Index Home About Blog
From: (John R. Mashey)
Newsgroups: comp.arch
Subject: Re: Selling ideas
Date: 12 Jan 2001 04:30:12 GMT

In article <93kvco$12h2$>, (Del Cecchi) writes:

|> |> Paul Kahler <> wrote:
|> |> > Hi,
|> |>
|> |> >   Suppose a guy has a great idea for say... a processor or memory

<various good pieces of advice>

|> There is a fair chance that what you think is new is in fact not new.  But maybe
|> it is, and maybe it is as valuable as the Amazon 1-click patent.  :-)

Yes, and for sure, do *not* try writing letters to computer&chip companies
	"I have a great new idea, and for $XXX I will send you the details."

People do this; I've gotten some at every company I've worked for,
and it is very unlikely that anything ever happens.  In fact, in many places,
if you send a letter that says: "The attached document describes this idea",
it will probably be returned unread/destroyed, to avoid possible later
legal issues.  [That was the usual rule at Bell Labs, because there had been
cases where somebody sent in some vague idea, then later filed suits
if anything ever appeared that they could imagine had been covered.]

Note that to make money from a patent like this:
	(a) It has to be patentable [of course].
	(b) The patent has to be defendable [which can cost serious $$],
	    and especially, has to be defendable against claims of prior
	    art dredged up by highly-motivated people who have more time to do
	    it than patent examiners.
	(c) It must be difficult to avoid, that is, the patent has to be
	    written generally enough to cover plausible alternate ways.
	    [This is why you sometimes see bunches of patents submitted
	    that together cover every way of doing something that people
	    can think of, so there's no easy way around the patent thicket.]
	(d) Somebody has got to go use it, perhaps in preference to
	    designs they created themselves.  The idea must still be relevant.
	(e) And the use should succeed economically.

This is hard, and it takes a while at best.  Suppose one gets a patent,
which takes, say 2 years.  Then, somebody has to decide it's great,
decide to use it, and then build products using it.  That's probably
a couple more years.

Alternatively, as "doc" mentioned, if you want to work in the computer
architecture field, and if the idea is actually good, then a more
immediate payoff may be:
	(1) Write at least one published paper, say for CAN [short-lead]
	and maybe an ISCA or IEEE Micro or Computer [longer lead].
	(2) About the time that comes out, post something about it in

Assuming that, and that the idea is good, you would certainly have something
useful to show for job-hunting, and maybe get famous.  Of course, as Del
notes, you might be surprised at the number of wonderful new ideas that
have unfortunately been invented 30 years ago.

-John Mashey EMAIL:  DDD: 650-933-3090 FAX: 650-851-4620
USPS:   SGI 1600 Amphitheatre Pkwy., ms. 562, Mountain View, CA 94043-1351
SGI employee 25% time; cell phone = 650-575-6347.

Index Home About Blog