Subject: Re: patents and web services
From: simo <s@ssimo.org>
Date: Sat, 03 Jun 2006 21:33:40 -0400

On Sat, 2006-06-03 at 14:05 -0700, Thomas Lord wrote:
> simo wrote: 
> > On Sat, 2006-06-03 at 09:42 -0700, Thomas Lord wrote:
> >   
> > > We're entering a period of big "black box" web services:  patents
> > > *might* be very helpful here.
> > >     
> > 
> > Let's see if, in the software field, this holds true.
> > 
> > You claim that re-use is important, and that patents may help.
> > 
> > Now that would be true if patent holders were forced to make their
> > patented methods available to third parties. This is not the case, a
> > patent holder is not compelled to let other use the patent until it
> > expires (20 years later at best).
> >   
> 
> What if it were 3 years?  or 8?   What if mandatory licensing was
> required and there were statutory limits on fees?

If there were a 3 years limit (from invention not from the time of
approval), than I would allow any patent, I would even relax the PTO
check even more than now. 3 years is a perfectly acceptable time for a
patent in the software field. 8 years would already be too much it is
more than 10% of the whole life of software research.

> I didn't say that every detail of today's patent law was right --
> only that patents in general may be quite useful in forcing
> disclosure.
> 
> There is no term limitation on trade secrets.   If Google came up 
> with "the most compelling, mysterious web service ever" and kept
> it as a trade secret -- they could hold onto that for much more than
> 20 years.  (Coca-cola)

And you think nobody would ever rediscover that again?
It is mathematics, it is not something that have to do with physical
process and taste. Anything really useful would be rediscovered once
somebody else would really need it IMHO.

> > I assume you are not interested in re-use after 20 years, but in
> > immediate re-use,
> > 
> >   
> 
> Nope.   Not immediate re-use in direct competition to a diligent
> effort
> to build a business around the patent.
> 
> The choice is between Google never voluntarily telling us how
> the latest trick was accomplished vs. all of us rewarding Google
> for a limited time for letting us in on the secret.   I prefer the
> latter.
[...]
> Hence the trickiness.   Basically, for a limited time, Google should
> not lose money because a customer has (forcibly) licensed Google's
> patent.  20 years does indeed seem too long -- and I don't think 
> Google really needs 20 years, anyway.

And do you really think that Google would ever discover such a secret
just for a 3 years patent ?
I don't think so, you do that only if you can keep an advantage for a
very long time (and 20 years in software is an very long time).

> > So what do we end up with?
> > We end up with no real important secrets revealed, 
> Only with a bunch of bogus assumptions you've made.  (Current
> terms, no forced licensing, etc.)

Same level of bogus assumptions you do, you seem to assume that a
Company would like to share a secret without taking a very big advantage
out of doing that.

> > * If you don't believe this think of a simpler case. Do you think that
> > the CocaCola Company would ever release their secret recipe if it were
> > patentable? Of course not, by keeping it secret they have enjoyed high
> > profits for decades, once patented their monopoly would end just after
> > 20 years, then competition would substantially drive their profits down.
> > 
> >   
> So, get into the kitchen.   Avoid mass spectrometers and other reverse
> engineering devices.   Apply your culinary skill.  Reinvent something
> really similar to and a fine substitute for Coca-Cola.
> Get some start-up capital.   Go into production.   Make history.
> They can't stop you.

They didn't stop Pepsi, they would have if they had a patent on "colas"
that lasted 100 years (which for physical goods I think is a time
comparable to 20 years in the software field).

Simo.