I don't think this is nearly as contentious as David fears it is. I see
a few very mundane points, mostly not new at all, coming out of this
latest thread. I don't want to comment very specifically on any other
individual, or their views or behaviour, for standard reasons. I do
think that David has made several things clear about Ashim's work that
were not nearly so clear in Ashim's own postings.
I will go a bit, on one point a lot, beyond this thread and link in some
impressions gained from other threads and private emails.
1. People who don't show their exact code are less likely to get good
answers to their questions, as people are less likely to understand the
exact problem and less able to give the advice that people want.
That's age-old on this list.
It is a little novel to get someone declaring that they are developing
proprietary code that they can't or won't show, but whatever the reason
for being coy, the result is much the same. Note that all I am saying is
"less likely": no more, no less. Several specific answers, including
some from David himself, appeared in this thread that Ashim appeared to
find interesting or helpful.
2. More generally, and as others have declared, Statalist works best in
open mode where everything (problem, data, code) is explicit. If people
are shy about any of those, it can be difficult to know what to say,
except "Tell us more", or what to do, except Delete with extreme
prejudice.
3. People who make sharp comments should not be surprised if others
disagree, and even if they get a little upset. People should know that
from childhood on. Conversely, the list is not a collective exercise in
diplomacy. People can be frank with each other and even disagree
strongly on some points and still maintain an atmosphere of general
mutual respect. Indeed, a list will be lively and stimulating largely to
the extent that that is so.
I was once on a list (name long since forgotten) in which venom and
vitriol seemed to seep out of messages daily, to the extent that I
withdrew quickly in disgust. In contrast, Statalist is a gentle
tea-party, and fortunately.
On a rather different note, I find it a little puzzling, although in a
sense also darkly amusing, that occasionally people who are manifestly
struggling hard with Stata and statistics should simultaneously fear
that people may want to steal their code. (This attitude is most obvious
in some private emails that reach me including expressions of
unwillingness to use Statalist as a medium.) That's about as wrong as it
can be -- setting aside any implicit slur on others' integrity.
First off, many of us most active in the Stata community write our own
code, because that's the way to get to do what we want to do. Stealing
what is likely to be incomprehensible, inefficient and buggy code is the
last thing on our minds. Second off, even bad people who steal code
(unless they are also idiots) should want to steal good, well documented
code from people with excellent reputations.
That's probably not a central issue for many. I do appreciate that there
are jobs in which it's an absolute rule that nothing whatsoever goes
outside the door. David explains all this and much else very clearly and
I have no serious disagreement with his analysis.
It's also worth noting that several people whose livelihoods depend at
least partly on private Stata-based consultancy have made enormous
contributions to this list over many years. It's a little invidious, but
I'll name Michael Blasnik, Joseph Coveney and Richard Goldstein as
excellent examples. So, there is no polarity between people in
"academic" mode and people in "commercial" mode. (The terminology is a
little unsatisfactory but should I hope be pretty clear.)
David asked specific questions:
1) Can persons working for private for-profit companies that may be
developing proprietary algorithms ask for help on Statalist?
This is easy.
The short answer is Yes, of course. There is not, and absolutely should
not be, any rule on Statalist against asking questions that are related
to code that will not made be public and is used largely for profit.
(Even if people thought that such a rule was desirable, it could not be
policed. Anybody can sign up to an anonymous account and cloak their
intentions.)
Nor is there any rule that determines how far people are able and
willing to answer those questions.
Here, as elsewhere, a good question for Statalist is one that someone
wants to answer well.
2) Should work by recipients of Statalist help be considered part of
the commons if it contains a solution derived from the commons?
Basically a Stata "General Public License" [ http://gplv3.fsf.org/ ]
clause.
I don't think that Statalist members either collectively or even
individually need have any opinions on such a question. Indeed many
might not want to discuss software development or dissemination in those
terms. I never worry about these licences myself. Do they really scare
bad people? I doubt it. In practice, there is really is nothing to stop
anyone benefitting from Statalist and then putting that benefit into
private or secret products. Nor is that necessarily anything to worry
about in principle. If you want your ideas to be protected, don't
mention them. It's as simple as that.
Nick
[email protected]
David Elliott
Frankly, I am disappointed at the turn this thread has taken and, from
an offlist discussion, know it has caused Ashim considerable personal
distress. By way of disclosure, I am the person who has been working
with Ashim on a problem of mutual interest. Namely, the creation of
customized HTML tables from program output. I have a similar need in
my work and so there was a natural convergence of interest. The
nature of Ashim's problem gave me an opportunity to work with him
towards my goal of making my -htmltab- program more generalizable.
He works for a company that is using Stata analysis of stock trades in
an attempt to gain an industry advantage. I understood and accepted
the fact that he could not send me his program in its entirety nor the
raw data upon which it acts. However, I have be working on an excerpt
that deals with the results display. One challenge, though, is the
recreation of the local macro environment that exists at the end of
the proprietary code so that my program can run with both the
resultant dataset and it's "macro milieu." Hence Ashim's request to
be able to save and recreate local macros.
The fact that Ashim has proprietary code upstream of the section of
interest is an absolute red herring, in my opinion. He is being paid
for a job where he must understand financial data and has developed
code that analyzes trading patterns to give his company a competitive
edge. This code is not his to share. However, he has not asked for
help specifically with his proprietary code. He has asked questions
that, while being somewhat exotic compared with more mainstream Stata
enquiries, do have general application and interest.
He is working on a problem where he desires to produce colour-coded
output tables. This inquiry started in the thread "How do you print
colourful SMCL files" [
http://www.stata.com/statalist/archive/2008-12/msg00964.html ] which
is how I got involved. I don't believe this is a unique or
proprietary request. It also highlights specific limitations in
Stata's capabilities of producing printable colour-coded output.
While Stata's viewer is colour-capable, it's printing translation is
strictly B&W. This has lead Ashim and myself to investigate an
alternate display format in which significant control of colour-coded
output is possible, namely through direct writing of HTML table code.
I'd like to address this issue of "non-symmetry." Because Ashim works
in a for-profit environment are his requests for help somehow tainted?
Should he be required to contract for services from a willing list
member? (I recognize Nick was just twigging Ashim a bit with the
payment comment, but there is a "truth said in jest" element here) Is
he leaching from the collective commons and giving nothing in return?
(Ashim, in fact, has made contributions in other threads) I have a
foot in both the government and academic worlds so I do feel I have
some insights worth sharing here. I'm also the administrator of an
web development on-line forum (and trust me, Statalist is sooo tame in
comparison).
Very few of us are independently wealthy and play with Stata simply to
stave off perpetual ennui. We use it because we have a job to do.
Jobs, hopefully, for which we are paid or achieve some personal
advantage. If we are students, we use it for coursework or analysis
for our dissertations which open the doors to our future gainful
employment. If we are academics we use it in teaching or in our own
research. Academic recognition and winning competitive grants to
further our personal research goals are the currency in which we
trade. Persons in government or private industry may be paid directly
for their programming activities. In each case, however, we are all
working with Stata for some form of personal gain, we are working for
someone, even if that someone is effectively ourselves. I can't see
how that concept is offensive.
Regarding the concept of openness and sharing. I believe in the
collective commons and in online forums and lists in which people with
skill and insight contribute their time and effort to help solve
problems and develop the abilities of other participants. Some people
only give. Some people only take. Most do both. Personally, I find
Sudoku boring and find working on Statalist problems more educational
and rewarding and rarely do I find a situation where providing a
response doesn't increase my own insight - so by giving, I gain as
well. At an institutional level, Harvard has made a continuing
contribution to the commons by providing hosting of Statalist for
free. It has done so in a spirit of altruism based on academic ideals
of boundaryless sharing of knowledge. Not all people can live and
work within that paradigm at all times. For example, academic
researchers are frequently unwilling to share their data or methods
with others until they have gotten their own papers published and
recognition for their work. "Freely available" in this context is
often painted in shades of grey and researchers often work in an
environment where the concepts of competitive advantage and
intellectual property play a part - this is academic realpolitik.
So here is the crux: I think the Statalist collective needs to address
the following questions:
1) Can persons working for private for-profit companies that may be
developing proprietary algorithms ask for help on Statalist?
2) Should work by recipients of Statalist help be considered part of
the commons if it contains a solution derived from the commons?
Basically a Stata "General Public License" [ http://gplv3.fsf.org/ ]
clause.
I think this is worthy of some thoughtful debate.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
