[Geowanking] LazyWeb request

[Tim O'Reilly]

In a number of talks over the past few years, I've lamented the fact
that MapQuest has never figured out how to become a platform. I
thought I'd repeat that lament as part of my keynote at our upcoming
Emerging Technology Conference (http://conferences.oreilly.com/etech).
(the talk will focus on my technology wishlist.)

But I'd like to give my skeletal thoughts a bit more meat, and I figure
you guys are the right folks to ask for ideas on how to flesh things
out.

Here's my premise:

- All of the "killer apps" of the first generation web (Google, Amazon,
EBay,) except for MapQuest have started down the path of turning
themselves into platforms, rather than just applications -- except
MapQuest and its imitators.

- What's more, Google, Amazon and EBay all leverage the behavior that
Clay Shirky outlined (I think this was in his talk Listening to Napster
but I can't find it in the published paper), namely, that the system is
architected in such a way users build the database as a side effect of
their individual "selfish" pursuits, rather than being paid or
volunteering. It's a bit more of a stretch with google, but you can
argue that pagerank is this kind of thing. Once again, mapquest and
its imitators are the only ones left out.

- Of all the internet killer apps, MapQuest is also the only one that
didn't become dominant. (maps.yahoo.com is run by vicinity, after they
dumped mapquest).

So, my question for you is this:

If we were to envision a next generation, collaboratively-enhanced
version of MapQuest, or Maps.yahoo.com, or mapinfo, how might we do it?
What features would lead people to naturally annotate maps?

What hacker work has already been done in this area? (I know there's
been some stuff naming cell towers so you can set alerts on them in
your phone, and there was the whole virtual london kind of thing) but
if we were really to brainstorm an ideal service that made it easy to
extend with the kind of rich commentary and added value that you see in
Amazon and EBay, what would it look like, and who's doing interesting
prototypes.

Thanks for your help.
------------------------------------------------------------------------
---------------
Tim O'Reilly @ O'Reilly & Associates, Inc.
1005 Gravenstein Highway North, Sebastopol, CA 95472
707-827-7000
http://www.oreilly.com (company), http://tim.oreilly.com (personal)

[Rob Hranac]

The key issue is geo-specific standards standards.
Geospatial features can only become a platform if
people follow standards. I would argue that this is a
step more complex than something like PageRank or
purchasing preferences because the information that
must be parsed to create a geospatial database is more
complex than links or purchases.

Furthermore, geospatial databases must be shared among
many organizations. They can't be crawled (Google) or
centralized (Amazon) because no one organization has a
handle on all geospatial data, nor can they. This is
why open standards are required - to leverage
inter-organizational geospatial data. In my opinion,
the best geospatial standards are from the OpenGIS
Consortium (http://opengis.org)

As to "where is the killer-app?" it doesn't exist yet.
But, the people on this list are searching for it and
there is a groundswell of feeling that it is out
there, somewhere. GeoURL is close.

Best Regards,
Rob Hranac

Post by Tim O'Reilly
In a number of talks over the past few years, I've
lamented the fact
that MapQuest has never figured out how to become a
platform. I
thought I'd repeat that lament as part of my keynote
at our upcoming
Emerging Technology Conference
(http://conferences.oreilly.com/etech).
(the talk will focus on my technology wishlist.)
But I'd like to give my skeletal thoughts a bit more
meat, and I figure
you guys are the right folks to ask for ideas on how
to flesh things
out.
- All of the "killer apps" of the first generation
web (Google, Amazon,
EBay,) except for MapQuest have started down the
path of turning
themselves into platforms, rather than just
applications -- except
MapQuest and its imitators.
- What's more, Google, Amazon and EBay all leverage
the behavior that
Clay Shirky outlined (I think this was in his talk
Listening to Napster
but I can't find it in the published paper), namely,
that the system is
architected in such a way users build the database
as a side effect of
their individual "selfish" pursuits, rather than
being paid or
volunteering. It's a bit more of a stretch with
google, but you can
argue that pagerank is this kind of thing. Once
again, mapquest and
its imitators are the only ones left out.
- Of all the internet killer apps, MapQuest is also
the only one that
didn't become dominant. (maps.yahoo.com is run by
vicinity, after they
dumped mapquest).
If we were to envision a next generation,
collaboratively-enhanced
version of MapQuest, or Maps.yahoo.com, or mapinfo,
how might we do it?
What features would lead people to naturally
annotate maps?
What hacker work has already been done in this area?
(I know there's
been some stuff naming cell towers so you can set
alerts on them in
your phone, and there was the whole virtual london
kind of thing) but
if we were really to brainstorm an ideal service
that made it easy to
extend with the kind of rich commentary and added
value that you see in
Amazon and EBay, what would it look like, and who's
doing interesting
prototypes.
Thanks for your help.

------------------------------------------------------------------------

Post by Tim O'Reilly
---------------
1005 Gravenstein Highway North, Sebastopol, CA 95472
707-827-7000
http://www.oreilly.com (company),
http://tim.oreilly.com (personal)
_______________________________________________
Geowanking mailing list
http://lists.burri.to/mailman/listinfo/geowanking

[Allan Doyle]

Post by Tim O'Reilly
In a number of talks over the past few years, I've lamented the fact
that MapQuest has never figured out how to become a platform. I
thought I'd repeat that lament as part of my keynote at our upcoming
Emerging Technology Conference (http://conferences.oreilly.com/etech).
(the talk will focus on my technology wishlist.)

Part of the reason for this is probably economic and not
technical. In order for MapQuest to become a platform, the single
biggest thing it would have to do is provide a location interface
that lets you enter some kind of a request and get back a map that
you could actually plot your own data on top of. That means that each
pixel you got back from MapQuest would have a known
latitude/longitude. That would be ducky for us, but it would drive
their data costs through the roof. They buy their data from other
companies like NavTech. If you could requst maps that have all their
locations known, then you could systematically step through the US or
the world and download your own map database.

Post by Tim O'Reilly
But I'd like to give my skeletal thoughts a bit more meat, and I figure
you guys are the right folks to ask for ideas on how to flesh things
out.
- All of the "killer apps" of the first generation web (Google, Amazon,
EBay,) except for MapQuest have started down the path of turning
themselves into platforms, rather than just applications -- except
MapQuest and its imitators.
- What's more, Google, Amazon and EBay all leverage the behavior that
Clay Shirky outlined (I think this was in his talk Listening to Napster
but I can't find it in the published paper), namely, that the system is
architected in such a way users build the database as a side effect of
their individual "selfish" pursuits, rather than being paid or
volunteering. It's a bit more of a stretch with google, but you can
argue that pagerank is this kind of thing. Once again, mapquest and
its imitators are the only ones left out.

There's actually a lot of database building going on out there. Much
of it can be had for free (at least in the US). There are several
places you can go to look for the data but unless you're into
geographic information systems, the hurdles are too high. There are
few good "consumer" level or even non-GIS geek level sites to play
with.

There are grass-roots database building efforts as well. They just
tend to be very, very hard to find.

Coincidentally, blogdex shows a new site http://www.world66.com/
that's being blogged about in a lot of places. Trouble is, the site
has been down all day so I have not had a chance to look at it
yet. The descriptions of it make it sound like some kind of
open-content data collecting site. Who knows. It might be your killer
app if they get their act together... Google does have some cached
views of the site.

Post by Tim O'Reilly
- Of all the internet killer apps, MapQuest is also the only one that
didn't become dominant. (maps.yahoo.com is run by vicinity, after they
dumped mapquest).
If we were to envision a next generation, collaboratively-enhanced
version of MapQuest, or Maps.yahoo.com, or mapinfo, how might we do it?
What features would lead people to naturally annotate maps?

People need to be able to trust the data they see. That means in an
open-content setting, people have to be able to develop reputations
and be able to see other peoples' reputations. The blogging world
seems to have implicitly figured this out. Using trackbacks and
comments, it's pretty easy to figure out whether a blogger is worth
reading or not based on each readers' sensitivities.

Developing data is a byproduct of many hobbies. Birdwatching,
geocaching, chasing degree confluences, wardriving. What is needed is
a way to easily serve that data and advertise its existence. Google
maybe isn't enough.

Post by Tim O'Reilly
What hacker work has already been done in this area? (I know there's
been some stuff naming cell towers so you can set alerts on them in
your phone, and there was the whole virtual london kind of thing) but
if we were really to brainstorm an ideal service that made it easy to
extend with the kind of rich commentary and added value that you see in
Amazon and EBay, what would it look like, and who's doing interesting
prototypes.

As part of a grant writing effort I was part of we put together a
little bit of information that might help think/talk about this some
more. The context of the proposal was to help relief organizations
but the concepts are more general. You can find it at
http://www.eogeo.org/projects/2003-09-30_osi/

Post by Tim O'Reilly
Thanks for your help.

Good luck...

[Tony Pryor]

I won't be there because I am a poor bastard.

At one point I thought that having a total visual catalog of what's on
the ground was a starting point- this was the angle I was coming from
with a prototype collab mapping project three years ago
(http://www.godseye.com/) - a reference model not unlike what people are
doing with rendered maps. Closer (.2m per pixel) full color orthos are
beautifully rich in detail. Certainly those two are mixed in GIS in
projects. But maps and orthos are not quite the compelling collaborative
model that they tease.

I think we're going toward a songlines/waypoint trust and/or locality
based collaborative filtering model with a power of patterning that will
back burner the reference approach, lending us organic segments that
lead us through vast swarms of invisible
ore/geospam/noise/chaos/meaninglessness/whatever.

some problems:

*Ubiquitous nodally-threaded geotagged image capture with effortless
contextual metadata- it's on the verge of happening for geeks, there are
even some early working examples. Overlay Reed's law with proximity and
the party isn't even scheduled, except maybe in NY and London... even
then the year of this huge mad party is fuzzy.

*Retrieval must be inferential/context strong. An enfiladic
psychogeographic viewer would be nice! Goggles, vibrating sneakers, I
don't care. Buttonless non-interruptive feedback. I haven't seen
anything at all like that. I feel that we are a bunch of reference dorks.

*Cell phones aren't designed for it as a natural activity. Forget PDAs.

my prediction framework:

Lazy p2p collaboration - trusted network friend of a friend queries
provide minimal information about people around you from many sources,
giving you an exceptionally clear picture.

I predict that wireless social trust networks that use lazy inferential
p2p collaboration will be here with disruptive force within one year.
Reality mine the people you haven't met in your immediate proximity by
querying your p2p trusted network. There exists a tangible demand for
that information, not unlike music sharing, that will ensure fast
growth. And hey, people are also actually funding social networks
development. Sneak it in!

I think the surge of demand for p2p trusted psychogeographic sharing
will be discovered by the lazy p2p collaboration ready public as a cool
thing That Your Friends Are Doing.

more babble fomenting here:

http://dev-bywater.media.mit.edu/wiki/borglab/Tony_Pryor

-t

Post by Tim O'Reilly
In a number of talks over the past few years, I've lamented the fact
that MapQuest has never figured out how to become a platform. I
thought I'd repeat that lament as part of my keynote at our upcoming
Emerging Technology Conference
(http://conferences.oreilly.com/etech). (the talk will focus on my
technology wishlist.)
But I'd like to give my skeletal thoughts a bit more meat, and I
figure you guys are the right folks to ask for ideas on how to flesh
things out.
- All of the "killer apps" of the first generation web (Google,
Amazon, EBay,) except for MapQuest have started down the path of
turning themselves into platforms, rather than just applications --
except MapQuest and its imitators.
- What's more, Google, Amazon and EBay all leverage the behavior that
Clay Shirky outlined (I think this was in his talk Listening to
Napster but I can't find it in the published paper), namely, that the
system is architected in such a way users build the database as a
side effect of their individual "selfish" pursuits, rather than being
paid or volunteering. It's a bit more of a stretch with google, but
you can argue that pagerank is this kind of thing. Once again,
mapquest and its imitators are the only ones left out.
- Of all the internet killer apps, MapQuest is also the only one that
didn't become dominant. (maps.yahoo.com is run by vicinity, after
they dumped mapquest).
If we were to envision a next generation, collaboratively-enhanced
version of MapQuest, or Maps.yahoo.com, or mapinfo, how might we do
it? What features would lead people to naturally annotate maps?
What hacker work has already been done in this area? (I know there's
been some stuff naming cell towers so you can set alerts on them in
your phone, and there was the whole virtual london kind of thing) but
if we were really to brainstorm an ideal service that made it easy to
extend with the kind of rich commentary and added value that you see
in Amazon and EBay, what would it look like, and who's doing
interesting prototypes.
Thanks for your help.
------------------------------------------------------------------------
---------------
1005 Gravenstein Highway North, Sebastopol, CA 95472
707-827-7000
http://www.oreilly.com (company), http://tim.oreilly.com (personal)
_______________________________________________
Geowanking mailing list
http://lists.burri.to/mailman/listinfo/geowanking

[Anselm Hook]

One of the blocking issues seems to be GML and OpenGIS itself - insofar as
they act as a bit of a tarpit for errant hackers who are interested in the
space. These are fine standards but tend to dominate where other lighter
standards might do better.

If you look at how well the blog space is taking off you see fairly
lightweight standards that tend to be pretty easy to play with - a lot
more people are able to get their feet dirty without getting mired down.

Actually it really does seem like these barriers to having a richly geo
annotated internet are pretty much history... if one views it from the
perspective of rdf/rss. Services like typepad will undoubtably begin to
offer location as a property attribute of blog posts and probably more and
more people will begin to decorate their RDF graphs with location
information. It looks like the real remaining barriers are just adoption
of the pattern.

What still impresses me about rss/rdf in general is just how robust an
idea it is. I can remember reading Berners Lee's article in Scientific
American and just not getting why it would be valuable - but now I see
that if you turn the entire internet into a database you get a system that
with vastly superior characteristics to that of any single hosted service.
You get peer to peer, scaleability, better trust, retention of ownership,
division of roles etcetera. The one hassle with distributed content
serving is that you need beefy aggregators to aggregate, organize and
present the content - but this provides an opportunity for services such
as feedster or google to add special value... and it doesn't diminish the
public weal.

Ben Russell raised the possibility of some low hanging fruit in the geo
app space that could be a 'killer app'... the idea of a geolocal stuff
trading service - where people would geoblog things that they wanted to
buy or sell and the service could have hordes of little software agents
scampering around the internet trying to match up complementary interests.

An app I kind of have a soft spot for (being so shocked by some of the
things that criminals get away with sometimes) would be a forensic crime
reconstruction tool where police officers could have all the citizens in a
neighbourhood blog everything they could remember about a point and
location in time. If the crime was particularily heinous and the evidence
unusually slim then perhaps a peek into the shared knowledge commons could
elucidate a few clues.

- a

[Matthew Mullenweg]

Post by Anselm Hook
Services like typepad will undoubtably begin to
offer location as a property attribute of blog posts and probably more and
more people will begin to decorate their RDF graphs with location
information.

Products like WordPress currently offer this. :)

--Matt
http://photomatt.net || http://wordpress.org

[Mike Liebhold]

Post by Tim O'Reilly
If we were to envision a next generation, collaboratively-enhanced
version of MapQuest, or Maps.yahoo.com, or mapinfo, how might we do
it? What features would lead people to naturally annotate maps?

Hello Tim,

Here are a few quick thoughts on your questions---

Maps are the currently dominant way to view spatial information, but
there may be others ( e.g. simple visible or audible 'proximity alerts'
for a selected quality of a place.)

I believe the most important constraint on the widespread use of spatial
information is the tautological lack of understanding of intrinsic
values of viewable spatial information. Right now lots of geo-hackers
are building map things just because maps are cool, and fun to hack -
not necessarily because anyone has any stunning vision of how this all
will make life better. Similarly In the beginning days of hypermedia,
we had Xanadu, and Memex as fuzzy models , but little else, but we built
Hypercard, and then the web anyway, and then along the way, almost after
the fact, people figured out how to do cool things with the web.Spatial
hypermedia may develop similarly. People just don't understand the
value of 'information in place', mainly becuase they haven't seen much
that's really interesting. As more cool demos are developed, people will
begin to engage.

There are a few important developments that are clearly prerequisate:

1.Ease of combining disparate data sets into one layered view, not a
separate map for each discrete data set.
- standard geocoded legacy web pages,
- new standard geocoded locative media,
- new standard opengis/wfs/gml data.
- non-standard legacy geospatial data

2. Geospatial Search engines, web services, repositories, and libraries
with common searchable APIs

3. Public access to geospatial data. We have great access here in the
US, and Canada but in Brittain and elsewhere, geodata is tightly
controlled by the government, if it exists at all.

4. Ubiquitious location sensing networks and software. Telephone
companies can locate a cell phone for emergencies, but, so far, don't
generally provide location information to users. Academics, hackers ,
and private ventures have figured out how devices can locate themselves
by triangulating nearby wi-fi base stations, but there are not yet and
widely availble databases of known wifi Access point locations. It is
also possible to locate a device by triangulating digital TV towers, and
perhaps radio stations.

5. Client side privacy utilities for users to mask their location from
the network.

6. Filtering tools to weed out 'spatial spam' ( documents that have
been falsely geotagged wiht many location coordinates)

7. etc...

Cheers,

Mike Liebhold
www.starhill.us

Post by Tim O'Reilly
What hacker work has already been done in this area? (I know there's
been some stuff naming cell towers so you can set alerts on them in
your phone, and there was the whole virtual london kind of thing) but
if we were really to brainstorm an ideal service that made it easy to
extend with the kind of rich commentary and added value that you see
in Amazon and EBay, what would it look like, and who's doing
interesting prototypes.
Thanks for your help.
------------------------------------------------------------------------
---------------
1005 Gravenstein Highway North, Sebastopol, CA 95472
707-827-7000
http://www.oreilly.com (company), http://tim.oreilly.com (personal)
_______________________________________________
Geowanking mailing list
http://lists.burri.to/mailman/listinfo/geowanking

[Raj Singh]

I think you're asking the right questions. I've had similar laments.
I'll try to answer without adding too many more questions.

1. Many people argue (like Rob Hranac just did) that geospatial content
is more difficult to manage, and the technologies/standards haven't
been there in the past for killer apps to be created. Is this true, or
have geospatial people been too proud to accept 'good enough'
technologies and are instead waiting for perfection?

2. It's tough to get people to add content--whether its through
'selfish' pursuits or otherwise. Geographic information is usually very
personal, and not always for obvious reasons. For example, Amazon will
often tell me, "people who bought X also shopped for Y." How would I
feel if they said "people who live in Cambridge, MA also bought Z this
month." That might be creepy. How would you feel?

3. Geocomputation is just plain expensive. For example, to draw the
boundary of a coastal state like Massachusetts, the computer may have
to parse tens of thousands of floating point numbers, translate them
into screen coordinates, color the shapes and draw them on the
screen--and users want this to happen in tenths of a second. This is a
far cry from rendering HTML or XML, and makes it tough to produce
cheap, commodity, easy to use/install/maintain software.

All that being said, I'm trying hard to overcome these challenges
as a volunteer on the Community Mapping Project:
http://mapbuilder.sourceforge.net/
and professionally on an MIT research project:
http://web.mit.edu/crcp/prinstru.html#mapping

--Raj

Post by Tim O'Reilly
In a number of talks over the past few years, I've lamented the fact
that MapQuest has never figured out how to become a platform. I
thought I'd repeat that lament as part of my keynote at our upcoming
Emerging Technology Conference (http://conferences.oreilly.com/etech).
(the talk will focus on my technology wishlist.)
But I'd like to give my skeletal thoughts a bit more meat, and I
figure you guys are the right folks to ask for ideas on how to flesh
things out.
- All of the "killer apps" of the first generation web (Google,
Amazon, EBay,) except for MapQuest have started down the path of
turning themselves into platforms, rather than just applications --
except MapQuest and its imitators.
- What's more, Google, Amazon and EBay all leverage the behavior that
Clay Shirky outlined (I think this was in his talk Listening to
Napster but I can't find it in the published paper), namely, that the
system is architected in such a way users build the database as a side
effect of their individual "selfish" pursuits, rather than being paid
or volunteering. It's a bit more of a stretch with google, but you
can argue that pagerank is this kind of thing. Once again, mapquest
and its imitators are the only ones left out.
- Of all the internet killer apps, MapQuest is also the only one that
didn't become dominant. (maps.yahoo.com is run by vicinity, after
they dumped mapquest).
If we were to envision a next generation, collaboratively-enhanced
version of MapQuest, or Maps.yahoo.com, or mapinfo, how might we do
it? What features would lead people to naturally annotate maps?
What hacker work has already been done in this area? (I know there's
been some stuff naming cell towers so you can set alerts on them in
your phone, and there was the whole virtual london kind of thing) but
if we were really to brainstorm an ideal service that made it easy to
extend with the kind of rich commentary and added value that you see
in Amazon and EBay, what would it look like, and who's doing
interesting prototypes.
Thanks for your help.
-----------------------------------------------------------------------
----------------
1005 Gravenstein Highway North, Sebastopol, CA 95472
707-827-7000
http://www.oreilly.com (company), http://tim.oreilly.com (personal)
_______________________________________________
Geowanking mailing list
http://lists.burri.to/mailman/listinfo/geowanking

[Anselm Hook]

oh, one perhaps totally nutty geo related idea:

the idea of a real sim-city has been raised so often by so many people...
i think i heard it first from S0ren Gr0nbech from Denmark. But it is an
endearing concept:

it would be nice to have a community geoblog everything they knew about
their neighbourhood with an emphasis on any algorithms or relationships
that they noticed and could quantify... coarse approximations would be
ok, and the emphasis would be not just on environmentally connected
observations but on legal ones - say the perceived repercussions of the
enactment of a particular law...

then a digital simulation could synthetically integrate and project
forward the entire state of that community into the future by some point
in time... and perhaps it could be used as a tool by lawmakers to see
what the side-effects were of some proposed legal action...

the great thing is that it would put corporations on par with
individuals... and argument would shift to the merits of each algorithm
rather than simply purely argumentative stances...

often (if not always) the side-effects of any law seem to always dominate
over the intended effects... housing projects become ghettos and spacious
country homes become inhuman suburban deserts... the world is too complex
to understand despite the hubris of politicians... and a digital
projection - even a short term one - might be more useful than the
unverifiable argumentative positions or demagogery we put up with these
days...

- a

[Chris Holmes]

Post by Tim O'Reilly
If we were to envision a next generation, collaboratively-enhanced
version of MapQuest, or Maps.yahoo.com, or mapinfo, how might we do it?
What features would lead people to naturally annotate maps?
What hacker work has already been done in this area?

It's not quite hacker work, but one interesting effort that I don't think
has been brought up on geowanking yet is project OneMap
(http://onemap.org), by some academics in Norway.
http://onemap.org/gw3/Rationale.html provides some good information,
basically they are making tools to create open collaboratively built
geodata, using open source and open standards. They also seek to have
'open management', some sort of distributed, non-hierarchical project
organization. It has a long way to go, but if successful it could be very
useful, as open and freely available base data seems essential to making
this stuff happen.

Chris

(I know there's

Post by Tim O'Reilly
been some stuff naming cell towers so you can set alerts on them in
your phone, and there was the whole virtual london kind of thing) but
if we were really to brainstorm an ideal service that made it easy to
extend with the kind of rich commentary and added value that you see in
Amazon and EBay, what would it look like, and who's doing interesting
prototypes.
Thanks for your help.
------------------------------------------------------------------------
---------------
1005 Gravenstein Highway North, Sebastopol, CA 95472
707-827-7000
http://www.oreilly.com (company), http://tim.oreilly.com (personal)
_______________________________________________
Geowanking mailing list
http://lists.burri.to/mailman/listinfo/geowanking

--

[Philip Abrahamson]

Hello Tim,

I work on mobilemaps.com which is superficially an open source
alternative to Google's Search by Location demo (we've been around
longer - which doesn't say much for our marketing:), but we believe it
has platform potential.

The critical factor is that it plots Web based data on the maps rather
than the privately compiled "point of interest" data or Yellow Pages
data found on the likes of MapQuest and other map sites.

This offers the opportunity for users to publish their own data to the
system, and effectively create their own application. We have
experimented with a public list of urls to which users submit their web
pages. The pages will have had a pair of geographic meta-tags added by
the user which we use to plot the location of the page while indexing
the page for the search functionality.

The obvious application is finding local information, but other
applications could include more real-time updates of location, such as
publicly tracking the location of someone or some object.

Why hasn't this been done by existing mapping companies? Maybe they are
too used to quality controlled geographic data to want to put something
unstructured on their systems. They prefer the closed, quality
controlled Yellow Pages data.

We do have a prospective business model based on location based text
advertising. A geographic extension of Google's/Overture's ad models.

Regards,
Philip Abrahamson

***@mobilemaps.com
http://www.mobilemaps.com

Post by Tim O'Reilly
In a number of talks over the past few years, I've lamented the fact
that MapQuest has never figured out how to become a platform. I
thought I'd repeat that lament as part of my keynote at our upcoming
Emerging Technology Conference (http://conferences.oreilly.com/etech).
(the talk will focus on my technology wishlist.)
But I'd like to give my skeletal thoughts a bit more meat, and I figure
you guys are the right folks to ask for ideas on how to flesh things out.
- All of the "killer apps" of the first generation web (Google, Amazon,
EBay,) except for MapQuest have started down the path of turning
themselves into platforms, rather than just applications -- except
MapQuest and its imitators.
- What's more, Google, Amazon and EBay all leverage the behavior that
Clay Shirky outlined (I think this was in his talk Listening to Napster
but I can't find it in the published paper), namely, that the system is
architected in such a way users build the database as a side effect of
their individual "selfish" pursuits, rather than being paid or
volunteering. It's a bit more of a stretch with google, but you can
argue that pagerank is this kind of thing. Once again, mapquest and
its imitators are the only ones left out.
- Of all the internet killer apps, MapQuest is also the only one that
didn't become dominant. (maps.yahoo.com is run by vicinity, after they
dumped mapquest).
If we were to envision a next generation, collaboratively-enhanced
version of MapQuest, or Maps.yahoo.com, or mapinfo, how might we do it?
What features would lead people to naturally annotate maps?
What hacker work has already been done in this area? (I know there's
been some stuff naming cell towers so you can set alerts on them in
your phone, and there was the whole virtual london kind of thing) but
if we were really to brainstorm an ideal service that made it easy to
extend with the kind of rich commentary and added value that you see in
Amazon and EBay, what would it look like, and who's doing interesting
prototypes.
Thanks for your help.
------------------------------------------------------------------------
---------------
1005 Gravenstein Highway North, Sebastopol, CA 95472
707-827-7000
http://www.oreilly.com (company), http://tim.oreilly.com (personal)
_______________________________________________
Geowanking mailing list
http://lists.burri.to/mailman/listinfo/geowanking

[Chris Heathcote]

Hi

Post by Tim O'Reilly
What's more, Google, Amazon and EBay all leverage the behavior that
Clay Shirky outlined (I think this was in his talk Listening to
Napster but I can't find it in the published paper), namely, that the
system is architected in such a way users build the database as a side
effect of their individual "selfish" pursuits, rather than being paid
or volunteering.

In my mind, this is key. I hate to be English about this, but this is a
big social conundrum to solve, not just a technical one. Give people a
reason to contribute. This is hard on the Internet - in the real world,
Zagat, for example, give you a free guide for filling in a review form.
The trouble is that there are very few pure-play consumer GIS services
that people want - geodata normally enhances other services (what would
eBay be like, for example, if it actually understood where sellers and
buyers were?).

The technical problem is the proprietary data. Amazon would be nothing
without its extensive book database. Geo services need a base level of
accurate data (including hard stuff, like roads). I've been played with
TIGER, and the US is far better off than most countries, but the
'difficult' stuff is still only kept up to date by a few companies.

A platform is definitely what is needed - or more likely, small pieces
loosely joined by a common data understanding. Example services needed
are geo-tagged data stores for text, pictures, sound, video, mapping
and map annotation services, and some machinery to turn real-world
places ("Bob's Big Burger Bar, 500 Lost Highway") into something
computers can deal with (-124.566, 32.567).

A platform that can link all these pieces together (understand and
advertise what and where each location-based service is good at) would
allow service creators to plug geography into their apps without having
to learn the intricacies of GML, geo licensing etc. etc.

c.

[Frank Leahy]

Hi Tim,

On Monday, January 26, 2004, at 10:47 PM,

Post by Tim O'Reilly
If we were to envision a next generation, collaboratively-enhanced
version of MapQuest, or Maps.yahoo.com, or mapinfo, how might we do it?
What features would lead people to naturally annotate maps?

I've just moved from Sausalito to Cornwall, and am interested in being
able to place my photographs on a map. For example, when I walk around
Land's End in the Spring I want to show a map that has little
thumbnails of each photo with links to the individual weblog entries
that I write.

I also want to make my photos more searchable, which is why I wrote
"How To Make Photos More Searchable" at
http://cornwall.backtalk.com/archives/000062.html. (I'm currently
working on this, and am hoping to have something working by
mid-February.) But what I really want in this arena is 1) for Google
to recognize geoURL data attached to my weblog entries, as well as to
each picture, and 2) for Google to show search results that have geoURL
data on a map (or at least provide the geoURL data so the results could
be mapped).

I wound up purchasing Microsoft's AutoRoute (the European version)
because it seemed to be the only product that let me place multiple
push-pins on a map (I haven't figured out whether it's possible to run
the program programatically -- anyone know if that's possible? -- which
is why there aren't any maps on my weblog yet) But I only purchased it
because I couldn't find an online product, which I would be quite
willing to pay for.

Hope this helps,
-- Frank Leahy
Weblog: http://cornwall.backtalk.com

[Sonny Parafina]

Some ruminations after seeing all the previous posts.

There seems to be two issues here. First is the creation of
geotagged/geolocated content. Second, the lack of a platform that makes use
of geotagged content.

While geoURL is a nice and compact way of geotagging content, populating the
coordinates requires more effort than a conventional reference such as an
address. For example, when blogging about a fantastic nouvelle Nepalese
restaurant, its much easier to cut and paste an address from a phone book
look up, instead of trying to give the entry a geoURL. To cut to the chase,
there are many forms of geographic reference that much easier to use than a
coordinate pair. Mechanisms for autogenerating geotagged content need more
development.

This segues into the second issue, the lack of a suitable platform. When you
look at most web mapping sites, what they do is render a set of data into a
map. They are essentially cartographic engines and not spatial engines.
That is, they lack spatial operations that allow folks to use them as a
platform for building the killer app. The OpenGIS standards are directed
towards building a complete platform, which is one reason why they seem
overly complex. For example, there is a discussion paper (long languishing)
on a geocoder interface that can address multiple location reference
systems.

While I agree with Allan Doyle and Rob Hranac about the need for standard
interface for geodata and geo-applications, I tend to find that OpenGIS
specifications focus too much on the location aspect to the point that the
location is the only object and other information is just an attribute. I
think this is the origin of the dissonance between the RDF folks and the
OpenGIS/mapping folks.

Below is a URL documenting a project I managed several years ago. While it
is certainly not a killer app, it does try to implement the concept of what
we called a Location Organizer Folder or LOF using the OpenGIS standards at
the time. The idea was that individuals would be able to send an email, web
page, or any bit of text and have the information geolocated automagically.
The user would then be able to pick out the bits that interested them and
save them into a "folder" which they could share with other individuals.

http://newgate.socialchange.net.au:8080/gfspilot/docs/index.html

regards,

sonny






-->-----Original Message-----
-->From: geowanking-***@lists.burri.to
-->[mailto:geowanking-***@lists.burri.to]On Behalf Of Tim O'Reilly
-->Sent: Monday, January 26, 2004 2:37 PM
-->To: ***@lists.burri.to
-->Cc: Clay Shirky; Rael Dornfest
-->Subject: [Geowanking] LazyWeb request
-->
-->
-->In a number of talks over the past few years, I've lamented the fact
-->that MapQuest has never figured out how to become a platform. I
-->thought I'd repeat that lament as part of my keynote at our upcoming
-->Emerging Technology Conference (http://conferences.oreilly.com/etech).
--> (the talk will focus on my technology wishlist.)
-->
-->But I'd like to give my skeletal thoughts a bit more meat, and I figure
-->you guys are the right folks to ask for ideas on how to flesh things
-->out.
-->
-->Here's my premise:
-->
-->- All of the "killer apps" of the first generation web (Google, Amazon,
-->EBay,) except for MapQuest have started down the path of turning
-->themselves into platforms, rather than just applications -- except
-->MapQuest and its imitators.
-->
-->- What's more, Google, Amazon and EBay all leverage the behavior that
-->Clay Shirky outlined (I think this was in his talk Listening to Napster
-->but I can't find it in the published paper), namely, that the system is
-->architected in such a way users build the database as a side effect of
-->their individual "selfish" pursuits, rather than being paid or
-->volunteering. It's a bit more of a stretch with google, but you can
-->argue that pagerank is this kind of thing. Once again, mapquest and
-->its imitators are the only ones left out.
-->
-->- Of all the internet killer apps, MapQuest is also the only one that
-->didn't become dominant. (maps.yahoo.com is run by vicinity, after they
-->dumped mapquest).
-->
-->So, my question for you is this:
-->
-->If we were to envision a next generation, collaboratively-enhanced
-->version of MapQuest, or Maps.yahoo.com, or mapinfo, how might we do it?
--> What features would lead people to naturally annotate maps?
-->
-->What hacker work has already been done in this area? (I know there's
-->been some stuff naming cell towers so you can set alerts on them in
-->your phone, and there was the whole virtual london kind of thing) but
-->if we were really to brainstorm an ideal service that made it easy to
-->extend with the kind of rich commentary and added value that you see in
-->Amazon and EBay, what would it look like, and who's doing interesting
-->prototypes.
-->
-->Thanks for your help.
-->------------------------------------------------------------------------
-->---------------
-->Tim O'Reilly @ O'Reilly & Associates, Inc.
-->1005 Gravenstein Highway North, Sebastopol, CA 95472
-->707-827-7000
-->http://www.oreilly.com (company), http://tim.oreilly.com (personal)
-->
-->_______________________________________________
-->Geowanking mailing list
-->***@lists.burri.to
-->http://lists.burri.to/mailman/listinfo/geowanking

[Allan Doyle]

Post by Sonny Parafina
Some ruminations after seeing all the previous posts.
While I agree with Allan Doyle and Rob Hranac about the need for standard
interface for geodata and geo-applications, I tend to find that OpenGIS
specifications focus too much on the location aspect to the point that the
location is the only object and other information is just an attribute. I
think this is the origin of the dissonance between the RDF folks and the
OpenGIS/mapping folks.

Oddly enough, I didn't use the word "standards" in my message even
once :)

In fact, I come down firmly on the grass-roots side of the
grass-roots vs. OGC debate.

Post by Sonny Parafina
While geoURL is a nice and compact way of geotagging content, populating the
coordinates requires more effort than a conventional reference such as an
address. For example, when blogging about a fantastic nouvelle Nepalese
restaurant, its much easier to cut and paste an address from a phone book
look up, instead of trying to give the entry a geoURL. To cut to the chase,
there are many forms of geographic reference that much easier to use than a
coordinate pair. Mechanisms for autogenerating geotagged content need more
development.

Your point about coordinates perhaps being inappropriate as a user
interface element is spot-on. Coordinates should remain hidden or at
least should not get in the way. A few days ago someone asked about a
web service that would take in a location string and produce XML with
the location encoded in it. That's the kind of service we need.

A real long time ago there was a little discussion about
c-squares. The tantalizing thing about that is that you might be able
to cobble together some spatial search functions using nothing more
than grep.

Post by Sonny Parafina
Below is a URL documenting a project I managed several years ago. While it
is certainly not a killer app, it does try to implement the concept of what
we called a Location Organizer Folder or LOF using the OpenGIS standards at
the time. The idea was that individuals would be able to send an email, web
page, or any bit of text and have the information geolocated automagically.
The user would then be able to pick out the bits that interested them and
save them into a "folder" which they could share with other individuals.
http://newgate.socialchange.net.au:8080/gfspilot/docs/index.html

I think the concepts in the Location Organizer Folder are nice, but
maybe they can be merged with RSS so that an RSS feed can hold
similar kinds of things.

Frank's Cornwall photos on an RSS feed are the right idea. Which, I
think brings us right back to RDF. (Which I guess I better learn more
about!)

[Raj Singh]

I've been trying to articulate my problems with the direction of OGC
specs for awhile, but haven't been able to find the right words until
now. IMHO, the most important technology principle is, "simple things
should be easy to do--complex things may be harder." OGC does a good
job of letting us do complex things, but the technology doesn't scale
down to make simple things easy.

--Raj

Post by Sonny Parafina
The OpenGIS standards are directed
towards building a complete platform, which is one reason why they seem
overly complex.

[Phil Gyford]

Post by Tim O'Reilly
If we were to envision a next generation, collaboratively-enhanced
version of MapQuest, or Maps.yahoo.com, or mapinfo, how might we do
it? What features would lead people to naturally annotate maps?

Most of the responses so far seem to have been the usual geowanking
thoughts about how to open things up, and the difficulties involved
in sharing and distributing geographical data. I'm not sure this is
really the issue here though. It's like discussing how to make it
easy for people to obtain their own database of books, rather than
how Amazon can use its database (and data gleaned from customers) in
new ways.

I assumed the question was not "how can MapQuest make the world a
better place for geowankers by sharing its data", but "how can
MapQuest make its service better by using data gleaned from users."
So let's think about what Mapquest (or whoever) could do on their
site before getting into all the usual stuff about semantically
sharing our collaborative meta-blogs or whatever.

Perhaps the biggest hurdle is that to get any useful information
Mapquest would have to get people to sign up for a free account, so
it can identify individuals. There's no point in people signing up at
the moment, but maybe they can be persuaded to...

Once we have signed-up users, what data can we get from their
behavior? Mainly, where they search for - why not store those? Is
there any useful datamining that can be done? If a user goes to the
'print' version of a map, that's presumably somewhere they're
visiting so that location is probably more important than every other
map they've viewed. Does that tell us anything?

What extra activity could people be enticed into? One day, someone
will discover a real use for social networking sites, and maybe this
is it. People certainly seem eager to sign up for them (if we ignore
any "Oh no, not another one" aversion).

Think of Matt Jones' idea about combining social networks with Amazon
<http://www.blackbeltjones.com/work/mt/archives/000842.html> How
about combining them with maps? Instead of asking people their
favorite books, films, etc, ask them where their favorite restaurants
and bars are. I already know what kind of movies my first degree of
friends are into, but I'd really like to know what restaurants they
recommend.

So, let me overlay all the maps I view with markers indicating places
that my friends recommend. Privacy issues aside for a moment, show me
where my friends live ("I'm going to visit this guy I know online,
let me just click his name in the sidebar and see a map to his
house").

Then you have "People who liked this bar also liked...", "People who
live near here recommend these restaurants...", etc, etc.

I'm sure there's plenty more we can think of, without getting into
the difficulties of understanding WGS84 or whatever - no one has to
understand the mechanics of Google's search engine to use their
services, or the publishing/retail worlds to use Amazon's.

Phil

(Also sent to Tim, Clay and Rael who may not be on the list?)

[Douwe Osinga]

Post by Allan Doyle
Coincidentally, blogdex shows a new site http://www.world66.com/
that's being blogged about in a lot of places. Trouble is, the site
has been down all day so I have not had a chance to look at it
yet. The descriptions of it make it sound like some kind of
open-content data collecting site. Who knows. It might be your killer
app if they get their act together... Google does have some cached
views of the site.

I happen to operate that site. Incidentially, the reason it is down
has to do with mapping; we posted a little app that let you put a map
of the world on your blog with the countries visited showing up in red.
A lot of people seemed to enjoy that, taking down the site after three
days of exponential visitor growth.

The idea behind world66 is not primarily map oriented; we just want to
have people collectively write a travel guide, sort of Wikipedia meets
Lonely Planet. Geo data is part of what we're trying to collect. So far
it has not been our main focus, but the success of the visited countries
thing proves that people like maps and they do cooperate if there's
something in it for them. Of course here the data we collect is basically
which countries are the most popular as holliday destinations, which
is maybe not that interesting, but it is something.

More related to this discussion. We we're thinking of creating a little
flash showing a map of a city + the favourite bar, restaurant, cinema
etc for one person. This person can show this thing on his website as
a 'my city' page. He can edit/add the bars/restaurants/cinema and the
map. Of course underneath the flash thing, there'll be a link to the
page where you can create these things, so more people will do it. If
enough visitors cooperate, you'll end up with a nice collection of geo
tagged data. The database is not distributed, but open content, so anybody
can build other applications on top of it.

Letting people put the location of bars on a pre-drawn city map is
of course one thing. Letting them draw the map itself from a browser
environment is another. We have been experimenting a bit with a flash
app that does just that, but you'll end up with at best an outline of
cityblocks/roads/parks. And it is probably very hard for random people
to draw city maps. More visitors might incrementally improve the quality
of maps, but it is less an obvious choice for open content than pictures
of your favourite travel destination.

There are of coure also issues as to who gets to decide where Ben's Burger
bar exactly is (the last person that put it on a map, some kind of average
of every person that positioned it) and how to map the map-space to world-
space, but that could probably be worked out later.

I'll be interested in your thoughts about this idea.

Douwe Osinga
http://douweosinga.com

[Richard Fairhurst]

Further to the discussion, we have a site with similar intent at

http://www.geowiki.com/

--
| Richard Fairhurst www.systemeD.net
| London is deluding itself if it thinks it can continue to
| dominate national consciousness: the centre is anywhere and
| everywhere, especially Uppingham - Iain Sinclair

[Clay Shirky]

Post by Tim O'Reilly
- All of the "killer apps" of the first generation web (Google, Amazon,
EBay,) except for MapQuest have started down the path of turning
themselves into platforms, rather than just applications -- except
MapQuest and its imitators.

Another line of thought: many of the killer apps had their handle problem
worked out in advance ISBNs for Amazon, PNRs for Expedia, URLs for Google,
and all of these identifiers were public. Geo-data, on the other hand, has a
history of ownership or governmental control (for reasons both economic and
political), so it may have been hard to rationalize a service model, with
lock-in so nicely handled by the upstream data sources themselves.

Post by Tim O'Reilly
What features would lead people to naturally annotate maps?

Credit, for starters. The karma engine of the Amazon book reviews is
astonishing.

Also: Make it a game, like geo-caching; or provide 'hook' services, so that
a group that wanted to mark up its neighborhood could advertise that markup
to the world; easy integration with services like World66 or movie location
scouts, so that the 'sponsorship of annotation' problem was shared by end
users.

-clay

[Rich Gibson]

Phil breaks this down into two parts. "How can MapQuest make its service
better by using data gleaned from users." And the part he discounts, "how
can MapQuest make the world a better place for geowankers by sharing its data"

My first answer to question one is to accept user feedback. I'd love for
that to be full blown collaborative cartography. "Users who traveled
between these waypoints have suggested these alternatives." And the
alternatives would be catagorized as 'corrections,' where the MapQuest
route just didn't work, 'optimizations,' where there is a faster route,
and some long series of scenic or theme based detours.

The error component in MapQuest data is especially apparant for those of
us who live in Sebastopol. If you know something about the roads up here,
then pull of these travel directions from my house to a friend near
Oakland.

http://www.mapquest.com/directions/main.adp?go=1&do=nw&cl=EN&un=m&ct=NA&1y=US&1a=&1c=&1s=&1z=&1ah=yVEiFc5BCGiRr48a22HVQTgIU%252bAFpesb6x%252bXChkoXcaXemWKx%252bR1Cdm8lCeTc2MvWX0YOMYlQpwbtLDZImAo0Q%253d%253d&2y=US&2a=&2c=&2s=&2z=&2ah=6LmnwPYnPjfJuAUrjgh3M8iBeXdCjdKzQnn8W6bTjNr84dxncu6%252b%252bNCcGOAkBFogy1jHZVPA%252fxgMo7F%252fA5q7%252bQ%253d%253d&formtype1=address&formtype2=address&idx=0&id=4016a5ec-0038e-031d3-400c330e&aid=4016a5ec-0038f-031d3-400c330e

Llano raod to get from Sebastopol to 101? MapQuest would be better if
there was a systemic way to correct these sorts of errors.

I would also like user supplied data to lead me to cool things along the
way, something in addition to 'show me the McDonald's near my journey,'
but showing user offered alternatives is my strongest example.

As for Phil's second point, I believe that the idea of MapQuest as
platform requires some form of shared data. A better MapQuest would be
fun, but MapQuest as usable platform would be revolutionary.

Cheers,
Rich

[Bill Kearney]

Post by Rich Gibson
Llano raod to get from Sebastopol to 101? MapQuest would be better if
there was a systemic way to correct these sorts of errors.

I'd be fascinated hearing whether or not the mapping databases take
'socio-economic' factors into account. As in, don't give user from area X
directions through neighborhood Y because it's a crime-ridden area. Talk about
hack potential, start giving people with Lexus or other high value vehicles
directions right into ambush territory. Or maybe I've had too much coffee this
morning...

Post by Rich Gibson
I would also like user supplied data to lead me to cool things along the
way, something in addition to 'show me the McDonald's near my journey,'
but showing user offered alternatives is my strongest example.

Me, I'd rather have /my own database agent/ understand geopositioning such that
it could nag me. As in, "Bill, you're near the dry cleaner and you've still not
picked up those shirts that were ready 8 days ago". Or, "there's a new Thai
place near here and you've not had a Thai dinner for several months now..."

There's no flipping way I've any hope whatsoever of this data being 'trusted' in
the hands of the demographers, marketers and other (expletives) masquerading as
professionals in their related fields. I know they're interested in making a
buck and I've a fair amount of respect for some of the tangental ways they can
extrapolate this from indirect data. But they've thus far demonstrated such a
remarkable degree of spineless contempt for the individual that they've little
or no hope of getting a shot at doing this sort of thing with any live genuine
data.

-Bill Kearney

[Phil Gyford]

Post by Rich Gibson
Phil breaks this down into two parts. "How can MapQuest make its service
better by using data gleaned from users." And the part he discounts, "how
can MapQuest make the world a better place for geowankers by sharing its data"

I don't mean to "discount" that part, but it seemed like everyone was
concentrating on that and ignoring the first part. Which is only to
be expected among geowankers I suppose. But the first part seems more
likely to create things that non-geowankers might find useful.

[Robert Woodard]

Just my 2 cents:

The geo killer app? I'm not really a geowanker so forgive me if my
terminology isn't correct.

[Allan Doyle]

Post by Robert Woodard
The geo killer app? I'm not really a geowanker so forgive me if my
terminology isn't correct.

[Raj Singh]

The key here seems to be Robert's Google vs. Allan's 'little enclaves'.
A strong, market leader like Google, Amazon or Ebay doesn't
'founder'--they execute.
It would help us if a highly visible market leader with relatively deep
pockets took
some risks and led the way here. Someone who already has all the data,
infrastructure, etc.

[Tony Pryor]

Some of the discussion issues (please add or replace):

The Intervisibility of Development Efforts Operating in Distant Vacuums
Data Layer Slurping / Media Servers / OGC
Transparent Indexability based on an Atomic Standard
The Killer App Question

[Tony Pryor]

Post by Raj Singh
The key here seems to be Robert's Google vs. Allan's 'little enclaves'.
A strong, market leader like Google, Amazon or Ebay doesn't
'founder'--they execute.
It would help us if a highly visible market leader with relatively
deep pockets took
some risks and led the way here. Someone who already has all the
data, infrastructure, etc.

There is an initiative fomenting among key chairs at Harvard/MIT for a
National Historical Database.

They at least have Access to deep pockets.

[Raj Singh]

I'm working on geospatial file sharing software. It's smarter than
existing stuff because it keeps track (through metadata) of the
original data source and can query that source on demand to check for
updates. Search will be keyword, title and geography based. It's not
perfect, but I hope it plant some seeds for better programmers to build
upon.

--Raj

Who is working on collaboration in a social/mobile/gps context, and
what kind of collaboration?
Not one more detail about maps themselves for the time being.

[Tony Pryor]

Sounds brilliant. Hate to badger you- but are you building any
trust/privacy levels into it? Encryption? Superservers? Proximal
aggregation pushes? Redundancy? I have more questions. :P

BTW, a slew of frontierish geo folks from different walks are getting
together at the Thirsty Ear this Monday at 8pm (
Loading Image...) to discuss paths to an image
geotagging initiative in the context of the National Historical Database
brainstorming session at the Mit Museum.

Interested?

If anyone else thinks they might be able to make it, please let me know.

-Tony

Post by Raj Singh
I'm working on geospatial file sharing software. It's smarter than
existing stuff because it keeps track (through metadata) of the
original data source and can query that source on demand to check for
updates. Search will be keyword, title and geography based. It's not
perfect, but I hope it plant some seeds for better programmers to
build upon.
--Raj

Who is working on collaboration in a social/mobile/gps context, and
what kind of collaboration?
Not one more detail about maps themselves for the time being.

_______________________________________________
Geowanking mailing list
http://lists.burri.to/mailman/listinfo/geowanking

[Raj Singh]

Post by Tony Pryor
Sounds brilliant. Hate to badger you- but are you building any
trust/privacy levels into it? Encryption? Superservers? Proximal
aggregation pushes? Redundancy? I have more questions. :P

nope. nope. nope. nope. Redundancy--kinda. Basically, what I get done
is a function of the number of students I have working on the project
(1), and the funder's need to have working, deliverable apps this
Spring.

--Raj

[Simon St.Laurent]

Post by Raj Singh
The key here seems to be Robert's Google vs. Allan's 'little
enclaves'. A strong, market leader like Google, Amazon or Ebay doesn't
'founder'--they execute. It would help us if a highly visible market
leader with relatively deep pockets took some risks and led the way
here. Someone who already has all the data, infrastructure, etc.

[Sonny Parafina]

-->-----Original Message-----
-->From: geowanking-***@lists.burri.to
-->[mailto:geowanking-***@lists.burri.to]On Behalf Of Allan Doyle
-->Sent: Tuesday, January 27, 2004 4:24 PM
-->To: ***@lists.burri.to
-->Subject: Re: [Geowanking] RE: LazyWeb request
-->

--> > 3) A Google like service that can tell me what 'layers' and
-->collections of
--> > 'points' are available to me. How much they cost, where to get
--> > them, etc.
-->
-->This is the sticking point. No one has quite figured out how to do
-->this yet and to actually do it on a wide scale. The efforts I've been
-->involved with have all foundered on the "service model". I.e. what is
-->the big picture in which these things swim. People keep trying to go
-->down the complexity path towards UDDI and similar schemes. No one has
-->tried defining a Googleable representation yet, nor have people tried
-->REST-based solutions Googleable or not. Thus the experiments tend to
-->happen in little enclaves and are not widely seen.
-->
--> >


Actually we have the catalog part (www.askthespider.com), as well as the
other parts WMS/WFS on the server side. On the client side, we have a
widget that implements the same catalog seach interface and lets the user
add the data to their client dynamically. Its about as close to a google
interaface over a catalog out there. What boogles my mind is the almost
fanatical adherence to Z39.50 by spatial data providers. It is the most
common interface to spatial data holdings but it is probably the least
useful because of its overly complex and poorly designed UI.

I'm not trying to plug my products, but we have everything that Woody
mentions, all built on OGC interfaces. This also includes things such as
annotation, transactional feature servers, and dynamic styling of features
(via Style Layer Descriptors) by the user, ability to serve and reproject
imagery through a Web Coverage Service and portray (select bands in a RGB
model and contrast enhancements) them through a Coverage Portrayal Service.
Our Studio product lets you even build a client in 5 minutes that takes
advantage of OGC services, this means you can use the client with any OGC
service, not just our servers. I'm starting to foam at the mouth so I'll
stop here.

Cheers,

sonny

[Philip Abrahamson]

Post by Sonny Parafina
I'm not trying to plug my products, but we have everything that Woody
mentions, all built on OGC interfaces.

While I have much admiration for the OGC standards when it comes to
mapping, I'll stick my neck out and say the standards aren't consumer
driven in some vital areas. The killer app isn't going to coalesce from
solely these standards.

For example, consumers aren't looking for categorised "features" such as
the OGC/GIS mentality suggests. I'll support this with an analogy: how
many people use dirctories on the Web, and how many use search engines?
Directories represent the categorised features of the OGC, and search
engines represent what people actually want - ways of finding
information inside complex objects - like Web pages.

The technical difference between these approaches cannot be brushed over
lightly. Finding nearby "features" is significantly different from
finding nearby information that matches consumers' keywords, and making
this scale requires a different approach.

Another example: the user interface is something the GIS community never
had to worry about. Traditionally GIS users are familiar with a
technical set of mapping tools and a few extra input boxes are assumed
to be perfectly acceptable. When it comes to consumers on the Web
however, the most successful user interface is a single box.

I disagree with the people who think the killer app will be created by
an existing mapping/GIS giant, and I think Tim originally was using
MapQuest by way of illustration. Innovation almost always starts at the
bottom and not the top, and the lack of it in this field has been caused
by all the financial barriers to those starting at the bottom - like the
expense of data (I live in the UK, but could I demo my application
there? No, I set up an American demo because America has free data),
and formerly the expense of good software to present this data (now the
open source MapServer helps alleviate this).

Here are some predictions, which are naturally biased towards the
software I work on.

A. Short term:

i. A base layer of maps will be provided by free OGC map servers, and
discovered from an OGC catalog server (although the sooner Z3950
disappears the better). However this will not be for performance
reasons, but simply because the existing map companies don't see the
opportunity quickly enough.

ii. The information that consumers will look for will be on geocoded Web
pages as provided by something like GeoURL, except it will be a public,
distributed list of geocoded Web pages, and be initially augmented by
geographic spidering of Web pages.

iii. The engine that finds the information will be a "Nearby Engine" - a
cross between a search engine's keyword index, and a spatial proximity
index. It will have a single box for input.

B. Long term:

i. The open source model of distributing software will win over the web
services approach of distributing services. Consumers will have the
mapserver software on their computers.

ii. Both Web data and map data will be distributed across clients and
servers, and updated continuously. Core data and indexes will reside on
consumers' machines, or at worst nearby machines if they have a mobile
device.

Philip Abrahamson
***@mobilemaps.com
http://www.mobilemaps.com

[Sonny Parafina]

OGC standards are not consumer driven, they are a set of technical
specifications based on a concensus approach from the industry, academia,
and public entities. In OGC, we have a saying, "We cooperate on standards,
and we compete on implementation." OGC standards are a framework for
building the killer app. I think that there is a general consensus that the
killer app is driven by the need to communicate with other users, for this
reason interoperability (i.e. OGC standards) will be part of the solution.

For the most part I agree with your predictions, and we have even
implemented the "mapserver" software on everyone's desktop paradigm for a
client (we call then personal WFS). I think that google will remain the
dominant spatial search engine for some time to come, until users start
differentiating between services and data.

sonny



-->-----Original Message-----
-->From: geowanking-***@lists.burri.to
-->[mailto:geowanking-***@lists.burri.to]On Behalf Of Philip Abrahamson
-->Sent: Wednesday, January 28, 2004 7:21 AM
-->To: ***@lists.burri.to
-->Subject: Re: [Geowanking] RE: LazyWeb request
-->
-->
-->Sonny Parafina wrote:
-->
-->>I'm not trying to plug my products, but we have everything that Woody
-->>mentions, all built on OGC interfaces.
-->>
-->While I have much admiration for the OGC standards when it comes to
-->mapping, I'll stick my neck out and say the standards aren't consumer
-->driven in some vital areas. The killer app isn't going to coalesce from
-->solely these standards.
-->
-->For example, consumers aren't looking for categorised "features" such as
-->the OGC/GIS mentality suggests. I'll support this with an analogy: how
-->many people use dirctories on the Web, and how many use search engines?
-->Directories represent the categorised features of the OGC, and search
-->engines represent what people actually want - ways of finding
-->information inside complex objects - like Web pages.
-->
-->The technical difference between these approaches cannot be brushed over
-->lightly. Finding nearby "features" is significantly different from
-->finding nearby information that matches consumers' keywords, and making
-->this scale requires a different approach.
-->
-->Another example: the user interface is something the GIS community never
-->had to worry about. Traditionally GIS users are familiar with a
-->technical set of mapping tools and a few extra input boxes are assumed
-->to be perfectly acceptable. When it comes to consumers on the Web
-->however, the most successful user interface is a single box.
-->
-->I disagree with the people who think the killer app will be created by
-->an existing mapping/GIS giant, and I think Tim originally was using
-->MapQuest by way of illustration. Innovation almost always starts at the
-->bottom and not the top, and the lack of it in this field has been caused
-->by all the financial barriers to those starting at the bottom - like the
-->expense of data (I live in the UK, but could I demo my application
-->there? No, I set up an American demo because America has free data),
-->and formerly the expense of good software to present this data (now the
-->open source MapServer helps alleviate this).
-->
-->Here are some predictions, which are naturally biased towards the
-->software I work on.
-->
-->A. Short term:
-->
-->i. A base layer of maps will be provided by free OGC map servers, and
-->discovered from an OGC catalog server (although the sooner Z3950
-->disappears the better). However this will not be for performance
-->reasons, but simply because the existing map companies don't see the
-->opportunity quickly enough.
-->
-->ii. The information that consumers will look for will be on geocoded Web
-->pages as provided by something like GeoURL, except it will be a public,
-->distributed list of geocoded Web pages, and be initially augmented by
-->geographic spidering of Web pages.
-->
-->iii. The engine that finds the information will be a "Nearby Engine" - a
-->cross between a search engine's keyword index, and a spatial proximity
-->index. It will have a single box for input.
-->
-->B. Long term:
-->
-->i. The open source model of distributing software will win over the web
-->services approach of distributing services. Consumers will have the
-->mapserver software on their computers.
-->
-->ii. Both Web data and map data will be distributed across clients and
-->servers, and updated continuously. Core data and indexes will reside on
-->consumers' machines, or at worst nearby machines if they have a mobile
-->device.
-->
-->Philip Abrahamson
-->***@mobilemaps.com
-->http://www.mobilemaps.com
-->
-->_______________________________________________
-->Geowanking mailing list
-->***@lists.burri.to
-->http://lists.burri.to/mailman/listinfo/geowanking

[Anselm Hook]

Post by Philip Abrahamson
iii. The engine that finds the information will be a "Nearby Engine" - a
cross between a search engine's keyword index, and a spatial proximity
index. It will have a single box for input.

I heard from the Nokia site that there is something like 200 million SMS
messages a day - one could easily imagine an equivalent quality of
geotagged messages per day within a couple of years.

- a

[Philip Abrahamson]

Post by Anselm Hook

Post by Philip Abrahamson
iii. The engine that finds the information will be a "Nearby Engine" - a
cross between a search engine's keyword index, and a spatial proximity
index. It will have a single box for input.

I heard from the Nokia site that there is something like 200 million SMS
messages a day - one could easily imagine an equivalent quality of
geotagged messages per day within a couple of years.
- a

Google's traditional search engine currently handles 3Billion pages.
It's just a matter of whether the index can be scaled. The added
component to Mobilemaps' Nearby Engine is a pair of space-curve indexes
for proximity, which likewise scale with a near constant lookup time.
We only have a little demo of a Nearby Engine at mobilemaps.com, with
half a million pages, because we're underfunded :-)

Serving the pages is less a problem than the time spent indexing them,
which can be more challenging depending on resources. However, the
beauty of geographic data is that it can be scaled geographically: local
data is usually sought by local users; so a Nearby Engine can choose to
index data only in its own geographic area.

Philip Abrahamson
***@mobilemaps.com
http://www.mobilemaps.com

[jo walsh]

sonny,

Post by Sonny Parafina
and public entities. In OGC, we have a saying, "We cooperate on standards,
and we compete on implementation." OGC standards are a framework for
building the killer app.

In my world, the development of standards is a completely open process
fully documented on the web, and hyperdiscussed and tested by many
different kinds of developers and interest groups.org. open software is
actively distributed and disseminated through communities like cpan.org

I don't see this as a space for competition - a future in which i
subscribe to and switch between sony augmented world and nokiasoft
augmented world doesn't make sense. Small services providing easily
re-implementable ideas, like del.icio.us, make sense. Web * Services, that
ought to be easily chainable, which can drop in out and nothing else will
detrimentally stop working.

There are tremendously powerful tools allowing us to model and monitor
our entire social lives implicit here. Not just spatial but all manner of
personal and collective metadata about our activities past present and
future. Whose hands the tools are in, and what questions they ask,
determines everything about how we map ourselves. In an atmosphere less
informed by data monopoly and vested interest i would not be so wary of
these complex, high-buy-in standards.


jo
--
"Common sense won't tell you. We have to tell each other." -DNA

[Sonny Parafina]

jo,

While it would be great to have a fully open standards development
organization for geospatial interfaces, there isn't one. OGC is the only
organization focused on interoperable standards that reduce the geospatial
gobbledy-gook created by the vendors into a reasonably coherent set of
interfaces. "Compete" means who can deliver the most functions, most
stable, most usable, fastest, etc software, all within the OGC standards
framework. I like to think about the whole endeavor as similar to writing
haikus.

Personally, I don't think the interfaces are particularly cumbersome, all
specifications use 3 basic requests: getCapabilities (service metadata about
the service offered), describeXXX (get data metadata), and getXXXdata (a
request to get the data from that particular service). My preference is
that mappers should do what mappers do best and allow others to hook into
mapping systems transparently. A nice example of this is the Inline WMS
client (http://sourceforge.net/projects/inlinewms) that allows users to
imbed a maps from multiple Web Map Servers and another non-geographic data
source.

sonny



-->-----Original Message-----
-->From: geowanking-***@lists.burri.to
-->[mailto:geowanking-***@lists.burri.to]On Behalf Of jo walsh
-->Sent: Wednesday, January 28, 2004 12:56 PM
-->To: ***@lists.burri.to
-->Subject: RE: [Geowanking] RE: LazyWeb request
-->
-->
-->
-->sonny,
-->
-->> and public entities. In OGC, we have a saying, "We cooperate
-->on standards,
-->> and we compete on implementation." OGC standards are a framework for
-->> building the killer app.
-->
-->In my world, the development of standards is a completely open process
-->fully documented on the web, and hyperdiscussed and tested by many
-->different kinds of developers and interest groups.org. open software is
-->actively distributed and disseminated through communities like cpan.org
-->
-->I don't see this as a space for competition - a future in which i
-->subscribe to and switch between sony augmented world and nokiasoft
-->augmented world doesn't make sense. Small services providing easily
-->re-implementable ideas, like del.icio.us, make sense. Web *
-->Services, that
-->ought to be easily chainable, which can drop in out and nothing else will
-->detrimentally stop working.
-->
-->There are tremendously powerful tools allowing us to model and monitor
-->our entire social lives implicit here. Not just spatial but all manner of
-->personal and collective metadata about our activities past present and
-->future. Whose hands the tools are in, and what questions they ask,
-->determines everything about how we map ourselves. In an atmosphere less
-->informed by data monopoly and vested interest i would not be so wary of
-->these complex, high-buy-in standards.
-->
-->
-->jo
-->--
-->"Common sense won't tell you. We have to tell each other." -DNA
-->
-->
-->
-->
-->_______________________________________________
-->Geowanking mailing list
-->***@lists.burri.to
-->http://lists.burri.to/mailman/listinfo/geowanking