[Manifold-l] Distance calculations

[Dimitri Rotow]

> I 'suspect' that large numbers of Manifold users are not 
> formally trained in GIS. This would be the result of low 
> product price and the fact that Manifold directly targets 
> high-end Office users.  I believe (rightly or
> wrongly) that most of the non-GIS professionals using 
> Manifold will want to work in non-projected space.  There are 

and...

> 
> I have never understood why Manifold makes working with 
> unprojected space so difficult. This is one area where lower 
> level software works well for many.

As Einstein once said, "Make things as simple as can be, but no simpler."
Any time one designs a package with professional conceptual power suitable
for expert use there will arise a tension as to how much one is willing to
risk deception by way of simplifications made for beginners (and, to some
degree hiding nuances always risks deception whether the beginner knows it
or not).

Measurements made over the Earth's irregular surface involve fundamentally
complex ideas, beginning with elementary notions such as the difference
between Euclidean and ellipsoidal space and continuing on through ever more
refined ideas such as the notion that there is no one ellipsoidal space but
rather many of them. [For that matter, there are plenty of applications
encountered by rank beginners where you can't do measurement without
choosing some DEM model as even a three-axis ellipsoidal model is not
enough, but let's not consider that right now as projections are enough for
this discussion.]  The traditional method of dealing with such notions is to
use projected coordinate systems. It's actually not a bad way to go, and
utilizing projected coordinate systems within Manifold has the very useful
benefit of tapping into millenia of prior experience, current educational
resources, well-understood characterization of various data sets and so on.

We have nothing against simplification when such is possible.  But some
things cannot be simplified no matter how much a beginner might want them to
be.  I can't tell you how often beginners ask for a view of the entire world
at once that doesn't distort anything.  But that's just not possible given
the difference between a 3D world and a 2D monitor.  

I grant you that providing an ellipsoidal distance using linear units within
a lat/lon coordinate system is something that could be done "close enough"
for most usages if one is willing to make certain assumptions on behalf of
the beginner.  The key question in my mind is whether such a simplification
does or does not do a disservice to beginners, especially those beginners
who go on to think in terms of an "accuracy" that is not equivalent to some,
say, coordinate system result generated by a colleague or other program.
It's hard enough for beginners to understand what projections are and why
they are used without tossing in a meta function that combines a bunch of
different notions at once. 

It is a highly fundamental notion that degrees are radial units and not
linear units to be used for measurement, and it is a very important part of
doing GIS (as opposed to running an atlas program like MapPoint) to
understand what coordinate system you are using.  Sooner or later you have
to get your head around such ideas or you will not be able to take advantage
of one of the major benefits of real GIS, the ability to tap into and
utilize millions of pre-existing data sets. It's also critically important
if you want to compare the results of fundamentally different calculations,
such as an ellipsoidal area computed using WGS84 in Lat/Lon in one part of
the Earth with an area computed using a UTM zone and different datum in some
other part of the Earth.   We don't want to give beginners the impression
that "don't worry about the details, it's the same thing" when it is not.

To date the Manifold approach has been to have faith that beginners could be
taught the necessary ideas.  It's not that much of a stretch to tell people
that if they are measuring distances they should be doing that in a
projection, not lat/long, and then those people who need to make precision
comparisons will be led (one hopes) into delving deeper into the details.
But nobody at Manifold will in any way resist the idea of making things
simpler where they can be if that is a greater priority for our user
community than other things.

Regarding priorities:  to date, Manifold has emphasized more professional
features for several reasons.  First, we felt that the market was already
well-served by elementary, inexpensive "atlas" programs, but it had no
inexpensive professional-grade software.  Second, we felt that creating the
apparatus for true professional capability was much more difficult than
consumer atlas stuff, so we thought it important to do that first.  Having
done the pro infrastructure you can always add consumer items, but not the
reverse.  Third, the barrier to entry for simple atlas programs is very low
but the barrier to entry for professional GIS is very high.   We like the
idea of being in markets where there is a very high cost of entry because
then we know that no one in their right minds will enter the market to
compete with us once we are in it as they are unlikely to recover their
entry costs given a price war with us.  Fourth, we saw a big prize in real
GIS that was available by decapitating the handful of previous occupants
(such as ESRI) and we saw the opportunity to do so by doing all the high end
stuff. Finally, we felt that just like the masses took to Borland turbo
Pascal at a time when experts felt that serious languages were not for the
masses, just so we believe that GIS can be taken mainstream and that there
is a much greater willingness to learn about GIS, if it is sensibly priced
and reasonably well implemented, than many people think.  That seems to be
the case, as there are very many more people from the mainstream who are
perfectly capable of learning and doing GIS even given the sophistication of
some of the concepts involved, such as projections.

I 100% agree that we could grow that market even bigger by taking time to
expand Manifold down more into consumer usages and by making some things
easier for beginners who intend to get going in real GIS but are nonetheless
just starting on the path.  In recent builds it's clear we are hunting big
game in Enterprise markets, what with our collaboration with Oracle and all
that.  But it's only a matter of time before you see expanded efforts for
various low end simplifications as well.  I totally agree with you that
simplification of measurement in lat/long projections is a candidate for
that.  How it should be done and where it fits into priorities is something
that is still unfolding.

Regards to all,

Dimitri