[OSGeo-Discuss] The OGC: clueless, uncaring, and still rocking to Prince.

[Adrian Custer]

Hey all,

One of you probably has some clue who this character is who spews forth
his "The OGC is Stuck in 1999" the OSGeo planet. If so, you might 
forwards this his way , if you feel he'd appreciate someone's reaction 
to his rant.



Mr. Geo-preneur,

Back in the early days of computer networking, we used to have these
things called 'names' and other things called 'addresses' which used to
be very helpful to help us interact. I discover, in the modern way of
following links through to your 'blog' and your 'about me' page, that
you have moved on from such primitive forms of being to things like
'twits' and 'inlinks'. Unable to deal with such modern ways and stuck on
the wrong side of those pesky login screens, I cast this out to the
community hoping it may float your way.

Lovely rant this, your "The OGC is Stuck in 1999;" a classic spew. Too
bad, really, because there are some nice links in there, a good video,
some ideas worth hashing out, and other goodness. But the overall tone,
nay shit-spew, reads like the ejection of a twelve year old, full of
energy, naive, and not really old enough to carry a discussion. As a
result, since you are pissing on my work without so much as asking what
is going on, I'll assume this mail will only serve to teach me something
and therefore, I'll write my reaction not addressed to you but rather
to, and for, myself.

You are surely full of knowledge, enthusiasm, and good intent. Perhaps, 
if we ever meet, we will find the similarities that link us. Until then, 
thanks for the useful stuff in your post; I'll now try to extract that 
from the rest of your text.



So Adrian, what's in this thing?



Standards should be written for the future – not the present

Lovely reminder of Dan's wise rebuke, back at the bar counter in
Berkeley, or maybe Albany: "Adrian, you keep using that word 'should.'
There's no 'should.' There's only what there is and what you are going 
to work really, really hard to make happen."



"an unecessary ... bureaucracy ... that is truly killing innovation"

This non-sequitur I can ignore since there is actually little
inefficiency in the adoption of standards. I can ignore the conclusion
since the lack of innovation is not because it is killed by the OGC.
Produced as a throw away sentence; I'll consume it by tossing it away.



OGC is making "standards" that are ...

Okay that's a core critique, well except for the 'whole bunch of
protocols' which is probably at the wrong level of the networking stack,
a confusion that comes back later.

   * outdated:
       => cheap criticism, no content, move on. And what 'standards' are
          we really talking about? Those that are outdated perhaps.

   * overly complicated:
       => indeed, which gave rise to the whole modularization push at the
          OGC currently. Good, that still seems like the right track.

   * reference implementation:
       => no such thing in reality, despite the name being used. Luis B.
          really needs to grok the confusion generated by the testing
          group's misuse of the 'reference implementation' meme. That was
          a real victory of the commercial folk over the best interests
          of the community at large.

   * a whole bunch of protocols:
       => this confusion is key. The specs were never really trying to
          define "protocols". We probably have done it for the open web,
          essentially by focusing on the lowest common denominator
          system of exchange system, but that was probably not our
          intent. Now, we should really be pushing for experimentation
          and encouraging the successes to land as additional
          communication options. Of course, this is the whole point of
          the modularization push, so we are on the right way forwards
          for that too. Still, there's something there that needs to be
          clarified to ourselves and to the public at large.

So it looks like there needs to be some clarification. Probably we need
some kind of, 'What the OGC is up to these years' since the young
probably don't have the generous impulse 'whatever they are up to
perhaps they are making progress and doing good work, we should ask
before assuming the worst'.



XML Datastores were all the academic hype back in the day ...

That's cool to think about; and about the fatal tendency of standards
writers chasing the latest hotness. Makes me wonder again of that
Semnatic Web, and will it ever fly or, if not, why not. And, of course,
it clashes directly with the 'write for the future' thread: yep, that's 
a tension in this line of work.



Filters === SQL === XPath !?

That sounds like it comes from experience. I don't get it but certainly
recognize the complexity is huge. Passing SQL queries as a solution?
Well that certainly violates to the 'don't impose an implementation.'
Perhaps that's just facing reality that they all *are* SQL databases. 
What of nosql, though? Well, this is out of my realm of expertise but I 
can see the issue.

Then again, the CartoDB comparison is maybe silly, since it is tackling
such a totally different scale of complexity. Giving everyone access to
my database schema does not equate to giving everyone a way to access
all the database schemas of anyone else.

Room for analysis based on this but the rant doesn't seem to have good
elements to get started. Oh well, we'll fry other fish.



Two year old post

Oh, a name! Emerging out of a comment response on a stack exchange
article; a classic of this time!

Hello Ragi Yaser Burhum.



Why don't the standards refer to?

Ooh, cool new projects. Ah, binary formats, that sounds like it'd be
useful for lots of information communities.

Here is a BIG CONCEPTUAL MISMATCH, that the OGC should be mandating over
the wire protocol. This goes back to the difference between a standard
and its use.

We at the OGC keep thinking that people using the standards will put a
little work into it. And of course, serious groups do. The open web does
not though, so we probably need to do that work for them. This validates
my idea for "WMS" to become "The Open Web WMS Profile" introducing the
pieces, and eventually extensions, for the wider community. That's cool,
it clarifies a bit my work on that and the motivation for the profile.
I'll have to hang on to this to help my explanations.

So far, it seems, we actually have standardized the 'protocol' if only
for the open web. That was to define a functional service. And it uses
the lowest common denominator which was, back in the 1990s, XML. But
really, it still is. I have yet to learn of any other data structure as
widely implemented, with representations which are self structured,
archival quality, widely supported, with canonical forms for
cryptographic signatures and encryption. Exchanging the archival format
is sub-optimal of course; the intent was merely that it would be
functional. The idea now would be to allow for extensions. But could we
have the efficient extensions without the archival format? Perhaps.
That's something to think about. Then again the filter message is 
generally not something that gets archived, so there is tension there.

The idea is still that those who need to go faster will, and will go
explore/build/debug new ways of doing things that could be re-acquired
later.

Protocol Buffer looks like a server to server thing. I'd be fun to look
at the byte level design. Mark for further review. The others look
similar. The language support is not fantastic so the question is really
how complex it is to re-implement the system for those on new fangled
platforms.



'Reference Implementations' that cannot be referenced

This is lovely; it shows how pervasive 'free software' has become.
Reference implementations used to be merely functional black boxes whose
behaviour could be examined. Here 'preneur's expectation is to be able
to see the inside!


OGC Standards DO NOT HAVE REFERENCE IMPLEMENTATIONS of course. That was
a backhand squat pushed for by commercial entities who got into the
testing reference pool and then re-used the 'R.I' name. The CITE testing
system now calls these 'reference implementations' but they are not from
the point of view of the standard. Luis B. is now pushing to have
multiple implementations at the same level with no explanation of
conflict resolution between them: i.e. they are not acting as 'reference
implementations,' but rather as 'known, example implementations.'

But 'preneur goes further!? 'Reference impelmentation' SHOULD be one
that he can own!? Fuck that. Who exists in this world first to write the
standards open for everyone, and then code reference implemtations for
everyone, and finally to give those others complete, unfettered rights
the software? Be nice that, if this world were full of such volunteers.
But who gives themselves the right to expect this? What a whako world
view this guy has. Of course, by the time you are releasing a public
domain reference implementation, why bother to write the standard? Poor
guy, he's really not going to like my terms of echange, if he ever comes
across them. Good, that's a bit the point; thanks again Latuff for
pushing me even further to the left!

To top it off we are either 'clueless' or 'don't care'! Lovely, I choose
'both.'



"Web sockets"

Lovely technology. Of course, very tied to 'web' and to 'web browsers'
but great for that slice of the comupting landscape. Will have to look
in to how they could work for server to server communication. Todo, bye
and bye.



"HTTPS"

Okay, now the critique has shifted to how deployments are done. Not my
issue.

Great video! Yeah, all the OGC security stuff that does not start with
getting certificates to users is pretty broken; assuming the browser
will come with a proper system for that is indeed stupid. Good to see
how broken things are though, after that, you can't skip it. (Forwards
to Anderas).


SPDY

No Way! Someone has invented something better than HTTP!? Impossible. A
line based, textual message must be the most efficient way of doing
things. Well, no, but it sure is easy to implement, hence well, the web.

SPDY extensions will be fun; once we know how to separate the protocol
specific from the rest. Hence modularization, hence the work that the
OGC actually *is* doing.



"Non-blocking servers"

Well if the OGC bends over backwards to avoid dictating implementations,
it is exactly so they can implement servers any way they like. What does
this comment do for anyone?



"specs should be about ... the future"

Yes, I made a similar point recently, didn't I? Specs should concieve of
their lifespan. Still, predicting the future is a fool's errand. So its
more that specs should define a really good basis today on which the
future is easy to graft. Hence modularity, hence what the OGC actually
*is* doing.



Why, why, why?

WebSQL - no clue, don't know if it was any good. Probably clashed with
          the wave of NoSQL hype at just the wrong time.

Spatial DVCS - ah! Indeed. One of the great questions of the day.

                G.R. seems to be pushing forwards on that, albeit without
                separating the particular DVCS from the fundamental
                problem. Hopefully, he'll make lots of progress and we
                can learn from his mistakes.

                The open question of course is what is the unit of
                geospatial data, akin to the line of text, and can we
                find a provably optimal diff for that unit.

                Why doesn't the OGC solve the problem? Well that's
                obvious, it doesn't know how. Nor does anyone, yet. Why
                hasn't physics gotten past the point charge? Ah, math...

Web stateful connections- I don't get the question, need, intent. Next.

Highly compressed geometries- No one has seen the need and done the work.

Editing capabilities- fucking hard to do; no one has done it.

Temporal- ah, the poor neglected child of GIS.

                So the short answer is because there was lots of progress
                to be made without it. Low hanging fruit and all that.

                The longer answer is the model of General Features and
                the fundamental relations of Feature properties. That,
                however, is the subject of a long discussion, one that is
                only worth having when I have the software to manipulate
                the model and show it working. Carthage, which came, then
                went, then is again; if we make a single feature, how do
                we represent its attributes? Does a feature still have
                properties duirng its time of non-existance?

MBTiles - yeah, its great. You want to use their one service, with their
           one data model, over the web, that's a good solution.
           Interoperability is a different ball game, different scale,
           different set of issues. Small and targeted can be quick and
           simple; the OGC is playing a different game.



So, that's his rant. It's kind of annoying as these things are but at
least has a bit of content mixed in with the confusion. Clarifying the
confusion won't happen over the network, not with the vehemence of the
position expressed. Actually discussing what the OGC *does* won't help
much either.

So, what's my take-away?

   * there are some cool projects around binary serialization for
     messaging,
       => look into Protocol Buffers and derivatives, Web Sockets

   * security by communication is risky; all secure communication based
     designs really still need to start with certificate distribtuion,
       => mail Andreas/OGC Security raising the cert. exchange issue
          again.

   * we need to clarify better what the OGC does not do,
       => figure out a way to express this protocol situation.

   * modularization and better written standards is still the best way
     forwards for the documents,
       => get back at it you lazy slog.

   * the 'reference implementation' term needs to die at the OGC, until
     we have software which is really acting as an r.i.,
       => lean on Luis some more

   * we need to talk more about what the OGC actually is doing these
     days, why progress has been slow, so maybe I *should* be talking
     about my work more despite talking about something that has not yet
     landed,
      => don't be so hesitant to talk about your work; be more certain
         that you will bring it to fruition some day, despite the total
         lack of involvement of anyone else.

   * the young are still full of fight, hope, and confusion.


Well, I guess it was worth slogging through the useless invective,
except that really nothing has changed. There's still no participation,
no money, and no gratitude for writing these standards, is there?
Keep focused and carry on. But if you need to go slowly and play on the
side, that's the cost of it being all volunteer work.


a clueless and uncaring,
   Adrian Custer

P.S.
     Adrian, you are now banned from internet discussions for a while;
     you have better things to do in this lovely world.

    Critical Mass, Montevideo, let's go for a ride!