Pleasant@Rutgers.arpa (01/11/83)
HUMAN-NETS Digest Tuesday, 4 Jan 1983 Volume 6 : Issue 1
Today's Topics:
Response to Queries - Computers and the Blind &
MIT Hacker's Dictionary,
Programming - Unix (5 msgs),
Computers and People - Human Memory Capacity (4 msgs)
----------------------------------------------------------------------
Date: 31 Dec 1982 0104-PST
From: GRANGER.RS at UCI-20A
Subject: Computers & the Blind
Sommers at RU-GREEN asked about micros and the blind. You should get
in touch with Ted Sterling at Simon Fraser Univ. in Vancouver -- he
knows a lot about this.
------------------------------
Date: Saturday, 1 January 1983 18:37-EST
From: MROSE@MIT-OZ at MIT-MC
Subject: HUMAN-NETS Digest V5 #111
In reference to a message asking about an MIT hackers dictionary--
there used to be such a thing on MIT-MC in the file gjs;jargon. I
don't know if it still exists.....
martha
------------------------------
Date: 30 Dec 1982 1343-PST
From: Henry W. Miller <Miller at SRI-NIC>
Subject: Re: HUMAN-NETS Digest V5 #111
I prefer the DEL key to either control-h or control-a
(control-a was the default delete key on our EXEC.) I modified the
EXEC to use DEL, and the users loved it.
-HWM
------------------------------
Date: 30 Dec 1982 14:31 EST
From: clark.wbst at PARC-MAXC
Subject: UNIX user interface
It is true that UNIX comes with it's configurable options in a
rather difficult to use state, almost a blank state. It would be
easy to change this, but doing so would tend to bias a given
installation towards that philosophy. This environment may not be
particularly well suited to the installation, but would tend to be
followed because it is a new system, and everyone would adapt to it
because it is new... they expect it. As it is now, when the system
is new, there is an uncomfortable time, but before long it has been
taylored according to the local way of thinking, with additions of
nice ideas from other unix systems; ones that are new to the new
unix users. After that, a new user to our now old unix system may
start with a blank account, but quickly picks up the appropriate
local configuration and lore, again biased by his experience.
The point is that no one is forced into something because it is
there. The only people left out in the cold are the very first
users, and to me, that is to be expected.
As for slow, I don't know what it is that makes distribution unix
slow, but most of my unix work has been on a system that is anything
but slow, with many more than 15 users. I suppose this is not the
place for it, but I would LOVE to hear an explanation of why it gets
bogged down and how it can be fixed.
--Ray
------------------------------
Date: 30 Dec 1982 15:06 EST
From: clark.wbst at PARC-MAXC
Subject: Use of keyboard characters and EFFICIENT user interfaces
First, a rebuttal... my general philosophy in a minute...
Control-H playing delete character on input does not
preclude it's use as a backspace character on output.
So why not delete? Well, it has to do with moving your hands... I
am sure you will get the idea from the following:
I have wanted to say this for a long time, this point merely got me
going...
I have done the majority of my editing on an adm3a (no extra keys,
ctrl next to the A) running the RAND editor. The funny command key
strokes were control characters. Why do I like this cryptic
environment ?
I am a touch typist. If I have to move my hand from their 'normal'
position, first, it takes time, and second, I have to realign them.
It slows things down. By using a control character, I can press the
control key and another key simply by wiggling my fingers. I can
edit like crazy, with no wasted effort or delays. It does make a
big difference, and that is why I like it.
As for cryptic, you learn; very quickly. People have done studies
(I would guess that people on this net have seen them) that show
that you can pick almost any random commands, and before long people
learn them. When a person is using a computer, they don't sit there
and think "I want to copy a file... the command for copy is xxx, the
arguments go in this order... etc." Before long, in fact after very
few times, most of this gets routed around. you have barely
realized that you want to copy a file and your fingers are already
moving... something in there does all the processing... it works
sort of like a cache (you forget if you don't use it for a while).
Someone once asked me if I really knew what buttons I was pushing,
or how did I think of what to do so fast. The honest answer was
that I really didn't know what buttons I was pushing.
It does not take long to learn either - a few hours to be usable, a
few days to be pretty good, and a few weeks to almost be a part of
it.
In general, the people who use computers tend to use it on a regular
basis. You want to have as efficient a translation from the first
glimmer of an idea to action as you can. Extra strings, characters,
arm movements don't help. Remember the letters that came to the
conclusion that a real fancy menu type interface is only wanted by
beginners, not old users?
People are much better at adapting to computers than computers are
at adapting to people. Since the idea is to get the job done, you
don't want any extra bottlenecks.
This is not to say that we should not try to make the computer
easier to use, just that we should realize when the efficiency of
the interface is being drastically reduced at the expense of making
it more recognizable to someone totally unfamiliar with the system.
After all, who uses the system, one who never saw it before or one
who uses it every day ?
Perhaps what we need to develop is:
1) efficient interfaces between brains and action.
2) good interfaces to (1) for very occasional users which
are capable of teaching him to use the 'native' mode if
desired, and performing at more efficient, less
'naive-user-friendly' levels in between.
3) The idea of a menu command interface that can be used or
not used or called up when you want it was an excellent
example of this. The beginner used the menu, the veteran
used the command interpreter directly.
4) My RAND editor had a simple keyboard map that could be
called up with one finger wiggle, and put back just as
easily. It was sufficient 95% of the time it was
necessary.
Anyway, to me all these nifty user terminals with all the buttons
all over the place only make my job harder, and craftily designed
rather than concise commands actually just make MORE for the new
user to remember and keep straight, and that extra information is
not really easier to remember, since even though it may look nicer
printed on paper, it is not any easier to guess when you can't
remember. It is fine to make a system easy to use for beginners,
but by definition, beginners do not use a system much.
I would greatly appreciate comments or questions or disagreements
(or agreements for that matter), as I think this is important.
--Ray Clark
------------------------------
Date: 30 Dec 1982 15:37 EST
From: clark.wbst at PARC-MAXC
Subject: One hand
This is to "RWK at SCRC-TENEX", who my machine never heard of...
1) I have two hands. So do that VAST majority of computer users.
There are facilities to change the erase character for those who
do not.
2) I can type all control characters with one hand, all but three
comfortably. With practice I am sure I could get good at it.
BOTH of these points are immaterial to what key is used to delete
characters.
A well designed keyboard/computer combination (varying placement of
keys makes them inseparable) has an erase key which can be easily
typed without large movements of the hands or careful placement of
the fingers, be it control-h, del, or blamo.
It would be nice if keyboard were semi-standard, but that will never
happen because I will always disagree with you and you will always
disagree with me. So, we must modify what key we use according to
what is convenient on our keyboard, since we are the one who has to
type it 859,364 times a year.
--Ray
------------------------------
Date: 2 January 1983 20:41 est
From: Frankston at MIT-MULTICS
Subject: Re: DEL vs ^H
If one must resort to ASCII toargument DEL vs ^H, remember that DEL
is supposed to be used to RUBOUT mistakes and paper tape and
therefore the proper interpretation of DEL is to ignore it.
There is NO "delete last character command" in ASCII. Personally, I
prefer ^H (Backspace) because it is closer to the standard
correction function on a typewriter. If you want accented
characters you can use some other method of entering it.
Remember also, that what you type on the keyboard and what appears
in the files has only an accidental relationship.
Systems MUST preempt control characters. Unfortunately, there is a
paucity of shift keys on keyboards and people have resorted to use
the control key as a "code" key. It often works. One shouldn't be
surprised it if doesn't always work.
------------------------------
Date: 30 DEC 1982 1702-PST
From: RTXENM at AMES-67
Subject: "RE BROADCAST OF BEYOND HUMAN MEMORY"
Danny's query (danny at mit-od) about the "size" of human memory
sparks me to pose a number of questions I have recently thought about
at UC Santa Barbara (grad school).
Rather than just looking at the "size" (in bits) of human memory, I
wish to also consider the extensions of human memory in the form of
personal libraries containing books, records, video tapes and disks,
computer media, etc. My motivations are from some discussions with my
grad adviser, Kay and Goldberg's idea of Dynabook, and the movie TRON.
The question is this - How many bits of data does a person need to
keep throughout a life time?
My discussion at grad school consisted of a number of office talks
concluding that 32-bit addressing such as on the VAX might not be
adequate for future "home" needs. Here at the Ames research center,
we are looking for a machine in the near future (ten years) to perform
10 GigaFLOPS (Floating pt. operations) per second (Or about 10,000
times that of a VAX-11/780). The machine must have 256 MBytes of
physical memory and have data transfer rates in excess of 500 Mbits
per second. When might something like this be available to fit in the
palm of one's hand?
Personal computers are flooding the market place (perhaps
prematurely). Many people posit the change from static hard media
such as paper to the Dynabook. I have worked on the "interim
Dynabook:" the Xerox Alto and I concur. In the movie TRON, the
computer characters (sic) carried "disks" which contained all of their
information. If a disk was lost, the character would be "derezed."
Perhaps our personal libraries, stereo recordings, VHS tapes, video
disks, etc. will all fit on a TRON "disk." We might read this disk
using a Dynabook. The question evolves: what is the capacity of this
disk? Is human memory the cache for this disk? What are the
requirements for storage (in bits) for one human lifetime?
An initial simplistic stab at an upper bound might be: suppose human
sight might be a window (from computer graphics) which might be a 1024
by 1024 color raster with a refresh rate of 30 frames per second.
Assume 24 bit color. Assume the average human is awake 16 hours per
day (grad students slightly more!). Times 70 years for an average
life span. This of course does not take into account other senses
(Sight is the primary sense.), forgetting, masking, etc.
The above might interesting in the light that CMU and others are
requiring advanced PCs for attending school. Atari 800s would come no
where near the capacity to adequately satisfy future needs of people.
This is especially important in the light that the GRiD Navigator
executive computer is about the size of the Dynabook. I ultimately
hope that I could reliably store my entire personal library (books,
records, tapes, images) in a small booksized box.
--eugene miya (rtxenm at ames-67)
------------------------------
Date: 31 Dec 82 03:35:35 EST (Fri)
From: Chris Torek <chris.umcp-cs@UDel-Relay>
Subject: Re: Human memory capacity
You might note that some people are very good at remembering one
"kind" of thing yet poor at another, e.g. someone who can remember
things like the phone number of some store called once while in
California ten years ago, but who can't remember the name of the
street he/she lives on. I, for example, have particular trouble
associating names with faces.
As an aside, I suspect that those who have eidetic memories are
better off keeping the fact to themselves. While I don't have a
photographic memory, I do have a good one for many things, and I've
noticed that people in general become annoyed with me when I
"remember too much".
- Chris
------------------------------
Date: 2 January 1983 00:14-EST
From: Zigurd R. Mednieks <ZRM @ MIT-MC>
Subject: HUMAN-NETS Digest V5 #113
The calculation of memory capacity from an estimate of raw data
taken in is very clearly wrong. That data could never get to one's
memory in the first place. There isn't the bandwidth down the optic
nerve. What gets to our brains, from our eyes, is heavily
preprocessed. Our eyes are not only receptors, they are organs for
seeing.
To play back a scene from our memory, we run through our minds the
mind's own coding of that scene. The raw bits were lost as soon as
the eye saw the next thing. Roughly the same thing goes for hearing.
This all has a lot of consequences for AI researchers. In
particular, we may be hoping in vain to "see" using
fourier-transform boxes attached to low resolution TV cameras with
god-awful optics. All sorts of things get done for us by our eyes,
like lighting and color correction, before we even get to
contemplate what it is we are seeing. I am not suggesting that only
humans can process information like humans can, or any mumbo-jumbo
like that. I am suggesting, however, that certain architectural
considerations have not been taken into account in the construction
of systems that hope to cognate.
For more information on this branch of thought see David Marr's
book, "Vision". If you are at MIT and want an interesting terms
worth of the same, take 9.36.
Cheers,
Zig
------------------------------
Date: 2 January 1983 20:41 est
From: Frankston at MIT-MULTICS
Subject: Re: Human memory capacity
I tend toward fairly low estimates of memory size because I feel
that the human brain would tend to be relatively efficient. Storing
a lot of information can be better done by storing the information
compiled. One can store a lot more information by remembering "Joe"
than having to keep track of every bit of a visual image.
Marvin Minsky has a good story (I would like to find the original
source) about use of hypnosis to find details. The problem with
such experiments is that no one usually bothers to check the
remembered details against reality. It turns out that a twenty year
old is able to remember his thirty fifth birthday just as well as
his fifth.
------------------------------
End of HUMAN-NETS Digest
************************