[okfn-discuss] Open Service Definition (revisited)

[Saul Albert]

On Tue, Aug 21, 2007 at 05:17:04PM -0400, Luis Villa wrote:
> Not only is lots of free software currently Windows-only, at one time
> *all* free software was proprietary-OS only, and my sense is that the
> current service situation is a lot closer to that period historically
> than it is to the modern situation (where there are multiple viable
> Free/Open operating systems, like Debian.)
> 
> This doesn't necessarily mean that we should allow non-free services
> in this way, but I think it means that (realistically) some of them
> have to be allowed into the stack at some point until open services
> become more common, which makes Debian a tough point for comparison.

That seems to be a good analogy. 

The line-drawing of open services between brand, service identity, API,
reliability, documentation, underlying systems, data licensing and other
considerations is hard but useful. 

The more aspects that can be seen as infrastructural, and therefore
needing to adhere to a debian-like rigorous openness, the better in my
opinion because that would allow greater flexibility. How those
services are then monetised or achieve 'brand recognition' is not
really the problem we're seeking to address. They'll work it out :)

However, what will win out is probably what works best from a
commercially competitive standpoint once the browser dies. I hope that
happens soon.

At the moment web browsers are currently a very clunky way to handle
complex data and relationships. Pretty soon, like yesterday, I'm going
to want to be able to interact with my data and its relationships in far
more interesting ways than most services make possible. I know some
interesting people building interesting and potentially very useful
software and services that should help me do that... but having lots of
incompatible and scattered APIs with dubious attempts structuring their
data isn't going to help them make it happen.

Imho, the valuable asset that the non-open service providers are sitting
on is the ability to make cross-references and complex queries on their
data sets, which they are probably only using to do 'clever' marketing.
So that value - the means of association - is really the core of what
the OSD needs to address.  Balancing this ability against privacy
concerns between users is the real issue - and coming up with facilities
that by virtue of their openness *just work better* and make more useful
things possible. That's when a landscape of open services could start to
threaten the walled gardens of web2.0 services.

Cheers,

Saul.

-- 
The People Speak   | 17-25 Cremer St.  London E2 8HD | http://theps.net
studio +44 (0)20 71007915 | saul: +44 (0)7941 255210 | ms at theps.net