I have no hard data to support the claims that I am presenting
below. This document is merely the first step in designing the
experiments needed to test these hypothesis. I'm creating this
document to help clarify my views so that I can identify the parts
that need to be tested. I'm using the tag 'I believe' to identify
the places where (I believe) testing is needed.
I am attempting to create a universal human computer interface that can be used by any person on earth without regard to the language that they speak.
A 'verbal' human computer interface would require multiple translation facilities for all of the major human languages. These translators would have to comprehend both irregular syntax and grammar, along with varying regional and personal accents. Instrumentation attempts to minimize this problem by using a common generic syntax and thereby reducing translation requirements to single words and simple conversational phrases.
Furthermore, a verbal interface could be insecure and annoying in
crowded environments and confusing in noisy environments. With
Instrumentation and a heads-up display (like Google
Glass), you could carry on a secure, silent conversation
with your hands (and keypads) hidden in your pockets (disclaimer:
large pockets not included).
I believe that the chorded
(or chording) keypad can be the most efficient type of keyboard.
In this design each finger has one and only one key and each key
is either on or off (there is no third possibility). I believe
that this creates the simplest digital interface
(in any sense of the word 'digital').
I believe that Instrumentation defines an 'optimal'
implementation of two, five key chorded keypads
(one for each hand). I call it optimal because (I believe that) it
addresses the largest practical space in the
most efficient way.
Finally I believe that the Instrumentation glyph
and the ontology that underlies it provides a very mnemonic
taxonomy which ties the entire design together.
In the remainder of this document I shall attempt to justify
these claims.
If all ten fingers (including the thumbs) were used to create a
single binary index, the address space would only contain two to
the tenth terms (or 2^10) or one kibi-term. This is much
smaller than Instrumentation's address space.
FYI: In binary exponential notation, kibi-term = 2^10 = 1024 terms, mebi-term = 2^20 = 1048576 terms, etc. I will be using this convention where possible for the rest of this document.
The Instrumentation address space is created by using eight fingers to specify a binary number from zero to 255 (or #0 to #FF in hexadecimal). The two thumbs then specify a two bit exponent or 'layer' of zero to three. In hexadecimal, 255 is #FF. This means that the right hand forms the high order #F0 and the left hand forms the low order #0F for each layer. I believe that this recasts these hexadecimal numbers as "counting on the fingers".
When all four layers are specified (#FF #FF #FF #FF ), this
creates an index for an address space of 256 to the fourth power
or 2^32 or four gibi-terms.
If six fingers were used as the base and four fingers were used as the exponent, it would create an address space of 64 to the sixteenth or 2^96 or 7.9X10^28 terms. This is much larger than the proposed space, but it ignores the natural partitioning of the hands (two thumbs and eight fingers) and the user would be required to specify sixteen layers of zero to 64 to generate a complete address. In essence, this method awkwardly creates an address space that is so large that (I believe) few users would be able to master it.
A base of four fingers and an exponent of six fingers would also
address 2^96 terms but it would require 64 layers of
sixteen for a complete address. A base of two and an exponent of
eight would address 2^32 terms but would require 256
layers of four. These schemes provide no additional benefit at
greater cost.
There are no other comparable addressing schemes that use the
fingers symmetrically, where the same number of fingers on each
hand are used to create the base and the exponent. The following
are the remaining asymmetric schemes.
if one thumb was use as an exponent we would have a base of 512 and an exponent of two, which would address 2^18 terms or 256 kibi-terms and would require two layers of 512. This might be workable, but I don't think the vocabulary would be large enough to contain all of the potential scientific and technical terms and functions needed by everyone everywhere for the foreseeable future.
if seven fingers were used as the base and three fingers were
used as the exponent, the address space would be 127^8 terms or 2^56
terms or 64 pibi-terms. This also might be workable, but (I
believe that) eight layers is too cumbersome, the lack of symmetry
makes the chording awkward and the added amount of space is
unneeded.
If the right hand was the base and the left hand was the exponent
(or vice versa), the address space would be 32^32 terms or 2^160
terms. I believe that this is the largest possible address space,
but we have long since passed the point of economy of scale and
are deep into the land of diminishing returns.
Here are all of the previous schemes in order:
| Base fingers |
Exponent fingers |
# of Layers |
Address Space |
| 10 |
0 |
1 |
2^10 terms |
| 9 |
1 |
2 |
2^18 terms |
| 8 |
2 |
4 |
2^32 terms (sweet spot) |
| 7 |
3 |
8 |
2^56 terms |
| 6 |
4 |
16 |
2^96 terms |
| 5 |
5 |
32 |
2^160 terms |
| 4 |
6 |
64 |
2^96 terms (same as 6 & 4 fingers) |
| 2 |
8 |
256 |
2^32 terms (same as 8 & 2 fingers) |
In summation, I believe that four layers of 256 creates a large but manageable address space in a naturalistic manner. I shall now attempt to show that Instrumentation uses this space in an efficient way.
The total Instrumentation address space of four gibi-terms is
intrinsically partitioned into 256 Areas by the Specialization
layer.
Each Area contains sixteen mebi-terms. Each Area is dedicated to a separate discipline which has it's own specialized vocabulary. Each Area is then partitioned into Divisions, which contain sixty-four kibi-terms and into Blocks, which contain 256 terms.
Since each layer is uniquely identified using the thumbs, it is not necessary to enter any layer that has a value of zero. Instrumentation takes advantage of this by putting (what I believe will be) the most needed features at the indexes that can be selected with the fewest chords.
I've used the qualifier 'should' above to describe the actual
vocabulary because I am sure that many of my current term choices
are far from optimal. I hope that the vocabulary will improve as the language matures.
If you hold your hands with the thumbs pointing upward, the little fingers correspond to the lowest elements (zero and four) in the glyph below, while the pointer fingers correspond to the highest elements (three and seven).
If an element (a spoke or a lath) is present (blue or black), the user would press the corresponding key. If an element is absent (yellow or invisible), the user would release the corresponding key. The presence and absence of elements can be demonstrated using the game.
Since each finger has a single key, the typist's fingers never
leave their 'home' keys. Since each key is either pressed or
released, the typist needn't worry about the intermediate
positions of the fingers. I believe that these features are the
major contributors to the simplicity of the keypad interface.
It would be possible to include a second thumb button for power, but that button would only be touched once at the beginning of a session and once at the end of a session. I suspect that a separate power switch is unnecessary, but that is a minor issue in any case.
Once the finger positions have been memorized, they can also
serve as a sign language (presumably with the 'signer' interlacing
their fingers to correct the positions of the right and left hands
from the viewer's point of view). If the keys on the keyboard
provide haptic feedback, the keyboard itself can serve as a
'headless' reading device for the blind and others. 'Headless'
refers to a system that has no visual display capability.
I assume that users would start in 'arpeggio mode' where each key would be pressed sequentially. As translation of glyph appearance (or 'aspect') to finger placement begins to become automatic, the user could progress to full fledged chording where all keys needed on a given layer are pressed at the same time.
Playing the Solitaire
game has taught me to automatically see the combinations of
elements as individual hexadecimal numbers. I believe that this
will also help to teach people to think of chords as single things
rather than combinations of fingers.
I believe that the trigrams of the
I Ching are a brilliant distillation of the states of the
human condition similar to the Kübler-Ross
model or the "Hierarchy
of Needs" of Abraham Maslow. I believe that the trigrams of
the I Ching can be used to create a symbolic ontology that can
encompass all human experience. Most importantly, there is one
trigram for each finger (otherwise I would have been forced to use
something else).
A glyph has four quadrants and each quadrant contains two octants. An index can be mapped to the four layers of the glyph. The inner spokes are the two lowest order numbers (00 00 00 FF) and the outer laths are the two highest order numbers (FF 00 00 00).
The four octants on the right are the 'active' (or Yang) octants and correspond to the four fingers of the right hand. The four octants on the left are the 'passive' (or Yin) octants and correspond to the four fingers of the left hand.
The four upper octants are the 'more controlling' octants, while the four lower octants are the 'less controlling' octants.
The four polar octants (two top and two bottom) are more stable, while the four equitorial octants (two left and two right) are more transitional.
Each octant has four mnemonic
types which are based on the trigrams of the I Ching (one
type on each layer). These types are associated with the
active/passive, more/less controlling and stable/transitional
states to help users remember their locations.
I believe that these somatic, visual and semiotic cues will help
users to remember the index
associated with an individual term.
I am not claiming that the grammatical structure I have designed for Instrumentation is the best fit for human expression within the limits that I have defined above. I have tried to make the language as simple and straightforward as possible to facilitate the learning curve, but there may be more universal, adaptable or expressive arrangements that I could have used.
For anyone who is still (pretending to be) interested, the
following are my "rigidly
defined areas of doubt and uncertainty".