Research on keys and keyboards

[Snarfangel]

I was surfing around the net, and I ran into something interesting I hadn't seen before from "The Handbook of Computer-Human Interaction" second edition. In chapter 54, titled "Keys and Keyboards," it give a ton of information on everything from key layout, keyboard slope, key force, and even chorded keyboards. Quite an interesting bit of information all in one place. Here is the link in context:
http://books.google.com/books?id=WuQbERgXR10C&pg=PA1285&dq=handbook+of+human-computer+interaction+keys+and+keyboards+chapter+54&hl=en&sa=X&ei=n4z5UrP0CMvNqwH434C4Aw&ved=0CDsQ6AEwAA#v=onepage&q=handbook%20of%20human-computer%20interaction%20keys%20and%20keyboards%20chapter%2054&f=false

And the particular chapter by itself: http://drjim.0catch.com/1997_KeysAndKeyboards.pdf

Anyway, I thought it was an interesting bit of information, despite being almost two decades old. I thought I would mention it in case someone else was interested on how different keys sizes, shapes, and formats were decided upon.

/Someone on Geekhack probably has mentioned this book at some point in the past, but hey, it's a fun read.

[Proword]

Interesting that although there was mention on page 1306 of the Kinesis keyboard (1991), there was nothing I could see referring to Lilian Malt, Stephen Hobday or the Maltron keyboard, which design underpins the Kinesis to a large extent, even though Malt's work was published in the mid-late '70s, and the first Maltron keyboard was produced in circa 1977.  I bought my first Maltron in 1986. 

Perhaps this was rectified in a later edition.

[Gid]

Thanks for mentioning this book.  It was nice of the keyboard chapter's author to share his work online.

[kod]

Interesting read, thanks for the link.  The part about slope is especially curious by today's standards.  No discussion of negative slope, recommended maximum of 35 degrees.

[Snarfangel]

Thanks for mentioning this book.  It was nice of the keyboard chapter's author to share his work online.

I actually emailed him last night to thank him for posting it, and he seemed surprised but appreciative for the interest. I wonder how many people read fun things like this and never bother telling him? (He has several papers on the site I haven't checked out yet. But soon!)

His signature shows a Ph.D. and a title of "Senior Human Factors Engineer" at IBM Software Group. Smart guy!

[jacobolus]

Thanks for the link!

Lots of the sections are less useful than they could be, because the body of literature they draw on doesn’t really study many important questions in depth, study designs are compromised or miss important confounding factors, sample sizes are tiny, etc. But it’s really great to have a comprehensive review like this; it gives a good idea of what papers are out there, and what conclusions can be drawn from those.

It would be really nice to see some larger and more methodologically sound studies of factors like: keyboard shape, keyboard and display positioning and chair/desk height and shape, key layout, keycap size and shape, keyswitch weight, keyswitch force curve, key travel distance, volume & type & timing of auditory feedback, etc., including sufficient time for training on the keyboard variants tested, and including groups such as programmers who need to type many special symbols and chorded keyboard shortcuts, and people with various hand sizes.

The literature cited there about optimal key travel and force, for example, leaves so much room in the “recommended range” to be almost useless.

In the absence of any kind of solid data, what we’re left with around here is lots of one-off experiments, anecdotes, and personal preferences. [But hey, that’s better than nothing!]

[jacobolus]

I like how the “bigass enter” key is called the “dogleg enter”. Also note: support for the claim that the ISO enter is strictly inferior to the ANSI or bigass variants. Muahaha.

[Snarfangel]

I like how the “bigass enter” key is called the “dogleg enter”. Also note: support for the claim that the ISO enter is strictly inferior to the ANSI or bigass variants. Muahaha.

If a scientific paper used the term "bigass enter key," I would quote it at every opportunity.

[Gid]

Section 54.2.4 mentions a "half-QWERTY" mirror layout designed for one-handed typing.  It sounds like an interesting concept (and at least one person has used this layout).

The part about slope is especially curious by today's standards.  No discussion of negative slope, recommended maximum of 35 degrees.

I thought the same thing.  It makes me wonder when common perceptions concerning slope started to change.  It would also have been nice to see some research concerning posture factors--forearm angle, sitting versus standing... Still, as a history piece, this is actually a really fun (not to mention authoritative) read.

I actually emailed him last night to thank him for posting it, and he seemed surprised but appreciative for the interest. I wonder how many people read fun things like this and never bother telling him? (He has several papers on the site I haven't checked out yet. But soon!)

I think I'll shoot him an e-mail, as well.

[kod]

Reading more, some of the key switch weights in that paper seem really high, unless they're measuring them a different way.

"Back in MY day we had to type with rocks in our pockets just to get the keys to depress, uphill, both ways, in the snow..."

[jacobolus]

One interesting note about that paper. When it was published there weren’t too many good studies about split “ergonomic” keyboard designs – they just had people try them for a bit and then asked about discomfort and preference.

But in the time since, there have been a bunch more studies that tried doing physical measurements of wrist strain, and following people for several months of using split keyboards, and have much more convincing evidence that a split design makes a big improvement in comfort for people who type several hours per day.

For example there are some good papers linked from http://ergo.berkeley.edu, or just do a google scholar search for 'split keyboard'.

[Unfortunately many of these studies still seem to just compare rubber dome keyboards.]

[Proword]

Have you read these?

http://www.maltron.com/keyboard-info/academic-papers

These date back to the mid-70s through to 1994.

[jacobolus]

Proword, yes, I have read those. You’ll note they have tiny sample sizes, mainly rely on anecdotal reports by a self-selected group of test subjects, and include basically no hard data, other than self-reported typing speeds. There’s also some theoretical analysis of finger load and speed of typing various digraphs, explaining the reasoning behind the Maltron design. I’m glad that Maltron published these testimonials and explanations, but they’re not really anything that could be called “science”, which is probably why they aren’t discussed in the paper linked above.

Jim Lewis, the IBM researcher author of the original paper linked here, wrote a more extended literature review in 1993 – http://drjim.0catch.com/1994_ACriticalLiteratureReviewOfHumanFactorsStudiesOfSplitKeyboardsFrom1926To1993.pdf

He mentions that he couldn’t track down a copy of an unpublished 1983 Malt paper that supposedly had some experimental results [I can’t find it either], so couldn’t evaluate it.

Here’s what he says about the Hobday 1988 paper: “This article makes no substantial contribution to the split-keyboard literature. The claims for the Maltron keyboard are impressive, but the author does not provide any experimental details or citations to support the claims.” I tend to agree.

Here is a nice explanation of how to critically examine such papers, from the beginning of that linked paper:

Key issues to look for when evaluating individual articles about split keyboards are:

* Do the claims in the article have any experimental support? Without clear experimental support, a claim is worthless. Scientists do not accept testimonials (quotations from users saying they like a product) as evidence.

* If the claims have an experimental basis, is there an adequate description of the experiment to allow critical evaluation of the experiment’s adequacy?

* Did the experimenters use proper controls, or was the design confounded? In a confounded design’s experimental conditions, two variables change at the same time, making it impossible to know which variable was responsible for any measured changes between the conditions.

* Does the article contain inappropriate or flawed analyses? Do the numbers in the tables add up correctly? Are the statistical tests appropriate? Are the results statistically significant?

* Does the article contain any conclusions that do not follow from the data?

* Are there other reasonable alternative interpretations of the results?

* Do the results support any claims regarding the superiority of split keyboards relative to standard keyboards? Some types of superiority might be faster typing speed, fewer errors, more comfortable typing, relief of CTD symptoms, or lower probability of the onset of CTD.

One more important note from that article is that most of the studies which found advantages in split keyboards were directly sponsored by inventors of split keyboards, using self-selected test subjects. It’s quite obvious that such a study design could lead to inaccurate reports from users/customers consciously or unconsciously trying to say something nice to a keyboard’s inventor, or rationalizing an expensive purchase, or experiencing a placebo effect, or changing other aspects of the workspace alongside the keyboard, etc. etc.

The Maltron keyboard may be the best thing in the world, but I haven’t seen any scientific study which convincingly demonstrates that it has any advantage. Note, personally I think the Maltron keyboard is pretty great, though not perfect. But that’s just based on personal preferences and intuitions, not any scientific evidence.

[Proword]

My initial post was simply remarking that the  Kinesis was mentioned, but not the Maltron.  And as far as I can determine there is even less primary documentation supporting it. 

If you wish to obtain Malt's 1983 paper (which I don't recall seeing) you could contact Stephen Hobday via Maltron.  He's still around, I received a Christmas card from him in December. 

[Proword]

Lilian Malt's 1977 paper Keyboard Design in the Electronic Era

http://www.maltron.com/keyboard-info/academic-papers/236-lillian-malt-papers

was presented to a conference of the Printing Industry Research Association

https://www.smitherspira.com/About-us.aspx

Perhaps they may have some more of Malt's papers.

[jacobolus]

I’m not too worried about finding Malt’s other unpublished paper. Again, the 1977 paper and the later Hobday papers are interesting, and I’m glad they are available, but they don’t contain much scientific evidence.