[Cryptography] Software patent lifetimes are the problem (Re: Hashgraph)

[Nico Williams]

On Tue, Jan 02, 2018 at 04:40:11PM -0500, Phillip Hallam-Baker wrote:
> The big problem with US patents is that you can use a technology for a
> decade while an application is varied to thread itself through the eye of
> the needle. So when assessing patent risk it is the potential claims, not
> the actual ones that are the issue.

No, the problem with software patents in (no matter the country) is how
long-lived they are.

The idea of a patent lasting N years is to give the inventor time to
recoup their investment and make a tidy profit while eventually
allowing the public to make use of it without further royalties.

The problem is that an N may be reasonable in one industry while being
entirely unreasonable in another.

17-20 years seems logical enough for pharmaceuticals -- at least it's
defensible.

But 17-20 years is several times too long for _software_.  Inventions in
software abound, and capitalizing on one can take much much less time
than with pharmaceuticals (not least because there's no FDA for
software).

Heck, the effective under-patent-and-available lifetime of a drug is
much shorter than 17-20 years.

I feel that setting context-/industry-specific patent lifetimes is
probably the easiest path to effective patent law reform locally and
worldwide.  I think a strong case for this can easily be made based on
equity.  This is just intuition, and my intuition may be wrong.

One could also point to a plethora of software patents where the
invention has been ignored by the industry until the patent expired, and
then argue that no value of N makes sense for software.  But I feel that
an argument against software being patentable is less likely to prevail
than one for setting N smaller.

N = 8 seems like plenty good enough to me, though I would prefer a
smaller value, say, 5.

Nico
--