Criticism of the SRI .geo proposalWhile I am in favour
of geolocated web services and content, I am not
convinced that this proposal
is the right approach. Its core benefit
may be as a marketing and publicity tool.
1.
There is the need to address disputed cells, such as the Kashmir
region, oft disputed
between india and pakistan. Letting either or both
countries control that area
and icann/sri/geo foundation is taking sides
in a long standing dispute
2. Because
the naming policy is clearly defined for machine generation
of domains, why bother
with a .geo TLD? Machines can just as easily use
subdomains of dotgeo.org during
a demonstration period.
3. Many existing web sites and services are implicitly
geolocated. Why
have a new heirarchy when a special geolocated UDDI registry would
find
located services?
4. I am concerned that the nameserver registration process
is just
another way of generating revenue for registrys. A 'reasonable fee' for
a
single location would scale up to be very expensive for anyone serving
dynamic
traffic or weather data for an entire continent. But keep the
fee low and you
encourage the virtual geode to be covered in spam
advertising for irrelevant content.
5.
The whole dynamic update mechanism is still 'to be done'. That means
that SRI
are applying for a TLD without apparently addressing many of
the implementation
issues. Given some of the examples have near
constant update -e.g logging the
location of parts of a moving
production line- the proposal implies a rate of
change many orders of
magnitude more dynamic than standard registrys, even that
of DHCP
servers on a LAN. This will be hard to deliver on.
6. The billing scheme
fixes the amount the gTLD registrar charges
cell registrars ($0.15-0.50), which
prevents countries offering their
regions content at lower prices (as .UK and
.ZA do today).
7. In the detailed proposal one submits one's position to a geoserver
and
it returns relevant geodata. But there are many other context items
which are
relevant in a search: velocity and bearing, user wants
("cinemas", "traffic",
"holidays") which are not included in the naming
scheme or the search terms: why
not? URLs like cinemas.n52w2.geo may
become valuable items.
One of the items
I provide in my spare time is geolocated data : my
favourite alpine hill climbs.
It is reasonably tourist centric for its
little niche? Would I put such 'amateur'
content in the .geo domain? Not
at $1-5 per registered point. The proposal is
not economic for larger
databases provided by amateurs -despite the fact that
a vast amount of
touring content is of that nature. Anyone supplying vast amounts
of
geolocated data may also have issues with the cost -offering at cost
registration
to academic content would be critical to ensuring that .geo
does become the home
of all GIS data.
The issues I have with execution (need for a new protocol for
dynamic
update of registry data, disputed territories, possible limited value
of
pure lat/long searches) can all be investigated without the need for a
new
TLD. Does SRI have a working prototype addressing these issues? If
they do so
then perhaps we can all be given access to it to see if it
addresses all of them.
To
summarise, I would like to reiterate that geolocated web information
can be of
profound benefit. But other I feel that unified geolocated
markup of existing
web sites will benefit everyone without the need to
fork off geolocation into
a separate heirarchy. If that is what is
needed for every 'complex' search term,
then all work on the semantic
web might as well be stopped right now, because
clearly it is obsolete.
Steve Loughran
www.iseran.com