GPL not encouraging new technology

[Niall Douglas]

On 29 Nov 2002 at 21:16, Marcus Brinkmann wrote:

> On Fri, Nov 29, 2002 at 08:54:56PM +0100, Niall Douglas wrote:
> > And its development was exclusively funded by a university (ie; a
> > government) as far as I understand it.
> 
> It makes sense that free software is written in a cooperation between
> the public and private companies, just as it happens in various
> sciences.

It does make sense, but it's not self-supporting. Governments have 
since time immorial paid for economically unviable things with long-
term benefits.

I would wholeheatedly agree that public expenditure should result in 
totally free results (I'd prefer BSD to GPL though for this). But my 
original point that the GPL does not encourage new technology still 
remains valid.

> > > So is the GNU Hurd, which is another
> > > innovative operating system.
> > 
> > You're right it is, but it was much more radical back in ~1990 when
> > it was conceived. My understanding of it is that it's quite similar
> > to the NT kernel.
> 
> Your understanding is wrong, it's very different.  Please refer to the
> http://www.gnu.org/software/hurd.  It's much more similar to Plan 9,
> although its root reach further in the past.

I had read that document prior to replying. I stand by my point - 
note I referred exclusively to the NT kernel which I believe *is* 
quite similar to GNU Hurd (you'd need to read the hard-to-find docs 
on NT's undocumented APIs). Plan 9 is a complete reimplementation 
which affects application space completely as well, and AFAICS GNU 
Hurd doesn't and won't do that.

> That might be one reason, I have a few others in mind but I think it
> would reach too far to discuss this here.  However, I fail to see how
> this proves your point, as you are comparing two GPL licensed projects
> here.  To prove something, I would think you need to compare GPL
> licensed projects with proprietary or other licensed projects. 

No it proves my point exactly because one is quite radical whereas 
the other used to be quite retrograde. The radical stuff just won't 
attract volunteers under the GPL - only money can do that.

There are plenty of examples of propriatary ventures making real 
innovation eg; Next, Xerox Parc, AT&T etc.

> > There's no reason why commercial software can't achieve the same
> > thing, it's just we've all grown used to not receiving the source
> > with our software.
> 
> I don't know if you are talking about commercial or proprietary
> software here.  If a license does not give me the four fundamental
> freedoms, my abilities to learn from the software and use the result
> of the learning is resrticted.

I was speaking of COTS as privately produced software usually comes 
with source. I appreciate the FSF's efforts to illustrate their 
concerns with clarity, but I fear people then become too entrenched 
on either extreme between free and closed. There is a large middle 
ground offering most of the benefits of free software but encouraging 
innovation and making profit. Among those benefits for sure are 
encouraging innovation in all its forms, including the production of 
the next generation of talent.

Alex Hudson wrote:
> > There's no reason why commercial software can't achieve the same
> > thing, it's just we've all grown used to not receiving the source
> > with our software.
> 
> I disagree with this; being able to look at source doesn't teach you
> much at all (I think Marcus hasn't stated the argument well, or
> perhaps he doesn't agree with me :). You only learn by doing, and
> participating in a Free Software project (for example) is far more
> 'doing' than being able to see source for X "commercial" (proprietary)
> program. Proprietary software that comes with source very rarely
> allows you to 'do', all you can do is review, play a little, perhaps
> bug-fix. 

Again, you're restricting yourself to the current business model 
mindset. It doesn't have to be that way. If a law existed which said 
that all COTS had to come with source and could be reused by anyone 
so long as they paid a percentage of the sales cost (including zero 
if they sold it for free) ie; in other words, if mandatory reuse were 
legistlated for, then this whole situation would change.

> > No, but money arrives to GPL projects much later (and less of it) 
> > than to commercial so therefore proving my point.
> 
> I don't think that's necessarily the case. Proprietary software tends
> to come to market fully-formed (i.e., developed already); Free
> Software isn't usually written and then marketed.

You mean here marketed = distributed. I'm not. I mean marketed = 
distributed for profit.

> I would imagine that is the more common scenario - if not now, in the
> future - than mature software being commercialized in the marketplace.
> "Mature" Free Software tends to address large-user-base COTS
> (commerial off the shelf) needs, rather than the bespoke software
> market (which is probably bigger), which is probably a contributory
> factor to the scenario where commercialized Free Software is also (at
> least partially) proprietized.

Except that all bespoke projects I've ever worked on tend to reuse a 
lot of COTS. Much bespoke software as well uses customised GPL 
software because the GPL permits not releasing source if you don't 
distribute the software. So, I think it's the opposite of what you 
say.

> > Both are better implementations of existing technology, there isn't
> > much radically different in them.
> 
> I think your argument that Free Software is inherently derivative is
> a) correct, and b) wrong. Yes, the tendency to derive is more likely
> in Free Software, but only because it is the natural tendency of
> software and is far easier with Free than proprietary. But, where is
> there real innovation? I would argue nowhere, not proprietary nor
> Free. It of course depends on your definition of innovation, but
> frankly, there's nothing new under the sun.

I'd disagree. If you read patent law, they have a strict definition 
of what is to be regarded as innovative (and unfortunately has been 
mostly ignored since the 1970's in the US and increasingly here). The 
UK govt's venture capital website http://www.nesta.org.uk/ has plenty 
on how it judges whether something is innovative or not.

And I think if you apply those criteria to non-subsidised free 
software, you won't find any innovative projects performing strongly. 
To my knowledge, they don't exist - and hence my original point.

> > Hence yes, I do assume the right to make money out of writing 
> > software, just as much as painting a wall or anything else others
> > consider of value.
> 
> You can assume the right all you like; your business will not succeed
> unless someone is willing to pay you :) 

Hence the "others consider of value" :)

> In a way, you are probably correct about proprietary licences. They
> are very good at mitigating the risk of developing software: I guess
> in the way it's a lot like the futures market, where you bet on the
> future value of something in terms of realizable revenue. However,
> this is only applicable in the sense of an entrepreneur risking
> investment in software development for future reward: i.e., the COTS
> scenario where the software is brought to market. From the bespoke
> point of view, it hasn't really mitigated the risk (the customer isn't
> really buying what you are developing), and certainly doesn't work
> from the skills-to-market point of view. 

I don't quite understand you here. Bespoke software is custom usually 
once-written software to perform usually one strictly defined task. 
So therefore surely the customer is exactly buying the software?

> The value of COTS software long-term is also pretty poor - this is the
> usual reason people think Free Software licences are poor at
> commericializing software. Once a piece of software has been
> developed, in order for a developer to continue work on it there must
> be reason for it's further development. The proprietary market has
> worked out it's own answers to this question (data lock-in,
> subscription software, etc.), but it doesn't get away from the fact
> that continued development on a piece of software in order to make a
> new value proposition is hard, and made harder the longer the software
> is developed. Free Software licences are poor at locking the user in,
> are poor at selling 'bells and whistles', etc. - if you haven't
> created something of new value, it's unlikely you could
> re-commercialize it. Hence, people think it's hard to commercialize
> Free Software. What they usually mean is, it's not easy to
> commericialize COTS Free Software on an over-extended product
> life-span.

No I agree with all that and always have done. The current locked 
down closed source system is bad for software, engineering, users and 
pretty much everyone except the CEO's of the software multinationals. 
I am just as strenuously opposed to them as Richard Stallman. 
However, I also don't think free software is the perfect rose-tinted 
vision of perfection so many in the GNU world believe it to be, and 
while it's ideologically comfortable it simply isn't practical IMHO. 
One of its main problems is that in my opinion, it stifles blue-sky 
innovation - hence my posing of this question.

> To be fair, he wasn't. While GNU was originally deriviative - probably
> more because the development systems it was being created on were
> UNIX, and it was being created piece-meal - it's not true any more.
> Yes, they were not ground-up innovative, but that doesn't preclude
> them being blue-sky: in fact, the amount of innovation is actually
> quite high, which is why they are trouncing their proprietary
> counterparts. Not because of development cost reasons (which are
> arguable anyway), but because they're better in many very real ways.

This may be your view, but no one in this list has yet shown me a 
piece of innovative strongly growing unsubsidised free software so 
I'm afraid I'm going to have to disagree. You're completely right 
that free software is trouncing its competition because its quality 
is higher (more love from the programmers, vastly better peer review, 
friendly cooperative competition) but it's merely doing better at 
functionally *cloning* the competition.

> > Don't get me wrong. Given twenty years of spare time by enough
> > people, you'll eventually get there. But vastly more people will
> > volunteer their time on established paradigms that radical step
> > changes, hence the GPL is not good for blue-sky projects.
> 
> If GNU projects were not progressive, they would not over-take their
> proprietary forebears.

No, completely wrong. They're *energetic*, but not innovative.

> Remember, the GNU project was *designed* as an
> OS replacement. Not an alternative, a replacement. So, it had to
> really implement all the functions in a substantially similar fashion.

And it's precisely that reason that they still haven't succeeded ie; 
because blue-sky stuff just doesn't sit well with free software. If 
they'd set out to write the perfect functional clone (which is what 
they have ended up doing anyway), their goal and ideology would have 
been compatible.

> What were the proprietary systems doing while GNU was being developed?
> Nothing that would keep them ahead, anyway - the only proprietary COTS
> operating systems with advantage over GNU/Linux are Windows and
> Solaris (arguably). Areas in which these systems are technically
> better are few and far between, arguable, and not worth debating here.
> I suspect Solaris won't be in the race much longer either. 

Much of the reason Windows got in there is because the competing Unix 
vendors refused to get their act together, and now most have gone to 
the wall. The current closed source tie-down-the-user business model 
is wholly to blame for this in fact - if by law software had to be 
reusable, the competing vendors could repackage their competitors 
software within their own and thus inoperability would have been 
reached.

> > > There are very few examples of "completely unique [sic]"
> > > (something is either unique or it isn't, by the way - it's one of
> > > those words ;-) IT technologies around at the moment.
> > 
> > I think there are more that many would believe. Acorn RISC-OS had
> > quite a number, as did NextSTEP.
> 
> If you believe that RiscOS is innovative, but the Hurd is not, then
> you either don't know the Hurd sufficiently well, or have a clear
> Acorn bias in your argument that you can't see past :)

You should have read my words more closely. I said the Hurd was 
innovative *in* *its* *time* but not really any more. RISC-OS was 
extremely innovative in 1988, but it sure isn't today. Innovation is 
100% relative to the context it's in.

John Tapsell wrote:
> Just a quick point - remember that the majority (90% or something
> wasn't it?) of software is not off-the-shelf software. For the
> majority (all?) of not-off-the-shelf software, the licensing probably
> doesn't affect much. (code for research, putting packages and support
> etc together for 'solutions', in house development, ...)
> 
> For the software that is off-the-shelf, I'll let everyone else cover
> that.
>  But remember that is a small percentage of code that is written.

Given a majority of the major projects I have done in my professional 
life have been bespoke, sure I'm well aware of that. But as I 
mentioned above, bespoke solution are increasingly using a 
combination of COTS and thus the COTS situation is very relevent.

A small example: I had to integrate one of the industry leading 
industrial testing packages into a bespoke solution. This was COTS 
and was riddled with bugs. Time and time again I would find yet 
another exception, serious memory leak among others and I would send 
a working example of how to trigger the bug to the manufacturer, a 
large US multinational.

But their repair rate was horrendously slow, and since we were a 
multi million euro contract for them I rang them up and threatened to 
switch to a competitor. We got an improvement, but I suspect at the 
expense of someone else.

Now if their COTS came with source, I'd just fix them manually and 
email the diffs as I do with Trolltech's Qt. The entire project would 
have gone quicker, wasted less of my time with workarounds and 
produced a cheaper, better bespoke solution.

The current system is crap for everyone and with software patents now 
being pushed on us by US multinationals, it's going to get a lot 
worse - especially as they'll win the next vote with the new EU 
members on their side. We 
*need* a third way acceptable to industry. Free software is not that.

All this discussion BTW has been for a point. I've been preparing 
that third way, and I'll publish it before Christmas. Thank you for 
your input, you've helped clarify a number of issues and I look 
forward to further comments with interest!

Cheers,
Niall