóú as secondary long-press.

The Portuguese has no direct ç, and á3897 like Spanish, with ãéíóú as secondary choices and õ even farther away.
 
Back
Top