Re: Economics and the GPL

On Mon, 10 Jun 2002, Anthony Taylor wrote:

> Deven T. Corzine wrote:
> 
> >It's voluntary for now.  But once you choose to use the GPL, you can't go 
> >back and change your mind later without starting over from scratch and even 
> >competing against your OWN code!
> 
> Not entirely correct.  You do not have to start over from scratch: you 
> own the copyright on your own code.  You may choose to release your own 
> code under another license as well.  (You cannot release other people's 
> code under another license unless given explicit permission.)  You 
> cannot remove your own GPLd code from distribution, so you *will* have 
> to compete against your own code, but you don't have to start from scratch.

It's not that simple.  You may own the copyright to your code, yet Stallman 
would argue that ALL of it constitutes a derived work, being based on the 
original GPL code.  I doubt he would relinquish his claim on the code just 
because you exorcise the original GPL code.  You might be left ENTIRELY 
with code that you wrote yourself, yet Stallman would probably still claim 
it as a derivative work and demand that it be kept under the GPL.

Now, he might not be correct, legally speaking, but if you're not prepared 
to fight that legal battle, you could indeed end up starting from scratch.

> You've discovered the core difference between the GPL and BSD: the GPL 
> is written with the idea that groups of people will collaborate on 
> software, with the idea that no-one can exploit the group effort without 
> participating in the group.  The BSD is written with the idea that 
> groups of people will collaborate on software, but anyone is allowed to 
> exploit the code.  That is the difference.

That's the difference in effect.  The BSD philosophy can claim to have more 
faith in human nature than Stallman evidently does -- hence Stallman uses 
coercive methods to force compliance with his standards, while BSD folks 
use social pressure and their faith in (at least some) people to do right.

> The GPL is hardly proprietary to the extent you claim; it does not allow 
> for code sharing back to other "open-source" licenses, and that is the 
> biggest flaw.  BSD folks get upset because the GPL is allowed to use 
> code from BSD, which is explicitly allowed withing the BSD license, but 
> won't give code back to BSD.  This is ironic, considering they don't get 
> upset when commercial developers do exactly the same thing.

The GPL is just as proprietary as any company.  That's why the BSD folks 
can't use the code.  But it claims to be the banner standard for "free" 
software.  Is it any wonder BSD proponents get upset about this, when they 
are more honestly free (as in beer AND speech) with their code?

> "Proprietary" means
> 
> *2* *:* something that is used, produced, or marketed under exclusive 
> legal right of the inventor or maker; /specifically/ *:* a drug (as a 
> patent medicine) that is protected by secrecy, patent, or copyright 
> against free competition as to name, product, composition, or process of 
> manufacture.
> 
> The GPL is none of these things, as users are granted a non-exclusive 
> right.  In fact, users are not allowed to make anything exclusive.  So 
> the GPL is innately anti-proprietary.  The BSD is pro-proprietary, in 
> that it does not forbid creating an exclusive license.  Microsoft 
> products are just proprietary.

Must we resort to wrangling over dictionary definitions because you don't 
want to accept the obvious?  It may be unpalatable to you, but the GPL is 
indeed proprietary in nature -- just in a unique form.

Okay, the following definitions come from the Oxford English Dictionary, if 
we must play this game:

> proprietary: 4. Of or relating to property or proprietorship.
>
> proprietorship: 1. The position or condition of a proprietor; ownership.
>
> proprietor: 2. [...] one who has the exclusive right or title to the use 
> or disposal of a thing; [...]

Most of the definitions relate to ownership of property.  However, look at 
the particular meanings from the definitions I quoted above.  Collectively, 
the GPL community "has the exclusive right or title to the use or disposal 
of a thing" -- NOBODY outside the GPL community may use GPL code.  To use 
GPL code requires joining that community be accepting the GPL's terms.  The 
membership in this community may be fluid, but there's a sharp line between 
the proprietors (those who accept the GPL and thus get to use it) and the 
rest (those who get no use of GPL code because they don't accept the GPL).

The GPL grants exclusive rights -- proprietary rights -- and demands GPL 
licensing of new code in return.  I'm not saying this isn't a fair contract 
or anything like that -- but it is most certainly a proprietary one.

Also, in a separate sense, the FSF is the proprietor of most GPL software 
as a whole, because most of it includes the "or any future version" clause 
that grants the FSF an exclusive right to determine the "use or disposal of 
a thing" by creating a new revision of the GPL.

However you slice it, the GPL's exclusivity marks it as a proprietary 
license, no matter how much Stallman would like you to believe otherwise.

> >The code still benefits more people if more people are using it, even if 
> >some of those users are paying for proprietary versions of the code.  And a 
> >proprietary version doesn't need to offer "many" compelling features, only 
> >a single one.  And that feature might well be support or documentation, not 
> >necessarily code.  Anything the user is willing to pay for.
> 
> This is fine, until someone in a dominant position takes the code and 
> extends it in an incompatible fashion.  This is what Microsoft attempted 
> with Kerberos last year.  In this way, because of their current market 
> dominance, they can make it painful to use any version of Kerberos but 
> their own, thereby making worthless the original code.

The problem here comes from Microsoft abusing its monopoly position, not 
from Kerberos being under the BSD license.  How would it be any better if 
Kerberos was appropriated into a GPL project and enhanced incompatibly with 
the free BSD version?  If the GPL were a monopoly, it would be just as 
damaging as Microsoft appropriating BSD code.

The problem is abusive monopoly power, not the freedom offered by the BSD 
license to all users, proprietary and free alike.

> The BSD license assumes everyone will play fair.  The GPL requires 
> everyone plays fair.

No, the GPL requires everyone to play the GPL game, not simply play fair.  
The BSD license probably assumes some people will abuse the license (and 
make proprietary offshoots) and that others will play fair and contribute 
back.  Some might view the danger of proprietary offshoots that don't give 
back as the price we pay for true freedom.  (Freedom always has a price!)

> Please define "fantastically successful."  First is Cygnus, a private 
> company purchased by Red Hat.  Cygnus lasted a decade customizing and 
> porting GCC to new platforms.  Cygnus never made a killing, but it was 
> profitable those 10 years.

I think they've been around closer to 15 years, but that's irrelevant.

Cygnus makes their money providing consulting/contracting services (such as 
porting GCC to a new platform) for single companies who found that route 
cheaper than buying or making a replacement.  As a matter of policy, Cygnus 
places all of their code under the GPL, in support of Stallman's political 
goals.  (I believe they also offerred additional proprietary licenses to 
companies who were willing to pay extra, but I'm not sure.)

There will always be a market for custom programming to meet the specific 
needs of a particular organization, so long as the cost of meeting those 
needs is outweighed by the value of having them met.  Cygnus just happens 
to restrict its target market to those clients who are willing to have the 
work they paid for be placed under the GPL.  Since there is valuable code 
already under the GPL (like GCC), there are companies looking to leverage 
that code, who are willing to accept the situation because it's cheaper 
than the alternatives.

The GPL consulting model that Cygnus uses (and Stallman advocates) amounts 
to "let some sucker pay for it, but let everyone use it."  Of course, it's 
better for your competitor to be the sucker who pays for the code you use, 
since that gives you a competitive advantage -- they're out the cost and 
you're not.

Cygnus was able to survive doing service and support for GNU software, 
because there was a need for SOMEONE to do it.  Do you suppose a thousand 
companies, all operating exactly like Cygnus, could be supported this way?  
I'm not sure there's THAT much GPL work out there on this basis.

Please, if you had an example that qualified as "fantastically successful", 
you and I both know you'd offer it without having to ask for a definition.

> Red Hat itself is not doing badly.  The fact it is still around is a 
> testament to the strength of their business model, when other OS 
> companies like Be have gone out of business.  ("But Be was driven out of 
> business by Linux!" I hear you cry.  Rubbish. Jean-Louis Gassee himself 
> attributed Be's demise to Microsoft's stronghold, in 
> http://lists.essential.org/1999/info-policy-notes/msg00005.html .)

Yes, Microsoft drove Be out of business.  There's no question about that.  
I've never suggested otherwise.  Red Hat may not be doing badly, but it's 
because of their revenue from boxed copies, service and support -- any GPL 
development they do is a beneficial side-effect.  Do you really think they 
would be profitable if they didn't do service and support, didn't sell any 
boxed copies, and only spent time and money on writing new GPL code?

> Right now, GPLd software will not make many millionaires.  I will not 
> deny that.  But what gives programmers the right to be millionaires? 
>  Most plumbers or electricians will never be millionaires.  Neither will 
> most accountants, nor teachers, nor even civil engineers.  What's wrong 
> with just making a good living?

If one programmer makes millions of people's lives better, I don't see any 
reason why they shouldn't be able to make millions for it.  Plumbers and 
electricians don't have that kind of widespread effect, even if their work 
is professional and necessary.  The same goes for accountants.  Teachers 
may have a wide indirect influence, if they teach influential students, but 
certainly not much direct influence.  A civil engineer may indeed influence 
millions; perhaps they DO deserve to become millionaires if successful.

Somehow it seems less objectionable for a programmer to make millions from 
a successful program that millions value than for entertainers to make the
millions that they routinely do.  But then, I'm certain I'm biased on this 
point, so take it with a grain of salt. :-)

> >Maybe one platform is proprietary and the other isn't, but each serves the 
> >same purpose ultimately, and for the purposes of the product in question, 
> >the user might not care about the platform.  Consider a device like the 
> >TiVo -- it's a standalone box which serves certain functions; it matters 
> >little to the user what the underlying platform is.  It's a black box.
> 
> Oooo, bad choice of examples.  Tivo is limited in some respects, and so 
> some people have hacked it.  Did you know you can now bypass the dialup 
> requirement by installing an ethernet card (or a *wireless* ethernet 
> card)?  Strange but true.  That's good for people like me who do not 
> have a local Tivo dialup number.  Other people like a web interface.  So 
> they've added a web server, and can control Tivo from work (say, they 
> just remembered Zalazny's Amber series starts tonight on SCI-FI and they 
> need to tell Tivo to record it).

TiVo has chosen to tacitly condone such hacking, as long as it doesn't harm 
the company.  They can't stop the hackers, but they can try to make it hard 
to do some things -- they try to make video extraction difficult because 
many content providers are still having nightmares of Napster.  They don't 
care at all if you increase the size of your hard drive, on the other hand.  

The Ethernet solutions are actually good for them -- it saves them the cost 
of the dialup access.  Surely you're not naive enough to think it happens 
without TiVo's knowledge and tacit consent?  Just how long do you think it 
would take them to block the connections from other providers and only use 
the dialup pool, if that's what they really wanted?  Think about it -- we 
can use an Ethernet in place of the dialup ONLY because they let it happen.

> Tivo the company wins because they had a ready hardware solution for 
> broadband connectivity when they released Tivo V2.  This was a win-win 
> situation for *everyone* involved.

They didn't include Ethernet in the Series 2, and they won't be using the 
existing hacked card.  If anything, they'll allow people to use USB devices 
which bridge to Ethernet for an out-of-the-box solution.  Whether or not 
this remains "unofficial" or becomes officially supported, we'll see.

> The reason it was so easy to hack was because we have some creative 
> people in the world, and because the underlying source code is freely 
> available.  This is *exactly* the sort of situation for which RMS 
> created the GPL: hardware is limited by its software, and creative geeks 
> fix it to their liking.  We are not left to the whims of the producer.

TiVo only makes the modified Linux code available, as required by the GPL.  
They don't provide source to their proprietary application.  Yes, intrepid 
hackers have worked around much of that, despite that limitation -- that's 
been happening since the dawn of proprietary software.  Do you really think 
TiVo could survive if the GPL forced them to release ALL of their code?

> If this is true, the GPL will die before it destroys the industry.  If 
> it is not true, then your reasoning is incorrect.  Either way, you win. 
>  We currently do not have enough information to definitively judge one 
> way or the other.  (Well, *I* have enough information for myself.  But 
> YMMV, as with everything in life.)

You don't know that.  The GPL codebase could continue to be an irresistible 
lure to one naive company after another, a siren call difficult to ignore.
It could then destroy those companies that succumb to the lure, but not 
before strengthening the lure even more for the next unwary company.  We've 
been warned by sailing legends, perhaps? :-)

Deven