Re: [Fsfe-uk] BECTA discriminate against FLOSS?

[ian]

On Mon, 2003-12-22 at 00:49, Robin Green wrote:
> On Sat, Dec 20, 2003 at 10:30:49AM +0000, ian wrote:
> > On Fri, 2003-12-19 at 22:56, Simon Waters wrote:
> > 
> > <SNIP>
> > > In free software it is inevitable that the earlier projects will be
> > > relatively more expensive than later projects,
> > 
> > It surprises me that the "investment" angle is never highlighted much.
> > Most companies will invest say double in year one if it means year 2 to
> > 10 give even a 20% pa payback in return. Its pretty obvious that it
> > doesn't take 10s of billions to maintain and distribute an Office suite
> > so taking those 10s of billions out of the equation simple reduces the
> > global cost. You don't need a TCO study to work that out. In fact it
> > would be more efficient for BECTA to do a specific analysis of what it
> > is in OO.o that is missing that is vital to schools and then for the
> > DfES to pay for the necessary development to fill the gaps (if any).
> 
> That's an interesting point. (Apologise, over-verbose post ahead.)
> 
> A pro-private-sector bureaucrat is, I would guess, likely to reply along
> the lines of "Assuming what you say is true, why won't/hasn't some corporation
> do/done that already - they could then sell it to us to recoup their
> investment?"

Sun has already done this in effect by buying Star Corp and releasing
the code. The reason why they haven't sold to the DfES I guess is that
the DfES has not asked them and they don't know what the DfES want. If
of course the DfES put out a tender for the work to be done, proving
there was a demand I guess there would be many bidders probably
including Sun.

I run a private sector company. I don't see this as a public/private
sector issue, its a method of software development issue. Private sector
companies have already put billions into FLOSS development, I'm just
suggesting that the public sector does so too in a focused way where it
is obviously beneficial to the taxpayer.

At present the public sector is actually tilting the market heavily in
favour of commercially licensed software through mechanisms such as
E-learning credits that are based on the assumption that all software
development is funded by license revenue. This is despite directives
from the E-envoy's office that procurement should be even-handed.

> My thoughts: Well, it could happen, but I cynically suspect that the kind
> of large IT corporates who typically bid for large government contracts would
> (if they were enlightened enough to risk investing in an open source dev 
> project
> in the first place) see it as an opportunity to over-charge for development,
> over-charge for support and lock schools in to support contracts.

IBM and Sun both support FLOSS in a big way and both bid for large Gov.
contracts (Ok they have their own interests at heart but that's to be
expected). I can't see that the Gov, defining some gaps in OO.o eg lack
of multimedia support in Impress and then paying some software house to
put that right in collaboration with the OO.o community any worse than
the current situation. Perfection is rarely attainable, incremental
improvement is usually more realistic.

>  Support
> contracts which might, in fact, lock out the ability to do just what 
> makes Free Software Free - i.e. change software, or hire a random
> contractor to change it.

If the source code is open, anyone with the expertise can hack it. Just
make it a condition of the contract that the software is released under
the same license as the main product and that there is a good standard
of documentation. 

> The reader may at this point wonder how Corp XYZ would force whoever 
> downloaded
> the code to buy a support contract. They couldn't, if - as we're assuming 
> here -
> it was all FLOSS code - but they might invest in the project on the basis of:
> 
>  (a) making a sufficiently large initial sale, in which the code is "ransomed"
>  - no sale, no code released (fair enough, if the price is reasonable)

So don't allow this in the contract. How much would it cost to do the
code for embedding video and audio into OO.o? Few 10s of thousand? Who
would bid for it? Many small companies - we might. Also write into the
contract that the DfES only pays on a satisfactory result and the budget
is fixed. So no risk to the DfES so they don't need an over bureaucratic
procurement strategy. In fact they could release developments as many
small contracts that then involve many more companies so they get an
idea of which are best at it without having to use the big brand names
with heavy price tags - in fact it is government policy to try and make
contracts more widely available to SMEs.
 
>  and/or
>  
>  (b) providing a "source only" version - no binaries - to comply with the
>  GPL, then trademarking their "distro", rigorously enforcing their trademark,
>  and subtly spreading FUD about the so-called "unsupported" nature of 
> white-box
>  compiled versions - i.e. their own "product", compiled by someone else and
>  without a support contract. This, amusingly/bizarrely enough, seems to be 
> what
>  Red Hat kind of does with its Red Hat Enterprise. The fact that it involves
>  effectively spreading FUD against nearly-bit-for-bit identical copies[*] of
>  your own "distro" is what seems so bizarre to me, but it seems to work
>  moderately well for Red Hat within their market.

You could use the same argument about OO.o and Star Office. The question
is whether this situation is better than one closed monopoly controlled
entirely by a single company. Yes you will get a more complex market
with plenty of marketing bo**ox from various sources but to me that's
infinitely better than the current situation. People have to start
thinking again ;-)

> [*] The only thing you have to change if you release your own copy of RHEL
> is the trademarked items, e.g. the images and text. None of this should cause
> the code to behave differently.

So educate people - Red Hat does have to stay in business though so its
unrealistic to expect them to channel resources into development without
some pay back. The issue is whether the payback is equitable.

> For example the Red Hat Enterprise support contracts, which appears
> to constrain users from changing GPLed software. (I haven't read it in 
> detail.)
> 
> I believe that what happens in practice if you "run an unsupported
> configuration" (i.e. change the code that you're running if any part of it
> forms part of the Red Hat Enterprise distro) and they find out,
> is that you don't get sued for breach of contract, but they just terminate
> your support contract.
> 
> Or they at least tell you off and threaten not to support you any more.
> Someone correct me if I'm wrong.
> 
> Because actually attaching penalties (beyond the penalty of "your support
> contract is terminated") or treating it as a copyright violation would
> be against the spirit (and in the latter case the letter) of the GPL.
> So I would think their support contract would be framed, and enforced,
> in a way that doesn't technically breach the GPL. ("We're not preventing
> you from exercising your rights, we're only refusing to support code
> that's been modified" - which is a fair enough, if slightly ostrich-like,
> position.) Again, someone correct me if I'm wrong. The signals coming out
> of Red Hat on this have been somewhat mixed.
> 
> Anyway, getting back to the original point.
> 
> Based on my thoughts above I don't think that waiting for the private sector
> to make FLOSS solutions "acceptable" to schools (which is kind of based on
> the flawed assumption that it isn't already acceptable in many cases)
> is going to be an efficient or cost-effective route.
> Because either
> 
>   (i) The private sector won't risk it, leaving it up to volunteers to do it
>   (or perhaps enterprising contractors to do it bit by bit as part of support
>   contracts).
>   
>   or (ii) A big corporate will risk it, but end up with some so-called
>   "Free Software" solution that isn't. Something like Red Hat Enterprise.
>   
>   or (iii) Some other government might pay for a large fraction of the work
>   itself - but that still means more time wasted in the meantime for MS to
>   gouge schools on price.
>   
> More efficient would be for BECTA and/or the DfES (or whatever it's called 
> this week)
> to take a more proactive role by:
> 
>   A. explicitly endorsing Free Software as a valid and valuable option in
>      software procurement for education. Or at least valid. This is the 
> essential part.

They have done. They just leave it to schools to decide what they use
while subsidising the commercial licensed sector at £100m a year through
ELCs. Rhetoric is cheap. £100m subsidies are what its really about.
Imagine £100m a year investment in education FLOSS each year every year.
   
>   B. employing people to improve FLOSS, or (more likely with this 
> pro-privatisation
>      government), put it out to tender but set down the terms of the contract
>      from the outset so that the contract is primarily paying for the 
> _development_,
>      and all the direct and indirect Freedoms of Free Software are retained - 
> ability to
>      change code, ability to freely select support contractors, etc.
>      
>      Who knows, instead of giving it all to one monolith like EDS and wasting 
> millions
>      they might actually have the insight to farm it out to multiple smaller 
> development
>      shops with pre-existing FLOSS expertise, no wasteful airhead 
> salespersons calling
>      themselves "consultants", and an ability to produce what is actually 
> required
>      (close to) on time and on budget?
>      Nah, maybe that's too much to hope for!
>     
> But while BECTA waits for open source to become "ideal" by the "magic of free
> markets" (which for sufficiently perfectionistic values of "ideal" is never
> going to happen), MS and others will continue to gouge massive profits out of
> the education sector. The NHS and many other governmental institutions around 
> the world are starting to wake up to the power of Free Software - BECTA needs
> to get with the program.

The DfES is already distorting the market. If they really believe in the
power of the private sector they would keep their oar out of *any*
commercial dealings. I have had some interesting debates with people
working in the public sector who try and tell me that its a commercial
issue not relevant to them yet they are subsidising one commercial model
over another. Its about understanding commercial realities, stimulating
competition and management of change. The socialist/capitalist issue is
really a red herring here.

> The major challenge would be communicating that kind of argument without 
> coming off
> as not _only_ Free Software partisans, but politically anti-big-business as 
> well.
> Needs some reworking I suspect.

I don't come at any of this from anti-anysort of business. Its simply a
matter of sensible use of resources and the fact that paying many times
more in license fees than the entire development cost of an application
makes no economic sense at all whether you are a company or a
government. Its just the size of Governments stack the economics in
favour of FLOSS long term to a greater extent than in individual private
companies. Its a simply inefficient use of resources to carry on as now
so look for ways to get round the problem. GPL does it rather elegantly.


> (Of course they might actually listen to that kind of argument and think 
> that it's politcally biased _but_ still likely to be true ;-)
> 
> > I suspect that the reason they don't think like this is that they can't
> > get their heads round the fact that the conventional commercial market
> > models are not automatically the most efficient or effective.
> 
> OK, I would summarise my argument as follows: The nature of FLOSS means that
> businesses can't do the standard thing of selling software per se to recoup
> their investment. Therefore, the risk is that either they won't/can't risk 
> doing
> these improvements (which could be quite marginal in some cases) due to lack
> of profitability or lack of ability to raise capital for the project
> - or they will but they'll try and attach conditions to make it profitable 
> which work
> against the Freeness of Free Software. If, OTOH you invest in the development
> yourselves, you have a better chance of getting it done in a reasonable time
> and budget and with _no_ strings attached.

The Gov. can put up the development finance without actually doing the
work itself. Put it out to tender and insist it comes in at the right
quality and budget and under the GPL or they don't pay. Keep it simple.

> I should note that the option of putting open source dev work out to tender
> isn't really that unconventional of a model, of course - it's only 
> unconventional
> in the _software_ context. Think of it like an old-school _non-PFI_ 
> construction
> project where the government owns the resultant road/building/etc. lock stock
> and barrel and doesn't have to pay a single penny more to the company in 
> future.
> (But then, as a bonus, copies of the product magically become
> available to others for merely the cost of a download and possible a compile
> ;-) You don't get that when you commission a new hospital!)

Quite so, but they have to make the transition to seeing things in this
way. Mind I would prefer software not to be under Crown copyright. GPL
would be much better. 

Regards,
-- 
ian <address@hidden>