🚀 go-pugleaf

Bernie Cosell (cosell@bbn.com) argues that algorithms, which are close kin
to padlocks, should be protected in the same way:
>   Say,
> instead of selling something that sorts a list you want to sell
> something that padlocks a box shut.  well, if you want to take
> advantage of Medeco's or Sargent's or whoever's innovations, you're
> free to buy one of theirs or license the technology.  If you don't want
> to do that, you can find an unrestricted type of locking mechanism, or
> you can invent your own.  The idea that you could take one of their
> locks apart [or read their advertising copy where they describe how it
> works] and then go and make, and sell, a knock off is outrageous.

There are several mechanisms used in toothpaste pumps.  Companies are
rightly prohibited from opening up a competitor's pump, copying the design
exactly, and marketing their own.  However, the concept of a toothpaste
pump is not protected by a patent on a particular one.  A competitor is
free to develop his own different mechanism for pumping toothpaste and
market that.  (In fact, this is exactly what has happened.)  Would you
argue that a patent on a particular type of ball-point pen should exclude
all other writing implements from the marketplace, simply because they have
the same function?

Now, suppose that I see a feature of someone else's user interface -- say,
pull-down menus -- that I wish to incorporate in my own.  I have not
learned how it works or how it was implemented, merely what it does.  I
should be free to craft my own distinct design for a mechanism with a
similar function (just as is the case in "'real' disciplines"), to invent
(as it were) my own type of locking mechanism.

If Apple or Lotus had patented the padlock, they would attempt to prevent
any other company from marketing any other securing device.  Count on it.
It is essential that the concept itself not be hostage by the owner of a
single copyright or patent; "look and feel" threatens to do this, as do
some software patents.

> }Quick Sort, Heap Sort, Shell Sort, etc. would all be patented.  YOU
> }COULDN'T SORT DATA WITHOUT PAYING A ROYALTY.
> 
> This is the sort of hyperbole that, I think, undermines the positions
> of the anti-patent arguers.  Some sorts would surely be patentable, and
> others would not.

If I understand correctly, the public-key cryptography patent covers all PK
cryptography systems, not just the RSA scheme.  Patent disclosures must
include an example implementation, and the RSA scheme is written up as such
in the PK cryptography patent disclosure.  Many other patents are written
so as to cover the method, not the particular algorithm (or simply the
prototype mechanical design).  [For instance, there was a patent granted a
number of years ago on an "interactive teaching aid".  The patent drawings
showed an extremely crude arrangement of a VCR with an attached monitor and
a few buttons for stopping the VCR, advancing to the next instructional
sequence, etc.  This patent, however, covers not just that (unworkable)
design but any others which achieve the same functionality.]  So it is not
unlikely that computer sorting could be covered by a patent, if the patent
office had been in its present state of mind a few years ago.

> The dumb sorts would surely all be unrestricted.

Why?  They were new once.

Using XOR to write a cursor on the screen is such an obvious method
that surely it would be unrestricted.  Right?

Apologists for algorithm patents commonly argue that that protection is
needed for complicated algorithms (where "complicated" varies by
apologist), while the simple ones will be somehow specially dealt with.
How?  Where will you draw the line between those algorithms which are laws
of nature and those which you claim are not?  If you, a computer
professional, cannot quickly and easily make this distinction, how can you
hope that a court of law or the patent office, both of which have amply
demonstrated their inability to make these judgments, will do so?