[PublicPolicy] Open Sourcing Government Code

Bastien Guerry bastien.guerry at data.gouv.fr
Tue Jun 2 21:55:03 UTC 2020


Hi Matthias,

thanks for feeding this interesting discussion with new input.

I need to quickly describe my job so that everyone better understands
from where I speak: I'm here to help public agencies publish their
source code, as the french law requires it.  I do this by discussing
the what and how with developers themselves or with project managers.

Matthias Kirschner <mk at fsfe.org> writes:

>> To play devil's advocate, the counter argument here is that since it
>> is Public Money it must be used in ways that benefit the public.
>> Creating repositories that no one looks at have no real value and is a
>> waste of taxpayer money.
>
> We heard that argument for many years and I agree that it is sad if code
> is just dumped somewhere. But my question meanwhile always is: what is
> the negative outcome of a simple code dump? How much does that cost? 

Well, there *is* a cost.  Believe it or not, but the number of people
dedicated to help public agencies publish their source code in France
is about ~1, and that would be me :/ With the support of knowledgeable
colleagues like Laurent, but still, we are very few on this front.

The time I spend pushing agencies so that they can "dump" the code
they just bought is taken from the time I could spend on more useful
tasks, like detecting the 1% public repositories that every other
agency should be aware of.

I don't say it is a complete waste of time: you learn a lot in the
process, but if this was my one and only task, I'd question myself
on the usefulness of what I do.

> Now if only 1 out of 100 times another public administration can make
> use of (parts of) software from such a code dump, isn't that already a
> positive outcome?

The long tail does not really work here: if you dump badly written
code with poor documentation on day 1, it's still useless code on day
2.  And if you add bit-rot to the equation, then you end up with many
things that feel more like a burden than a "chance".

The only real benefit you get from the "no code left behind" strategy
is to make a point, and to show everyone that *this is possible*.  But
we still need to make point about "this is something that you, as a
public agency, will really benefit from (one day)".

I believe we can -- otherwise I would not continue, but we need to be
patient until we really understand what strategy/arguments to adopt.

> There is a huge disadvantage if people in public administrations have to
> evaluate the usefulness of code before publishing it, 

This I fully agree with.

> or that they would
> have to create a contributors community before they can do so. 

There is no such thing as a private repository with contributors.

But public repositories with contributors are very few too, we need to
keep this in mind.

> This adds
> too many additional steps which are nice to have but if not done might
> also prevent the publication of useful modules for others, decreases
> transparency towards citizen, and decreases information to be able to
> compete with existing implementations. 

Transparency is the other thing where I agree with you.

> So my recommendation would be to publish by default and if you have
> budget, motivation, skills, etc. take the other steps like making it
> easy to reuse and contribute to code (e.g.  by providing good readable
> license information https://reuse.software/), documenting it, answering
> questions, including other contributors, etc. 

My recommendation is to publish by default too and to foster a sense
of ownership among developers so that they feel responsible for what
they publish.  But you still have to consider the efforts needed for
them to make this step and point at very real rewards, not just hope.

All best,

-- 
 Bastien Guerry



More information about the Publicpolicy mailing list