Interesting question. One that I have wondered about too. Maybe Speedmorph can run some tests?

I remember reading somewhere that Qwerty scored higher than the alphabetical layout, but I can't remember where now...

Perhaps http://millikeys.sourceforge.net/misc/why-qwerty.pdf ? (it's on Colemak's media coverage page)

He unfairly leaves the semicolon on the home position instead of placing T there:

ABCD EF GHIJ
 KLMN OP QRST
  ;UVW XY Z,./

It's still not great, but it's a little better.

I wouldn't think so; sounds more like an academic interest.  :)

However, there are people who seriously think the alphabetic layout is a good alternative to QWERTY!
http://abckeyboard.co.uk/
http://www.newstandardkeyboards.com/index.html
http://www.kidtech.com/

This tells me that a) some people have got it all ass-forward!, and b) Jag's question is of uncanny interest. Is the QWERTY really not much better than this, at least?

but that's exactly what these guys do!!!
http://www.kidtech.com/

what these guys have done is like you suggest and looking at how they position the keys on the board it makes a bit more sense
http://abckeyboard.co.uk/

but they must have been smoking something when they wrote the marketing hype!

if there are actually companies out there trying to market these boards it makes you wonder why someone has not yet made a go at a Colemak keyboard.

vVv: David Piepgrass (the author in question) refers to the ABCkeyboard and not an arbitrary alphabetic layout, as is easily seen on page 14 of his article which I suppose is the passage you're referring to.

I see that he is a bit unclear on whether QWERTY is better than any alphabet layout or not, but this has to be seen in context with the first paragraph of that section:

He is in reality addressing the claim that QWERTY is more or less the worst you can get unless you actually strive for badness, which isn't quite fair.

So if he's being unfair to the alphabetic layout in general, it would only be in terms of not delving deeper into its possibilities than what is already physically implemented. Maybe he didn't consider it to be within the scope of his article, or maybe it didn't occur to him to expound on that particular topic. I think it interesting enough to use that ABCkeyboard as an example, because it is actually implemented and marketed in that very horrible state!

I too think that the PLUM would be one of the easier layouts to learn for a keyboard naïve student. But I wouldn't use it myself, and furthermore I don't think that such naïve learners exist in significant numbers in these days when you see QWERTY everywhere and not just on actual keyboards either! I just saw a coffee mug decorated with QWERTY as a purely graphical element. Maybe there are white spots on the map that haven't been overexposed to QWERTY yet, but even those must be dwindling by now.  :)

Then there's Dexen deVries' fun 'fox' layout and a few others of the same kind...  ;)

quick,brown[
f0x.jvmpd;
thelazy]*g

According to this website, the alphabet layout has less horizontal(?) Finger Travel Distance and more %home keys.
But what does 'Overall Effort' mean?

Surely if the alphabetic layout has more percentage of words on the home keys and less Finger Travel Distance (only the horizontal?), that makes it better than Qwerty?

http://www.siteuri.ro/dvorak/SampleText … pedia.html

What do you mean by a random layout?
What would a random layout look like?

Thanks for those tests, SpeedMorph!

But how can the Alphabet layout be slightly better than the QWERTY if it doubles the same-finger ratio (which is already high in QWERTY)? I'd think that breaks a layout worse than a 19% higher travel distance?

In the www.newstandardkeyboards.com keyboard there is A, E, I at the poor left pinky! Ok, the right side seems quite good, but anyway, ergonomicaly it is not better than QWERTY at all.

According to my program, Alphabetic is substantially worse than than QWERTY.  Alphabetic's main downfall is its phenomenally poor hand balance.  For instance, on my corpus, the left index finger does roughly 32% of all typing, while the right index finger does 8%.

As far as QWERTY slowing typists who were using Alphabetic, I imagine it would do that for hunt-and-peckers; an alphabetically ordered arrangement is easier to pick letters out of than QWERTY's comparatively random arrangement.

Oh, and in my version of Alphabetic, I stuck all the punctuation at the end:

abcde fghij
klmno pqrst
uvwxy z,.;/

Here are some evaluations of interest:

Alphabetic: 0.91
QWERTY: 0.68
Dvorak: 0.35
Colemak: 0.26
My Best: 0.13

The score is a composite based on key preference, finger preference, finger balance, two-row jumps (jumps over the home row), and same-finger.  Lower scores are better.  Sometime I'll get around to writing about what all this means.  (Yeah, same-finger is fairly obvious, although in my implementation it persists across hand swaps, so, e.g., "RJF" on QWERTY would count as same-finger.)

Off the subject, but according to my scoring function, it's possible to beat Dvorak just by swapping J-N and K-O on QWERTY:

qwert yuikp
asdfg hnol;
zxcvb jm,./

That scores 0.35 (more precisely, 0.348 to Dvorak's 0.354).  I'm not sure how many people would agree that's superior to Dvorak.  I should mention that my scoring function is NOT anti-Dvorak.  In fact, if I optimize for two-row jumps and ignore everything else, I get layouts like this:

cvbjk ;zqx/
auoie ntsrd
fmpwg y,h.l

That home row looks awfully familiar, doesn't it?  I was really surprised the first time I saw that.  Most people attribute Dvorak's home row to maximizing hand swapping and home row usage, so I was quite surprised to see it rediscovered totally outside the context of either of those things!

Amusingly, that layout still scores 0.350--better than Dvorak itself.  We can manually rearrange the columns into a more sensible arrangement and hit 0.27:

jcbkv /z;qx
iaoeu dtnsr
wfpgm l,yh.

ANYWAY....