xref: /freebsd/usr.bin/compress/doc/NOTES (revision 7899f917b1c0ea178f1d2be0cfb452086d079d23)
1
2From: James A. Woods <jaw@eos.arc.nasa.gov>
3
4>From vn Fri Dec  2 18:05:27 1988
5Subject: Re: Looking for C source for RSA
6Newsgroups: sci.crypt
7
8# Illegitimi noncarborundum
9
10Patents are a tar pit.
11
12A good case can be made that most are just a license to sue, and nothing
13is illegal until a patent is upheld in court.
14
15For example, if you receive netnews by means other than 'nntp',
16these very words are being modulated by 'compress',
17a variation on the patented Lempel-Ziv-Welch algorithm.
18
19Original Ziv-Lempel is patent number 4,464,650, and the more powerful
20LZW method is #4,558,302.  Yet despite any similarities between 'compress'
21and LZW (the public-domain 'compress' code was designed and given to the
22world before the ink on the Welch patent was dry), no attorneys from Sperry
23(the assignee) have asked you to unplug your Usenet connection.
24
25Why?  I can't speak for them, but it is possible the claims are too broad,
26or, just as bad, not broad enough.  ('compress' does things not mentioned
27in the Welch patent.)  Maybe they realize that they can commercialize
28LZW better by selling hardware implementations rather than by licensing
29software.  Again, the LZW software delineated in the patent is *not*
30the same as that of 'compress'.
31
32At any rate, court-tested software patents are a different animal;
33corporate patents in a portfolio are usually traded like baseball cards
34to shut out small fry rather than actually be defended before
35non-technical juries.  Perhaps RSA will undergo this test successfully,
36although the grant to "exclude others from making, using, or selling"
37the invention would then only apply to the U.S. (witness the
38Genentech patent of the TPA molecule in the U.S. but struck down
39in Great Britain as too broad.)
40
41The concept is still exotic for those who learned in school the rule of thumb
42that one may patent "apparatus" but not an "idea".
43Apparently this all changed in Diamond v. Diehr (1981) when the U. S. Supreme
44Court reversed itself.
45
46Scholars should consult the excellent article in the Washington and Lee
47Law Review (fall 1984, vol. 41, no. 4) by Anthony and Colwell for a
48comprehensive survey of an area which will remain murky for some time.
49
50Until the dust clears, how you approach ideas which are patented depends
51on how paranoid you are of a legal onslaught.  Arbitrary?  Yes.  But
52the patent bar of the CCPA (Court of Customs and Patent Appeals)
53thanks you for any uncertainty as they, at least, stand to gain
54from any trouble.
55
56=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
57From: James A. Woods <jaw@eos.arc.nasa.gov>
58Subject: Re: Looking for C source for RSA (actually 'compress' patents)
59
60	In article <2042@eos.UUCP> you write:
61	>The concept is still exotic for those who learned in school the rule of thumb
62	>that one may patent "apparatus" but not an "idea".
63
64A rule of thumb that has never been completely valid, as any chemical
65engineer can tell you.  (Chemical processes were among the earliest patents,
66as I recall.)
67
68	ah yes -- i date myself when relaying out-of-date advice from elderly
69	attorneys who don't even specialize in patents.  one other interesting
70	class of patents include the output of optical lens design programs,
71	which yield formulae which can then fairly directly can be molded
72	into glass.  although there are restrictions on patenting equations,
73	the "embedded systems" seem to fly past the legal gauntlets.
74
75	anyway, I'm still learning about intellectual property law after
76	several conversations from a Unisys (nee sperry) lawyer re 'compress'.
77
78	it's more complicated than this, but they're letting (oral
79	communication only) software versions of 'compress' slide
80	as far as licensing fees go.  this includes 'arc', 'stuffit',
81	and other commercial wrappers for 'compress'.  yet they are
82	signing up licensees for hardware chips.  Hewlett-Packard
83	supposedly has an active vlsi project, and Unisys has
84	board-level LZW-based tape controllers.  (to build LZW into
85	a disk controller would be strange, as you'd have to build
86	in a filesystem too!)
87
88 	it's byzantine
89	that Unisys is in a tiff with HP regarding the patents,
90	after discovering some sort of "compress" button on some
91	HP terminal product.  why?  well, professor Abraham Lempel jumped
92	from being department chairman of computer science at technion in
93	Israel to sperry (where he got the first patent), but then to work
94	at Hewlett-Packard on sabbatical.  the second Welch patent
95	is only weakly derivative of the first, so they want chip
96	licenses and HP relented.  however, everyone agrees something
97	like the current Unix implementation is the way to go with
98	software, so HP (and UCB) long ago asked spencer Thomas and i to sign
99	off on copyright permission (although they didn't need to, it being pd).
100	Lempel, HP, and Unisys grumbles they can't make money off the
101	software since a good free implementation (not the best --
102	i have more ideas!) escaped via Usenet.  (Lempel's own pascal
103	code was apparently horribly slow.)
104	i don't follow the IBM 'arc' legal bickering; my impression
105	is that the pc folks are making money off the archiver/wrapper
106	look/feel of the thing [if ms-dos can be said to have a look and feel].
107
108	now where is telebit with the compress firmware?  in a limbo
109	netherworld, probably, with sperry still welcoming outfits
110	to sign patent licenses, a common tactic to bring other small fry
111	into the fold.  the guy who crammed 12-bit compress into the modem
112	there left.  also what is transpiring with 'compress' and sys 5 rel 4?
113	beats me, but if sperry got a hold of them on these issues,
114	at&t would likely re-implement another algorithm if they
115	thought 'compress' infringes.  needful to say, i don't think
116	it does after the above mentioned legal conversation.
117	my own beliefs on whether algorithms should be patentable at all
118	change with the weather.  if the courts finally nail down
119	patent protection for algorithms, academic publication in
120	textbooks will be somewhat at odds with the engineering world,
121	where the textbook codes will simply be a big tease to get
122	money into the patent holder coffers...
123
124	oh, if you implement LZW from the patent, you won't get
125	good rates because it doesn't mention adaptive table reset,
126	lack thereof being *the* serious deficiency of Thomas' first version.
127
128	now i know that patent law generally protects against independent
129	re-invention (like the 'xor' hash function pleasantly mentioned
130	in the patent [but not the paper]).
131	but the upshot is that if anyone ever wanted to sue us,
132	we're partially covered with
133	independently-developed twists, plus the fact that some of us work
134	in a bureaucratic morass (as contractor to a public agency in my case).
135
136	quite a mess, huh?  I've wanted to tell someone this stuff
137	for a long time, for posterity if nothing else.
138
139james
140
141