5:50 a.m.
As for the main topic... I fully agree with Dr. Kelly and others about
the GUI. As I have also mentioned, the GUI and the menu structure
needs some serious attention based on a modern and consistent design
direction. LaTeX output and graph pages need to work in all cases
since such are the things that make gretl so "cool" to use. Also, it
would be nice to add a few items for easily performing what applied
people do nowadays. For example, it would be nice to have Peseran's
ARDL bounds testing in addition to gretl's nice VAR, VEC, Johansen GUI
facilities.
The above are the sort of things that most people love, use, and will
start using gretl for. It is not some super duper matrix programming
language that no one has heard of or care about. I am increasingly
thinking that Hansl is a bad idea. There are already good free
alternatives with dozens of developers. Why compete with R?
I am a serious follower of gretl since 2005 (which is not as long as
some people on this list but still pretty long) and I see how design
direction has changed over the years. Initially, gretl was about
combining the various nice and available open source technologies such
as ESL, GTK, gnu MP, gnuplot, x12arima - tramo/seats in an intuitive
little package that can deliver. This is something that can be done by
a single dedicated individual namely you (and a small number of
contributors), who along this purpose added over time popular methods
as well as very very nice databases, example scripts, textbook data
sets, and translations; which has attracted many people on this list
to gretl in the first place.
In the last 4 years, however, it has become more and more about Hansl
and developing the most elegant scripting language ever. This is a too
big task even for you. It is not even that important. Much less work
would go into adopting R for scripting (just like how gretl adopted
gnuplot and GTK before) and that would put the program in a much more
advantageous position today. Just imagine being able to use some 4200
packages in CRAN.
As for today... Well, R is devouring everything and that is a good
thing. So what gretl needs to do is to position itself as
complementary to R so that it will remain relevant in the foreseeable
future. On future, Google Chief Economist Hal Varian says (New York
Times, 25 February 2008)
"So what’s getting ubiquitous and cheap? Data. And what is
complementary to data? Analysis. So my recommendation is to take lots
of courses about how to manipulate and analyze data: databases,
machine learning, econometrics, statistics, visualization, and so on."
What this means is that there is and will be a hugely increased demand
for good and easy-to-use GUI based econometric packages such as Eviews
and SPSS. Gretl really has a shot here because it is already
comparable to these programs in terms of ease of use and functionality
(although it still has considerable catching up to do). Plus, the
commercial programs are expensive and hard to use in terms of the
installation, registration, upgrade hassles. On the other hand, gretl
cannot and doesn't have to compete with R to become the platform on
which new methods will be developed. So the future of gretl lies with
students as well as corporate and government people, so deal with it.
Accordingly, the development has to shift focus on future trends: (1)
Intuitive GUI based modern econometric analysis, (2) localization, (3)
teaching, (4) portability.
As you know, I am writing all this because I really care about gretl.
I know about software, I have research on it, and I am saying now so
that, if gretl becomes a footnote in scientific software history (such
as EAL, ESP, DAMSEL, EPS), noone will say "if you knew, why didn't you
tell?"
Sincerely
Talha
On Mon, Jan 7, 2013 at 3:55 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Mon, 7 Jan 2013, Andreas Noack Jensen wrote:
> Even though I can always find what I am looking for (which is a
> problem in OxMetrics) I find the GUI a little unintuitive.
> Unfortunately a cannot really say what to change in order make it
> more intuitive to me. Only, that I don't like the category
> "Nonlinear models" as I think the categories should be positively
> defined (like the story about the two animals of the world:
> Elephants and non-elephants).
That's a fair point, and easy enough to fix, or at least improve.
Having a "Limited dependent variable" category in the model menu
might help?
> A more real problem is GTK maybe. It just doesn't look good in
> general and in particular not on Windows and Mac. Again, I have no
> idea about how to change that as I guess it would take too much
> work without any functionality gains.
Using a toolkit other than GTK for the GUI would be a huge amount of
work and not something I personally would undertake. But GTK 3 has a
more "modern" look, and as it matures we can consider switching for
the Windows and Mac packages. As I've mentioned before, using the
Quartz-based variant of GTK would (in principle) make gretl much
more Mac-like.
> I have thought about what I think the obstacles for Gretl are in getting
> more attention. The statistical functionalities in Gretl are really great.
> Within my own area I think that the cointegration stuff is really great and
> better than most of the commercial packages (but the unrestricted models
> should still go :-)). At my department we teach SAS but faculty uses Stata
> and I don't think it will change anytime soon. Maybe people like the idea
> about a supporting institution (enterprise) for the software to ensure
> continuity which is again a problem that is hard to deal with.
True, we can't do the "enterprise" thing. But we could perhaps think
about a more formal release process, with more systematic and
widespread pre-testing -- and more "advertising", if we can think of
a way of doing that. Maybe do one "LTS" release per year, in which
case we could maybe reinstate automated updates for such releases.
Not sure how much this would help, but one thing I've noticed with
colleagues who use gretl in their teaching is that they and their
students tend to be using rather dated versions of gretl at most
points in time.
Allin
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
“Blessed are the young for they shall inherit the national debt.” –
Herbert C. Hoover (1874-1964)
--