Path: utzoo!utgpu!watserv1!watmath!iuvax!mailrus!accuvax.nwu.edu!nucsrl!telecom-request
From: wmartin@stl-06sima.army.mil (Will Martin)
Newsgroups: comp.dcom.telecom
Subject: Re: Computerized Collect Calls
Message-ID: <3593@accuvax.nwu.edu>
Date: 5 Feb 90 18:26:22 GMT
Sender: news@accuvax.nwu.edu
Organization: TELECOM Digest
Lines: 93
Approved: Telecom@eecs.nwu.edu
X-Submissions-To: telecom@eecs.nwu.edu
X-Administrivia-To: telecom-request@eecs.nwu.edu
X-Telecom-Digest: Volume 10, Issue 81, message 1 of 6


>From: stank@cbnewsl.ATT.COM (Stan Krieger)

> ...such fraudulent usage can clearly be traced.  Obviously, the voice
>recordings can be kept; all the telco or AOS needs to do is listen to
>each day's recordings and if a message appears where the person's name
>should've been, they can assume the call was made, completed, and
>accepted, and bill the receiving phone anyway.  I'm sure the
>technology exists, or can easily be developed, to allow a playback of
>only the rejected calls, so it's a lot less listening.

>True, someone may try to "prank" someone else by using an automated
>collect phone to leave such a "message", but clearly how many people
>who really did try to leave a message instead of their name would do
>it again if they were caught?

I agree that a consistent pattern of repeated abuse of the "record-a-
'name'" computerized collect-call process would result in the parties
being caught and eventually forced to pay for the calls. However, I
think it is highly unlikely that random or rare indulgers in this
would ever be caught.

1) Given the telco's increasing push for automation, I think it is
unlikely they would ever assign a human the task of listening to these
recordings, assuming they are actually kept for any length of time. (I
really would guess they are recorded on disk or solid-state digital
memory and written over again as soon as they are used.) To keep track
of the recorded voice along with the billing info would be possible,
but would it be considered cost-effective? [What is needed here is the
NSA's fabled "monitor all call contents" voice-content-recognition AI
program. :-)]

2) They could intermix the automated collect-call processing with a
random and rare actual human operator. You'd never know *for sure*
that you are talking to a computer or a person (especially as
synthezsized voices get better); if you gave a human "I'll be home at
8" as your name, they could then flag the called phone (and the
calling one, if not a payphone or hotel phone) for surveillance of the
collect-call patterns.  They could also give you a severe
tongue-lashing, which would take you aback if all you expected was a
computerized standard response! :-) However, given point 1, this seems
unlikely, unless they assign a human to handle calls from a region of
high abuse (airports, train stations, etc.). Since those areas are
also ones of high volume, which is what they'd want the computer to
handle, it still seems unlikely.

3) There are some legal issues here. Could the telco legally use an
accumulation of such recordings, if they *did* keep them, as
evidence?  Wouldn't they have to get a wiretap authorization and only
use such recordings gathered after such a warrant was issued, in order
to use these as evidence in a legal case? This starts to sound
expensive, and not worth the possible monies that could be recovered.

4) Wrong numbers are not at all unlikely. This acts against BOTH the
abuser and the possible telco methods of detecting such abuse. That is
one reason I distinguished between occasional vs. repeated use. If I
try pulling this trick, and misdial the number, some stranger hears
"Will you accept the charges for a collect call from 'I'll be home at
8'?"  They are going to say "no" or hang up (unless they are totally
crazed).  My message to my household did NOT get through, but I don't
know this.  All I know is the call didn't complete, which is what I
expected. In order to have some reasonable chance of assurance that
the message REALLY made it to my home, I have to play this game twice,
at least.  This starts eating up enough time that it might be cheaper
in real terms to just pay for the stupid call!

On the other side, if the telco tries to stick me with a charge for a
call based on their having a recording of a call from a payphone to my
number, which I rejected, and where the text was "I'll be home at 8",
all I need say is that it was a wrong number. Unless they can show a
repeated pattern of such calls, or start playing high-tech games like
voiceprinting the message and comparing that voiceprint with those of
voices on my line or making calls from my number, there's no way they
can claim this is not an isolated instance. After all, the called
party did nothing illicit or contra-tarriffs; they just said "no" or
hung up!  Only the caller can be charged with anything.

The telco has a history of charging called numbers in cases of
long-distance abuse, like blue boxing and college-student fraud, but
there are major differences between billing the called party in cases
where the evidence is many many-minute-long calls all to that number,
versus cases in which the called party didn't do anything but
explicitly reject a call, or just hung up!

In short, I don't think the telco will make much effort to counteract
this fraud, at least until it can build in enough AI to its computers
so that they can make some guess as to what is a "name" versus what is
a "message". (And how would they handle foreign languages?) It just
isn't likely to repay the effort involved.


Regards, Will
wmartin@st-louis-emh2.army.mil OR wmartin@stl-06sima.army.mil