Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!bu.edu!wang!tegra!mcdougal
From: mcdougal@tegra.COM (Steve W McDougall)
Newsgroups: comp.dcom.modems
Subject: Re: hayes lawsuit
Summary: Better start writing those royalty checks...
Message-ID: <2387@riddler.tegra.COM>
Date: 14 Jun 91 11:53:22 GMT
References: <676362343.49@egsgate.FidoNet.Org>
Reply-To: mcdougal@tegra.UUCP (Steve W McDougall)
Followup-To: comp.dcom.modems
Organization: Tegra-Varityper, Inc. Billerica, MA
Lines: 67

In article <676362343.49@egsgate.FidoNet.Org> Bill.Mayhew@f98.n250.z1.FidoNet.Org (Bill Mayhew) writes:

>... only things such  as
>... mechanical assemblies, manufacturing processes
>and computer algorithms may be patented under additional
>stipulations of uniqueness and originality.
>
>I do not understand exactly the way the (wait one second) +++ (wait
>one second) in band data link escape used by Hayes modems is
>patentable.  I could understand a specific microprocessor algorithm
>or hardware timer chip to implement the escape could be patented,
>but not the escape sequence itself.  There obviously nearly
>limitless numbers of ways the escape could be implemented with
>hardware and software.

Patent protection is a bit broader than that.  You can patent an
"invention" or a "process".  It doesn't have to be a manufacturing process.
In this case, it's a process for controlling a piece of communications
equipment.

I doubt that Hayes is claiming protection for this invention as a computer
algorithm.  In fact, there is a good case that this process does *not*
constitute an algorithm, since it can be demonstrated to fail on certain
(legitimate) data streams - i.e. any which happen to contain the
wait+++wait sequence.

Hayes' invention may seem obvious, but many inventions do - after someone 
else has invented them.  That's part of what patent protection is all about.
To see the non-obvious nature of this invention, think about what it is
really based on.

Hayes wanted a way to send out-of-band data on the in-band data stream.
They looked at the serial communications protocol and noted that certian
timing relationships - specifically, the intervals between bytes - were
not specified.  Then, they observed typical modem traffic, and noted
that certain patterns of intervals - e.g. 1 sec delay, 3 bytes, 1 sec delay - 
rarely occurred.  Based on this, they constructed their wait+++wait
escape sequence.  Looked at this way, perhaps it seems a bit less obvious.

It is true that there are many other ways to control a modem.  As has been
noted in a previous posting to this group, RS-232 defines a set of modem
control signals that suffice for most purposes.  Anyone who wants to can
build and program a complete RS-232 interface.  Royalty-free.  So why
do so many people want to use Hayes invention?  Its cheap.  DB9 instead
of DB25.  Fewer ICs.  Smaller cables.  Easier to program.  In short, a
useful invention.  

Other software implementations?  Well, it depends what you mean software.
Hayes' patent may cover the entire concept of controlling a modem by
sending out-of-band data on the in-band data stream.  In that case, just
using a different escape sequence, or finding a different way of encoding
the out-of-band data wouldn't get around the patent.

Is it novel?  Probably.  Hayes was one of the first mass-market modem
manufacturers.  I doubt that anyone else was using a system like this 
way back then.  A previous posting noted that some (Sperry?) paper tape
punch used a similar scheme.  Fine.  Hayes took that scheme and applied it
to the control of modems.  Patent law protects new applications of
existing inventions.  

I'm generally opposed to software patents.  But the Hayes patent isn't for
software, or an algorithm.  It's a legitimate process.  Come on, lighten
up.  It probably expires in 10 years.  If you're really unhappy, write your
congressman.  Better yet, write the Patent Office.  They are soliciting
public comments on current issues in patent law.  The request for
comments has been posted around the net the last few weeks.  Check
sci.crypt, misc.legal, comp.dcom.modems and maybe comp.gnu.* for a copy.