Services & ResourcesWolfram ForumsMathGroup Archive

[Date Index] [Thread Index] [Author Index]

Re: Re: Re: Thoughts on a Wolfram|Alpha package

• To: mathgroup at smc.vnet.net
• Subject: [mg101934] Re: [mg101912] Re: [mg101885] Re: Thoughts on a Wolfram|Alpha package
• From: George Woodrow III <georgevw3 at mac.com>
• Date: Thu, 23 Jul 2009 03:57:00 -0400 (EDT)
• References: <h41f31$rfv$1@smc.vnet.net> <200907202321.TAA26029@smc.vnet.net>

I am very happy to read these comments.

First, I have David Reiss's "A WorkLife Framework", not David Park's  
"Presentations" package, at least not yet.

I have noticed that both these add-ons are usually ready with updates  
at the same time that new versions of Mathematica come out. My  
negative comments were not directed at these apps.

Parenthetically, you have (or will shortly have) made a sale. It  
appears that I cannot get my colleagues to install the free player  
app, so any hopes of building an in-house user base was just a  
fantasy. I end up sending them PDFs of notebooks, or converting the  
interactive graphics into Quicktime. Life is too short to try to re- 
invent the wheel when you have done the job so elegantly.

Your comments on trustworthiness of the code (without WRI's vetting)  
is to the point. However, WRI is not doing anyone any favors by not  
updating the software that they badge (even if it is not written by  
them) in a timely manner. My bad experience has made me very wary. And  
you are right about the cost. I did not have to pay for the packages I  
got through Wolfram, but they seemed to be overpriced for what they  
delivered. In most cases, their use were justifiable when they saved  
me enough time, but not when the useful life of the software was  
shortened by incompatibilities with new versions of Mathematica.

I am 'glad' to learn that other people have problems with Workbench.  
(Not really schadenfreude since I feel the pain as well.) I thought it  
might have been just me, coming from a mac background and expecting  
some sort of elegance in the tools. The documentation is poor, I get  
incomplete or un-useful answers for tech support about Workbench, and  
the operation of the program is a kludge.

Were it not for the fact that I really want to make decent Version 7- 
style documentation, I'd roll the packages by hand. Making a package  
hasn't changed that much since version 2 so I can use Roman Maeder's  
excellent books.

I think that documentation building is a major problem. I am awaiting  
the release of the new version of mathStatica. This is a book/software  
combo. If the comments on the website can be trusted, the main  
sticking point is the documentation. Based on my brief exposure to  
trying to trying to document a much simpler package, I can understand  
this.

I am sure that internal to WRI there are any number of cheat-sheets  
that let people write the documentation. It would be great to expose  
this information to the users.

So, the upshot to all this is that anything worth charging money for  
needs proper documentation, and there are great advantages to using  
the version 6/7 help system. However, the threshold to making this  
documentation is so high that it simply may not be worth it.

That, I think, is a critical issue. It is one thing to make a package  
for one's own use. Once one gets beyond a certain level of  
sophistication with Mathematica, this type of activity is inevitable.  
However, if one wants to share the work -- for free or not -- then  
having decent documentation and having an easy way to update software  
is essential. If these steps take too long, then the cost of  
development of a 'shrink-wrapped' package would be prohibitive.

george woodrow


On Jul 22, 2009, at 2:04 PM, David Park wrote:

> I will tiptoe in with my comments since George has made enough  
> substantial
> points to enliven the discussion.
>
> Developing 3rd party Mathematica packages is not hopeless, but  
> neither is it
> a living. In my case it pays for Mathematica and computer (one)  
> expenses but
> not much more. There are other compensations:
> 1) I really do like to have a package for my own use that extends  
> the way I
> can use Mathematica as a development, study and communications  
> medium, and
> have everything in one place. I think it is useful enough to share  
> with
> other users for a reasonable price.
> 2) I like to interact with customers, and with MathGroup, I get many  
> ideas
> that help me improve my packages. I actually learn some math and  
> physics
> from them.
>
> My experience is that WRI is not a great help to 3rd party  
> developers. Of
> course, they don't mind users providing free software on MathSource.  
> They
> have a very different attitude between academic and non-academic  
> people and
> this overwhelmingly determines their response.
>
> The WRI 3rd party business model does not seem correct to me. It is  
> really a
> book publisher's business model. WRI takes the vast proportion of  
> proceeds
> and leaves only a small amount for developers. For a book publisher  
> this
> might be appropriate because they have to pay for the actual  
> production,
> distribution and marketing of the books. It is not at all  
> appropriate for
> software that is downloaded and where they have to provide nothing  
> but a web
> page. I find that selling through an independent web store such as  
> Kagi is
> far better. WRI does not provide ANY value added. Kagi handles all the
> transaction business, collects any required taxes, markets all over  
> the
> world, and takes only a small reasonable fee. They don't have any  
> dealers in
> different parts of the world that also require a cut, and they also  
> pay more
> promptly. WRI has 21st century technology with an 18th century  
> business
> model.
>
> I seriously doubt the claim that the number of active Mathematica  
> users is
> in the millions. That might, with some stretch, be the number that  
> have
> Mathematica available to them in some form, but I would guess that the
> number of active users is an order of magnitude (or two) less.  
> Otherwise, I
> believe that the number of questions on MathGroup, and the number of  
> active
> responders would be far greater. Also, if there were really a  
> million users,
> it would probably be far easier to sell 3rd party software. One  
> might pick
> up a lot just from compulsive shoppers!
>
> George has a very valid point when he talks about the problem of  
> making
> notebooks written with 3rd party software available to other users  
> who do
> not have the software. I'm certain this hurts the sales of  
> Presentations,
> especially since it is so much attuned to using Mathematica as a  
> technical
> communication medium. That is why I was so excited when I first saw
> Mathematica PlayerPro. I even produced a test PlayerPro version of
> Presentations to see that this is easy to do. I believe it would be  
> a huge
> breakthrough if WRI could provide a free PlayerPro that was  
> analogous to the
> Adobe Acrobat reader. Because Mathematica is so much more expressive  
> than
> static or semi-static documents I believe that eventually the  
> technical
> community would gravitate to Mathematica as a publication medium.  
> Then there
> truly would be millions of active users.
>
> A number of purchasers of Presentations have enquired about making
> Presentations a standard package in Mathematica, and a number of  
> them have
> written to WRI about it and informed me later. I did make a  
> recording for
> WRI on it, but was cut off before finishing and I don't think any  
> developer
> ever looked at it. So far, no one at WRI has been willing to look at  
> the
> package.
>
> WRI was not very helpful to developers in producing documentation. The
> Version 5 Help Browser has serious problems in reproducing dynamic  
> content
> within the Browser window. It took WRI two years to provide a  
> reasonably
> effective Workbench for Version 6 documentation. The Workbench  
> documentation
> itself is quite poor. Workbench does not work well for applications  
> that use
> the DeclarePackage construct, and they have a difficult time  
> grasping user
> applications that go beyond the immediate WRI documentation needs.  
> (For
> example, writing electronic textbooks, writing courseware, and  
> managing
> large research projects.) I am presently writing Presentations  
> documentation
> in Version 6 style and at least I am now at the point where I no  
> longer have
> to fight with Workbench.
>
> There is a problem with charging anything for software. People just  
> think
> they can dash off whatever software routines they need and it cost  
> them
> nothing. Or that they should be able to get it for free. Some people  
> think
> it is actually unethical to charge for software. I think this is very
> misguided. (I did check prices on the Wolfram web site a few years  
> ago and
> the packages ranged in price from a few hundred dollars to many  
> thousands of
> dollars. Maybe with those prices one would balk. These prices might
> especially be irksome if the packages are not keep up to date, or  
> are not
> well documented. One should remember that academic people do not get  
> a lot
> of credit or career advancement from software.)
>
> Presentations sells for $50. (One big booster advised me to sell it  
> for much
> more and depend on a few high priced sales. I would prefer to make it
> available to more people.) So, if you are working for the minimum  
> wage, how
> many hours do you have to save to justify the price? Maybe about six  
> hours.
> And how much do students pay per year to attend university,  
> especially some
> of the more expensive private ones? How much debt do students leave  
> college
> with these days? How much is their time worth? $50 seems to me like a
> trivial price - if the package really is useful and saves time on  
> more than
> a single problem.
>
> In many cases I have provided solutions to questions on MathGroup with
> Presentations, and often when no other regular solution was  
> provided, but
> this hardly ever results in a direct sale. In one case a person  
> posted a
> problem and I had a number of communications with the poster, and  
> finally
> provided a solution with Presentations. I sent a PDF to him, and  
> also posted
> the solution on MathGroup and at Peter Lindsey's site:
> http://blackbook.mcs.st-and.ac.uk/~Peter/djmpark/html/
> But after purchasing a package was mentioned all communication  
> stopped, and
> a week later the person was back on MathGroup looking for a free  
> solution.
> So even $50 or a 3rd party solution without any WRI sponsorship, or  
> belief
> that the solution couldn't be trusted was too great a barrier to  
> overcome.
>
> So there is a large credibility problem in selling even inexpensive
> Mathematica software.
>
> Finally, some comments on factors that I believe contribute to a good
> general purpose package.
> 1) The package should be a natural extension of Mathematica without
> interposing a new user interface. The regular Mathematica interface is
> pretty good and it is just an annoyance to users to have to learn a  
> new one.
> 2) The package should have 'hierarchical depth' in the sense that  
> rather
> than providing large set-piece solutions, perhaps with many options,  
> it
> should provide routines with which users can easily craft their own
> solutions. A user shouldn't have to rewrite a routine to apply to  
> his or her
> own case.
> 3) It should be well documented. Documentation is equally or perhaps  
> even
> more important than the routines.
>
> (And George is talking about the other David. He's never purchased  
> any of my
> packages, which all work with Version 7.)
>
>
> David Park
> djmpark at comcast.net
> http://home.comcast.net/~djmpark/
>
>
>
> From: George Woodrow III [mailto:georgevw3 at mac.com]
>
> This is my take on the situation:
>
> Over the years, I have purchased any number of add-ons for
> Mathematica, including stuff from Wolfram and some third party
> packages, including David's.
>
> I've been disappointed time after time. (It wasn't *my* money most of
> the time, so the time saved by using the winners usually made up for
> the bad choices.)
>
> First, even for Wolfram applications, there is the inevitable lag
> between a new version of Mathematica and an update to the packages.
> This problem was worst with major changes, such as Mathematica 3 and
> 6. Many of my add-ons are still not working with Mathematica 6 or 7.
>
> This is understandable for a one person shop: the transition from
> Mathematica 2.x to 3 and 5.s to 6 were substantial. However, it has a
> serious impact on the perceived value of the add-on.
>
> Second, as the capabilities of Mathematica have improved, it is simply
> easier to write my own code. I am sure that this is the case with many
> Mathematica users.
>
> There is also enough free stuff out there that 90% of the code for any
> specific application is already done.
>
> Also, I would generally want to customize the code. If I did the work
> myself, then I'd already understand what is going on. Trying to alter
> an existing program is more difficult. My code might not have all the
> bells and whistles of a polished app, but generally, I only need a
> subset of the code to get the job done.
>
> If I want to send the program out to my co-workers, I cannot assume
> that they also have the package or add-on, so anything I want to
> deploy cannot have any expensive dependencies.
>
> I looked into this possibility as a source of income (and something to
> do) in my retirement. I came to the conclusion that there are only two
> viable business models:
>
> 1.	Develop something and sell it to Wolfram. An example of this is the
>
> Classroom Assistant in Mathematica 7 (I'm assuming that there was some
> compensation involved.)
>
> 2.	Develop a turnkey application using Mathematica for an application
> that does not require any knowledge of Mathematica. Player Pro is a
> delivery system for this application model. There are certain
> advantages to using Mathematica over developing a program in c/c++.
> However, this is not the type of program we are talking about here.
>
>
> It is easy to be wrong about this. After all I do not have any numbers
> about Mathematica users. Wolfram states that the number is in the
> millions, but I am not sure that the vast majority of these are
> students. I am surprised that it is hard to reach a break-even
> userbase of as little as 200-300 users.
>
> In many ways, using Mathematica is similar to the way it was in the
> 1970s, before there was any 'commercial' software. I use the Apple ]
> [ as an example, but it was only the first commercial PC.
>
> Mathematica takes the place of the built-in BASIC interpreters. People
> shared programs. There were books of programs that you could type in.
> (A *BIG* program at the time was 300 lines of BASIC code.) There was a
> lot of sharing. People were happy to show off by providing neat code
> examples.
>
> I missed those days. With Mathematica (and especially the newer
> versions), I can recapture the essence of using the computer to do
> mathematical things without all the overhead of learning c, Xcode,  
> etc.
>
> I may be an aging idealistic hippie, but I think that there will
> always be enough people who will delight in sharing what they know
> that the market for paid mathematica add-one will always be undercut.
> The nature of Mathematica itself, and Wolfram Research's ecosystem
> that encourages this type of thinking.
>
> ======
> So to summarize: I think that there is little market for Mathematica
> add-ons for the following reasons:
>
> 1.	Poor support in most cases. Code that breaks with each new version
> of Mathematica is a poor value.
>
> 2.	Free code that is 'close enough' trumps paid code unless the paid
> code is really exceptional.
>
> 3.	The mind-set of the users who might buy mathematica add-ons is
> biased towards either rolling their own or finding something free.
>
>
> george
>
>
> On Jul 20, 2009, at 7:21 PM, AES wrote:
>
>> In article <h41f31$rfv$1 at smc.vnet.net>, David Reiss
>> <dbreiss at gmail.com>
>> wrote:
>>
>>>
>>> Sadly 3rd party application for Mathematica sell very poorly--rarely
>>> even beginning to recoup the cost to develop them.  But I could do
>>> this if there were a couple of hundred users willing to purchase it.
>>>
>>
>> I'd welcome your thoughts (or anyone's) on _why_ this is the case?
>>
>> (I have my own ideas, but reality checks are always useful.)
>>
>
>
>

• References:
  • Re: Thoughts on a Wolfram|Alpha package for
    • From: AES <siegman@stanford.edu>