Re: Re: v.7.0 issues
- To: mathgroup at smc.vnet.net
- Subject: [mg93951] Re: [mg93910] Re: v.7.0 issues
- From: "David Park" <djmpark at comcast.net>
- Date: Sat, 29 Nov 2008 04:32:26 -0500 (EST)
- References: <gggqff$31i$1@smc.vnet.net> <200811261010.FAA19484@smc.vnet.net> <gglsmu$8ai$1@smc.vnet.net> <16620439.1227870662502.JavaMail.root@m02>
Wow! This just illustrates some of the serious problems that Mathematica
faces.
As many people who know my views are aware, I regard Mathematica as a
revolutionary medium for learning, doing and communicating technical
material with high mathematical content. But it is a revolution still in the
making (but far enough along to actually use) and it comes up against larger
difficult problems that must be solved.
I view the revolution as a chasm that must somehow be leaped across. On the
near side is the static-documents medium of working and communicating. Do
your mathematics with pencil and paper. Use Mathematica as a calculator or
programming language to do calculations and produce initial plots. Then copy
these results to other media, add supporting writing and embellishments and
publish. It is unfortunately a hard fact that this is the ONLY way right now
to merge Mathematica into the publication process. Mathematica notebooks are
not today an acceptable publication medium.
If you are on the near side of the chasm, and plan to stay there, then most
of the graphical and dynamic improvements of Mathematica 6 and 7 are
useless. They are for private enjoyment but can't be included in
professional output.
On the far side of the chasm are literary, active and dynamical Mathematica
notebooks that can be exchanged with any other technical person. These are
so far superior to static mathematical documents that I believe they will
eventually totally eclipse them. It is like comparing the floor plan of the
Parthenon to the Parthenon itself.
So the first larger problem that needs to be solved is the ability to
'exchange with any other technical person'. This really means a free
Mathematica Player Pro. One that allows anyone to read, but not necessarily
save or print (reading online is enough), notebooks produced by those who
have Mathematica. I know there are difficulties and risks here for WRI, but
the risks have to be weighed against the gains. This is, after all, a great
advertising medium for Mathematica. It is a method of one-way communication
that may have a high conversion rate to two-way communication. (As an
interim step WRI might consider letting Premier subscribers recommend
recipients of free Player Pros. This would direct Player Pros to likely
purchasers and test the effectiveness of conversion to paid subscribers.)
AES wonders 'who [WRI] views as their customer base'. It ought to be anyone
who uses mathematics in their studies, work or communications. WRI should
aim for a much larger customer base and be dissatisfied with the present
situation. Mathematica should be dominant and widely used. I know that then
Wolfram will be accused of being a monopolist and things like that. But the
very nature of the product requires widespread use - to enable easy
communication with others and to entice young people to make the effort to
learn it.
It is not a trivial task to use Mathematica as something more than just a
calculator and programming language. It is not easy to write literate,
active and dynamic notebooks that effectively communicate ideas to others.
It is difficult enough to write literate non-mathematical documents. There
are schools that teach it, and consultants that make a living at it. It is
MUCH more difficult to write mathematical documents that utilize the active
and dynamic features of Mathematica. There are many ways to misstep. It is
easy to fill a document with 'computer junk'. I can't, for the life of me,
see how a single Manipulate expression with a few sentences of explanation
could effectively communicate a mathematical idea. Mathematical ideas
require development, exploration and explanation. It takes experience,
practice, knowledge of the tools, and judgment to exploit the far side of
the chasm.
Our educational systems spend years teaching students language skills. It is
just as important to teach students who may pursue technical careers the use
of Mathematica. They have to start learning it early. It is absolutely
misguided to expect students to learn Mathematica at a university when they
are also expected to learn difficult technical material at the same time.
Mathematica, except as a calculator, is just too complex to learn quickly.
It is because early exposure is needed that it is necessary that Mathematica
have a wider reach. It will be difficult to implement early instruction
unless students and administrators know that Mathematica is a must-have,
must-know application.
For students reaching college and for professional people, the goal should
be this: when using Mathematica they should be spending 90% of their time
thinking about their technical material and 10% of their time thinking about
Mathematica. This probably sounds utopian, but with early training and with
'tuning up' and further development of Mathematica it is achievable.
I agree that Mathematica is too expensive. (But those who are regular users
should seriously think of being a Premier subscriber and get the free
updates. How important is Mathematica in your work or studies? How does the
cost of Premier service compare to your annual expense of car maintenance,
or various insurance costs, for example?) Mathematica should have, and
deserves, a much larger base of subscribers. Economy of scale is the main
method of lowering the cost. (The idea of selling Mathematica with built-in
advertising somehow doesn't appeal to me. Especially in an educational
setting.) The NSF should be strongly supporting Mathematica as an important
component of technological development in the United States. It would be a
much better investment than all the money they pour into education as
entertainment and statistical studies of sociological factors in technical
education that I often see written up in Science magazine. A free
Mathematica Player Pro would do a lot to expand the paying base. And if that
brings the cost down, hacking to some kind of inconvenient usage is not
terribly economical for the hacker. I do think new business models need to
be explored but it is terribly difficult to second guess people at WRI who
probably think about this all the time.
Now for the documentation. Many users on the near side of the chasm, with no
desire to jump over, hate the documentation. That is because the
documentation is on the far side of the chasm. Since all technical
publication today is on the near side of the chasm it is a real disconnect.
But it is the way things are going to go. It is the future. All I can say
is: "Get used to it." It is no longer possible to provide static printed
comprehensive documentation for Mathematica. It would have to be a
multi-volume set and would be totally useless for all the dynamic
capabilities. By and large, I think the organization of the documentation is
fairly good, especially since WRI has provided multiple paths into it.
WRI has done a lot to improve the documentation from the initial Version 6
release. Much of this was in response to suggestions and criticisms on
MathGroup. For example, they added the Function Explorer and Virtual Book as
additional ways to access the documentation. In Version 7 they redesigned
the main Documentation Center to a more usable format. They added a list of
the Standard Extra Packages. A new feature in V7 is a list of 'How to'
tutorials.
My complaint about the documentation is mostly in the examples, and
sometimes that it is incomplete. But there is a massive amount of
documentation! I can just visualize the people at WRI 'grinding it out, day
after day'. Sometimes the inspiration is going to sag. I would like to see
more 'useful' examples as opposed to 'formalistic' examples.
For example when looking at the Parallelize examples, the first ones seemed
too fast to notice any effect. Except that I did get a brief message that
did say something to the effect that two processes were established. (I have
a dual core processor.) Then further down I found an example that looked
promising. So this was the result, with timing added:
Parallelize[Count[Range[10^6], _?PrimeQ]] // Timing
{3.12, 78498}
And then without Parallelize.
Count[Range[10^6], _?PrimeQ] // Timing
{0.78, 78498}
Where's the advantage? Maybe there is some caveat? How about an example that
will actually show off an advantage to Parallelize. Maybe some simple
obvious case that will work with two cores and maybe run in something like 1
minute instead of 2 minutes?
My second complaint about the documentation is that WRI still has not
produced a workable Workbench so that users can produce the new style
documentation. The last version I saw still had many flaws and could not
produce documentation for types of packages and applications that could be
done in the old Help Browser.
I know that many people have a love-hate relationship with Mathematica. New
versions always present many new things to learn. Version 6 was like
learning Mathematica anew. (But if you are happy in staying on the near side
of the chasm you can ignore all the new dynamic features and some of the new
graphics. Just take advantage of the improved functions, programming and
classical CAS features.) Cost is still a major problem but the only real
solution is to broaden the customer base. (And a free Player Pro is the key
that will turn the lock.)
Revolutions are messy, expensive and wasteful. They take unexpected turns.
They are difficult to keep up with. But they produce something new and
exciting and better. This is the very beginning of a golden age of technical
communication. Take part in it. Help guide it. Don't get discouraged.
Mathematica is not getting worse. It's getting better.
David Park
djmpark at comcast.net
http://home.comcast.net/~djmpark
From: AES [mailto:siegman at stanford.edu]
On Nov 25, 7:18 am, "alexxx.ma... at gmail.com" <alexxx.ma... at gmail.com>
wrote:
> I thought it could be a good idea to start a thread where to detail
> all the issues (both positive and negative!) discovered by early
> adopters of ver.7.0:
> detail your experience here!
>
> Alessandro
Alessandro:
Are you limiting this to "early adopters" only -- or can "non-adopters"
chime in also?
If the latter, then the overwhelmingly negative issues for me are not
any of minor or routine (but still damaging) bugs and glitches that will
come along with any major new version of any major software package (as
people have already been reporting for M6 in this thread), but a set of
much more basic issues:
1) M7 brings another massive set of changes in old routines or new and
increasingly complex capabilities that one presumably has to learn or
adapt to, before the dust has even settled on the massive (and massively
disruptive) changes associated with M6.
2) And yet, _still_ no adequate user documentation, at least for new or
occasional or "ordinary" users of these changes and new capabilities --
the same as was the case when M6 emerged. (And, presumably, any
third-party efforts that might have been under way to develop such
documentation for M6 have now been thrown off track, or at least further
delayed, by the emergence of M7).
3) Beyond this, as a working engineer, long-time university faculty
member, and educator, I'm fully in accord with the response that says
> Like all [recent] versions of Mathematica, version 7.0 is
> ridiculously expensive for the average user...
"Ridiculously and impossibly expensive" might be more like it.
4) And so I expect I'll be going along with another recent response
that says:
> Probably, I am not upgrading anymore for this and the
> reasons below, despite that I have been a loyal customer
> since the earliest v2.1 and upgraded most of the
> subsequent releeases.
In my case, it's been "loyal customer __and active proponent__ of
Mathematica since the earliest v1.0." Just dug out of my files acouple
of memos dating back a decade or more ago, from me to deans of schools
and to IT people in my university, urging major university support for
widespread adoption of Mathematica across my university. Where
opportunity presents, I'm voicing exactly the opposite views today.
5) I might be a bit more restrained in expressing these highly negative
views, here and elsewhere, had I not had an accidental encounter a few
weeks ago with a senior colleague in my department whose multi-year
academic involvement with Mathematica up through v5 has included doing
very extensive analyses and numerical calculations of complex quantum
phenomena; writing several memos and reports for his graduate students
and colleagues on the techniques he's developed for this work; and also
developing, teaching and writing class materials for an undergraduate
course titled "EE141M Engineering Electromagnetics with Mathematica".
(He also happens to be a member of both NAE and NAS.)
I discovered that, totally without any contact with my own views and
rants, he had purchased multiple copies of M6 when it came out, and
installed them on his home and lab computers. After wrestling with M6
for some period of time, he then removed and trashed all his copies of
M6; dug out his old installation CDs for M5 and reinstalled it on all
his machines; and doesn't plan to worry further about MX where X>5.
I've also had private email responses to some of my early posts on this
group following the emergence of M6 that have been supportive of of my
criticisms of M6.
Bottom lines:
* I absolutely don't pretend to understand Wolfram's product and
marketing and branding strategy for Mathematica, and especially who they
view as their customer base. But at this point, it doesn't include me,
or any groups I can think of; and I sure as hell wouldn't invest in
Wolfram myself, or advise anyone I know to buy Mathematica.
* And as an entirely separate issue, I don't understand Wolfram's
strategy or tactics so far as user documentation is concerned. Their
approach simply seems utterly crazy to me, and has since the arrival of
M6.
* But I also retain a great admiration for the technical capabilities
and the immense knowledge base embedded in Mathematica -- post M5
included -- which means the two preceding comments are sources of
sadness, not gladness. I'm just not going to ride the Wolfram train any
further myself.
- Follow-Ups:
- Re: Re: Re: v.7.0 issues
- From: Andrew Watson <andrew.b.watson@nasa.gov>
- Re: Re: Re: v.7.0 issues