daemon@ucbvax.UUCP (08/16/84)
From @MIT-MC:Telecom-Request@MIT-MC Wed Aug 15 16:03:10 1984
TELECOM Digest Thursday, 16 Aug 1984 Volume 4 : Issue 67
Today's Topics:
EasyLink
New Toy
----------------------------------------------------------------------
Date: Tue, 14 Aug 84 23:38 EDT
From: Dehn@MIT-MULTICS.ARPA (Joseph W. Dehn III)
Subject: EasyLink
EasyLink definitely has problems, but it is not quite as bad as it
looks at first. All your questions about phone numbers and who you
can send to will be answered when you get your user guide. That is,
assuming you actually read through it. When I got mine, I looked at
it a little, and decided I didn't want to. It had all these
strange-looking command sequences, with plus-signs after them, and I
decided they were crazy to be offering something like this to the
general public. So I didn't even try it.
Some time later I got a call from one of their sales reps, who was
surprised when I told him that I was already signed up (apparently
they don't bother to check their subscriber list when following up
sales leads). Unlike the robot-person who collects the sign-up
information, this person tried to be helpful, and caused me to
reconsider reading the manual. I did so, and found that it is not a
total loss. In fact, in some ways it actually is "Easy"er than MCI
Mail.
The thing that I find most frustrating about MCI Mail is the excessive
prompting. (I know, I can pay extra and have less. Amazing marketing
strategy.) EasyLink has just about none. Although the syntax looks
strange at first, being derived from Telex, it is very concise and
regular (except for the "computer letter" services that have been
kludged on). For someone used to computer languages, it is simple
once you read the manual. And if you ARE a computer (e.g., a mail
system), it is definitely much friendlier; this is probably a
consequence of the fact that Telex messages are often sent
automatically using paper tape.
Unfortunately, the average user will probably find this too much to
learn in one step. If Western Union expects to sell this in
competition with MCI Mail, they will have to do one or more of the
following: (1) make an optional prompting mode, (2) make a front-end
program for personal computers that masks the syntax (they are selling
something called EasyLink Instant Mail Manager, but I am not sure if
it does this or just provides word processing and terminal emulation),
(3) completely re-do the documentation so as to introduce the new user
to the essential features step by step.
Another thing that makes the service more confusing than it needs to
be is the existence of two different mailbox identifiers for each
user: an "EasyLink mailbox" number, and an "EasyLink Telex" number.
The second is intended for use by regular Telex subscribers who want
to send you a message. However, since EasyLink subscribers can send
to any Telex user by simply specifying the Telex number, they too can
use your "EasyLink Telex" number to send you a message. There is
nobody who needs to use the "EasyLink mailbox" number! This is
apparently a vestige of a previous policy where some more deliberate
action was needed to connect EasyLink and Telex, but now it just adds
confusion.
One more comment on electronic mail companies in general: they don't
seem to understand that electronic mail is a way to communicate. I am
constantly getting paper messages from MCI announcing this and that;
never have they sent me an electronic message, except (sometimes) in
response to a message I have sent to MCIHELP. As for Western Union,
when the sales rep offered me a phone number where I could call if I
had any questions, I asked if there was some way I could reach him via
EasyLink, so he gave me a Telex number. When I sent a question to
that Telex number, I got a reply (from a different person - it was a
general customer service department or something) telling me that they
were unable to answer my question because they didn't have my
telephone number!
-jwd3
------------------------------
Date: 14 May 84 21:38:13 EDT
From: Hobbit <AWalker@RUTGERS.ARPA>
Subject: New Toy
I recently bought myself a dialer, and seek to share my experiences
with it. This is a pocket-size unit with a flip-open case, and
doubles as a clock, calculator, memory or manual dialer. Officially,
it is the Dictograph Dial-It II, and can be had for ~60 clams from DAK
inc. [The other catalog houses wanted $70!]
Now, *I* have no real use for a memory dialer, since I am reasonably
good at remembering numbers and can easily outstrip this sucker for
speed. This thing has 100 ''locations'' capable of holding 32 digits
each [but see below]. Why so many digits, I ask?? I still haven't
figured that one out - do you know anytime you would dial 32 numbers
to call somewhere??
So the sucker finally showed up in the mail, and if it had a
personality and wanted peace and quiet, it came to the wrong place.
What do any of us do when we get a new machine? We hack away at it
until we discover first its weaknesses/shortcomings, and then the
workarounds to overcome those [meanwhile submitting SPR's]. I removed
it from its box and examined it. Click, the case opens from the *top*
- weird! Actually it turns out that this configuration makes it
easier to hold and type buttons with one hand. The display was blank.
I pressed a likely-looking button and got a ''d'' in the rightmost
digit. Then I figured What The Hell, they gave me this nice manual
along with it, might as well read it. The documentation told me the
basic syntax of commands, and I took it from there.
The unit does indeed produce touch-tones from a very small speaker
built into the bottom. This unit is a tad thicker than a typical clock
of its type; its batteries are somewhat tall and there must be room
for the speaker. A small array of holes cut through the bottom of the
case lets the tones out. They are the typical tones generated by that
dialer chip - more square-wavey than a regular TT pad and mixed up
with clocking glitches. This tends to reduce performance because the
Bell tone parsers are touchy and want tons of volume. Because this
must pass through the carbon mike, acoustic interfacing and tone
volume/purity become somewhat important. The manual claims that if
you hold the handset such that the microfern is sitting in a vertical
position, it will work better - and indeed, this is the case. Holding
a carbon mike that way does increase its transmission capabilities -
How, I have no idea. They also mention the well-known trick of
pounding the handset on the wall to break up the carbon particles.
So, as I was playing around with it, storing things, deleting them,
trying to do recursive invocations, whatever... I discovered lots of
shortcomings, which I will not hesitate to pass back to the
manufacturers. Neato things include a password you can enable to turn
it on, a downcount timer, an upcount timer, 24-hour mode, 24-hour
alarm, a slow-dial hook for flakey fern systems, and a Manual mode in
which you press button, unit sends that tone just like a regular TT
pad.
Following are excerpts from the resultant flame I sent off to these
people.
-----------------
The unit is a really good idea, and can be quite useful even to one
such as I who doesn't need 100 memories for phone numbers. With some
minor fixes and improvements, this thing could be far and away the
best dialer concept on the market. Let me, therefore, run down what I
found wrong with it. You will see that I am using this approach
because what I have to say will never fit on your Warranty
Registration Card.
I got your 800 number in Buffalo [the one you so thoughtfully *didn't*
supply in the manual] and talked to someone who knew all about the 99
bug. He informed me that the designer resides overseas and is hard to
reach; perhaps this can be forwarded to him through whatever
channels?? The 99 bug is the one that bites when you attempt to
modify Location 99 with a digit string of *shorter* *length* than the
current contents. If you use a longer or equal string, it works okay.
Otherwise the unit does really strange things with memory, loses your
current storage, creates one or more locations containing *extremely*
long strange sequences, and basically crashes, the only fix being
power removal. You'd have to look at the microcode for the thing to
begin to fix this one; I assume the aforementioned designer is
responsible.
The unit could use a Date register as part of the clock. This may not
be built into the processor you use - but a suitable software
workaround could probably be created without too much trouble.
You advertise the capacity of the thing as 100 locations of 32 digits
each. [That length, although *very* handy for some things, is a tad
longer than most people would utilize for telephone numbers.] 100 x
32 4-bit digits is 3200 possible stored digits. Memory is kept in a
1Kx4 RAM, and allowing for location-pointer overhead, you actually get
somewhere around 930 digit capacity across all the memories. This
works out to around 30 *true* 32-digit locations. I notice that
memory is used in dynamically-allocated chunks instead of fixed
partitions - *nice* feature, but to live up to the advertising, it
should have a 4K memory or so in there. The manual also fails to
mention that an attempted SET returns the ''d'' in the display if
memory is full.
I find it regrettable that one cannot use the * and # tones within
stored numbers. I would greatly favor using other keys for SET and
PAUSE, and allow the * and # equivalent tones to be stored in a
location as well as 1-0 and L and C. 4 bits will address 16 possible
keystrokes, so bus capacity for the extra keys shouldn't be any
problem. You may not believe it but this has its uses, just like 32
digits do.
A somewhat blue-sky idea: Why not, instead of making 99 and 98
special, allow the in-stream insertion of *any* other location?? That
way, if you have more than one long-distance carrier service, you can
program more than one access code. With Bell's divestiture, there
will come a day when each call will be cheapest via a certain carrier.
The Dial-it could not only store a number, but using the
''insert-location-XX-here'' feature, the user can program the cheapest
calling method in on top of it. Once you get people to understand
what this feature could do for them, they would *welcome* a dialer
with the capability. Added security would be provided by the fact
that someone else wouldn't know where the person stored his personal
access codes. When more of these things hit the market, all someone
has to do is say ''oh neat, let me look at that'', type 99 or 98, and
remember the person's access numbers, unless they are stored in some
other place selected by the owner.
I like the ''lock'' feature, but its usefulness diminishes when all I
want to do is check the time. I therefore would only use the lock if
I *know* I'm not going to be looking at it for a while, or there's a
chance it would fall into the wrong hands. I haven't come up with a
defeat for a locked unit yet, but give me time....
The tones leave something to be desired. The dialer chip is known for
imposing a lot of clocking glitches on the signal and producing
something less pure than the sine waves from a good ole Western
Electric touch-tone pad. The fact that the signal must pass through
the carbon mike compounds the difficulty. I found that my unit, as
shipped, would not *reliably* dial my home phone [which has a
brandy-spanking-new mike in it], and was completely useless on public
fones. Bashing the handset and holding it vertically helped a
*little* but I'd still have trouble. In an effort to fix this, I did
the following: First, I installed a resistor in parallel with the one
going to (-) for the output transistor. Halving the supplied
resistance makes the tones louder [that's 50 ohms, supplied by you,
down to 20 or 25 now. I suppose it'll drain the batteries faster!],
and this somewhat improved matters. But after the carbon mike, the
key to success is not just noise, it's still purity. I noticed that
when I held the dialer atop a roll of electrical tape which in turn
sat on the mike, performance was very good. The inside of the roll
created sort of an acoustic chamber which did the right thing to the
tones. I can't carry a roll of electrical tape everywhere I go, so I
did the next best thing. As supplied, the configuration of holes in
the back of the unit is flat and tends to rock around on the middle of
the bulge of the mike piece. Since the edges therefore are open to
the air, the tones escape. I sat the unit down on a small round
object and bent the center of the hole pattern upward [into the unit]
enough to clear the mike hump. Then I made a ring on the back out of
string and duct tape. Although public phones still give me trouble,
the unit works better than stock. I therefore offer the following
suggestions: Build, into the back, some kind of rubber gasket that
will seal around the microphone and create the right kind of resonant
chamber between it and the dialer. This, if done right, won't add
*too* much to the thickness. Perhaps there is an even flatter speaker
out there in the market that will help? Increase the tone volume,
and, if possible, high-filter the output so it's more ''pure''. I
haven't figured out how to do that last bit yet; fiddling around with
capacitors and things didn't work. Look into the chip that Rat Shack
uses in their pocket dialer - I haven't checked but it may be
different than the one you use, and I know that one does a *real* good
job on any phone held in any position. I'm considering replacing the
dialer chip if they are pin-compatible. Also, Rat Shack does have a
rubber gasket on the back of theirs which lies quite flat and greatly
aids transmission.
The calculator section needs some work. Just about any $9.95 LCD
calculator you pick up today will do constant holding on at least
multiply and divide. That is, if you type 2 X = = = = you will see
building powers of 2. This thing doesn't do that, requiring more
typein, and if that wasn't bad enough, typing = twice is an implied
*minus*!! Try typing 5 = = 3 =; you'll get 2. This is a definite
*bug*. While you're at it, at least one memory on the calculator
would be a real convenience. If you upgrade the memory to 4K, you
could hold *lots* of extra numeric memory.
I mentioned that the memory is dynamically partitioned. This is fine
as far as capacity goes, but if you have lots of numbers programmed
into it and try to read 99 or some higher-number location, the unit
takes a *long* *time* to find that location. Fixed partitions might
actually be more efficient and would fit in 4K, including length and
insert-loc-here headers.
An extra window should be installed in the lid, to keep dust out of
the display.
There should be a way to abort a long sequence, for those times where
the phone missed a digit or something and you must otherwise wait for
the entire sequence to play out [including pauses, etc]. This will
become necessary, if you enable the insertion of any other location in
a true recursive manner. For instance, if location 12 has 4 6 2
<insert-12> 5 in it, you'll get 4 6 2 4 6 2 4 6 2 ..... As it stands
right now, 99 and 98 are recursive only one level deep, and only for
the duration of *digits* within the invoked location. That is if 99
has 4 2 L 3 3 in it, invoking 99 will produce 4 2 4 2 3 3. True
recursion would be more desirable [and more fun!], as long as there's
an abort key.
The stronger you make the case, the better. These pocket toys often
get sat on, bent, and thrown around. The case as it stands is
reasonably tough, but you can never be too safe, especially when they
want $60 of my hard- earned green stuff for it.
---------------
My inclination is to say Go Out and Buy One. It is a neat toy and has
its uses, the discovery of which is left as a reader exercise.
I wonder if I should have included a copyright notice along with all
those ideas??? Yar, har.
_H*
------------------------------
End of TELECOM Digest
*********************