Licensing of EGS4/EGSnrc
Iwan Kawrakow
Dear EGS users,
Jan Boucek and Keith Wise have raised the issue of the
EGSnrc license. I am not a lawyer, but I would like to try
to communicate my understanding of Copyright law, "free software",
"public domain" software, etc.
When someone writes a piece of code (the Program), he/she becomes the
Copyright owner of the Program. If the Program was written during
work hours, then in most cases the employer, no matter whether
a government lab, university, commercial company, or whatsoever,
becomes the Copyright owner. Some employers even extend this rule
to work after hours, if the Program is in some ways related to
the employment (and so, in the case of EGSnrc, The National Research Council
of Canada and SLAC are the Copyright owner, not Dave Rogers and
myself).
The Copyright owner, and no one else, has the right to copy,
distribute or modify the Program, to translate into another computer
language, etc., etc.
The mere act of making a Program publicly available (e.g. on a web server),
does NOT change the Copyright ownership and does NOT transfer any
rights to the user. In particular, the Copyright owner of the Program
has rights on any "work based on the Program".
"Work based on the Program" means either the Program or any
derivative work under Copyright law: that is to say, a work containing
the Program or a portion of it, either verbatim or with modifications
and/or translated into another language.
Let's now look at James Satterthwaite's suggestion that
there are "free" alternatives and that Geant4 is a very
promising one. I have searched their web site but didn't
find any information other than that ``the source code
is freely available". Sure. To do what?
To download it? To use it for research? To use it to
design weapons? To include it in a commercial package?
Anyway. I downloaded the source and found the
following statement on top of each source file:
// This code implementation is the intellectual property of
// the GEANT4 collaboration.
//
// By copying, distributing or modifying the Program (or any work
// based on the Program) you indicate your acceptance of this statement,
// and all its terms.
Hmm. Does James, Professor Bielajew, or any other list member PRECISELY know
what the legal implications of the above statement are?
I am not a lawyer but my interpretation is that the Geant4 collaboration
IS the Copyright owner and has all rights reserved, as discussed above.
Browsing their web page more carefully one finds that indeed, they too,
have had discussions about commercial use of Geant4 (see
http://wwwinfo.cern.ch/asd/geant4/collaboration/tsb_executive_02Nov99.html
http://wwwinfo.cern.ch/asd/geant4/collaboration/cb_executive_27Sep99.html)
In the second link we learn that some of the Geant4 members would have
liked to see Geant4 distributed under the
GNU General Public License (see below).
But it didn't happen. So, James, and all others planning a commercial
application of Geant4: I wouldn't do it without their express, written
consent (which will, most likely, come in form of a license granting
to you certain rights. The license may, or may not, be associated with
licensing fees).
The GNU General Public License (check http://www.gnu.org/copyleft/gpl.html)
is intended "to guarantee the freedom to share and change free
software--to make sure the software is free for all its users."
In addition, the GNU General Public License protects the rights of
the author(s), in particular their reputation, by requiring that
any changes made to the Program are clearly indicated.
I personally believe that the GNU General Public License (GPL) is the
best way to share software between researchers.
That's why my colleague Matthias Fippel and I decided
in 1997 to distribute VMC,
an "EGS-free" code specifically designed
for Radiotherapy type Monte Carlo simulation, under the GNU GPL
and 120+ people have downloaded the code since then. One might object that
VMC is no longer publicly available. There are variety of
reasons that led to the decision to stop the free VMC distribution,
one of them being the remarkable similarities between "major innovations"
introduced in a very recent PMB paper and the techniques employed in VMC.
Has Professor Bielajew, who "vehemently disagrees with the NRC licensing
initiative", ever distributed a software package, written by him,
under the terms of the GNU GPL, or any equivalent modification thereof?
I'm not aware of such a software.
Let's now come to the EGSnrc/EGS4 license.
As Jan points out, user downloading EGS4 don't need to accept
a licensing agreement. This fact has created the perception
that EGS4 is "free software" or "public domain" software.
However, if one bothers to look in egs4.mortran, egs4mac.mortran
and egs4block.mortran, one finds the statement:
" Copyright (C) 1985 by the Board of Trustees of the Leland "
" Stanford Junior University. All Rights Reserved. "
That is, Stanford is the Copyright owner, and as no rights
are being transfered to the user, they retain all rights
on the "Program, and work based on the Program" as discussed
at the beginning.
(Meanwhile, in recognition of the tremendous
contributions by Dave Rogers and Professor Bielajew (during his
times as a NRC employee) to the EGS community, the Copyright
of EGS4 and EGSnrc is jointly owned by SLAC and NRC and there is
an agreement between the two parties concerning this. )
The EGSnrc license is therefore not an attempt to restrict
your rights to use the code, it is an attempt to provide
a clear definition of the rights being transfered to the
user of the EGS package. From a legal point of view,
the EGSnrc license is a great improvement of the status
of an EGS user compared to the original EGS4 distribution
scenario, at least this is my interpretation of it.
In an ideal world, EGS4 and EGSnrc would be distributed
under the GNU GPL. I hope, this will happen sometimes
in the future. Ironically, the talk about the
great commercial value of electron/photon Monte Carlo
simulation packages that Professor Bielajew gave at SLAC a couple
of years ago has made the SLAC management aware of
the possibility of large royalties and has thus prevented
the distribution of the EGS system under the GNU GPL, at least for now!
Professor Bielajew's elocution on the license topic almost
deserve a full paper in a scientific linguistics journal
studying the subtlety of his formulations. For instance,
he writes
" ... as far as we are concerned, you are free to take the code and
develop your own tools for it "
and later
" It may have been better if I had said "As far as I am concerned...""
Sure. As far as Microsoft is concerned, once you have bought Excel
you are free to take it and develop your own tools for it, e.g.
macros for processing a certain type of data.
Later on he is careful to say
"I believe that EGS (in all of its forms), that is to say, the source codes
which model the electron and photon transport physics, should be available
for research AND commercial purposes, free of charge."
and
" ... I am working very hard with my SLAC colleagues to keep EGS5 free."
Professor Bielajew: could you please clarify to the list member
whether the whole EGS5 package, including its geometry package,
source routines, etc., or just the
"source codes that model the electron and photon physics" will be free ?
He also writes that
"Certainly, SLAC and NRCC will claim that the licensing fees are necessary
to maintain their research programs."
Where, dear Professor Bielajew, did you see licensing fees mentioned
in the EGSnrc license? The license merely states that the code
is free for research and education and that any other application
requires the "express, written consent of NRC". Did you, dear
Professor Bielajew, ever tried to obtain a license from the NRC
to do "exploratory research" using EGSnrc that may lead to
a "commercially-sponsored project" ?
And so on, and so on.
I would like to conclude this lengthy mail by addressing the
specific concerns of Jan Boucek and Keith Wise.
Keith, in case you decide to commercialise your tool,
what exactly are you going to sell ? I presume, not the entire
package (your macros and other pieces of code + EGS), since that
would be illegal, even if EGSnrc was distributed under
the GNU GPL. What you would be selling is the software package that YOU
have developed and you (or the institution you work for) have
the right to do so. It is then up to the vendor that takes
your package to ask for permission to distribute EGS along
with your software and it is up to the EGS copyright owner
(SLAC and NRC) to decide if, and under which circumstances, they
will grant this right to the vendor.
Jan, there is a simple construct that would allow you
to use the EGSnrc package for dose calculations for
fee-paying patients. As you can notice, the EGSnrc
package comes without any
"warranties, expressed, implied or statutory, of any
kind or nature with respect to the software, including without limitation
any warranty of merchantability or fitness for a particular purpose."
and
"NRC shall not be liable in any event for any damages, whether direct or
indirect, special or general, consequential or incidental, arising from
the use of the software."
If you wish to charge your patients for the calculations done using
EGS, you should be charging for the warranty YOU are providing
that the software is capable of performing that particular type
of calculations. Even the GNU GPL permits charging fees for the
provision of warranty, as we can learn from the GNU home page.
It is up to YOU to convince yourself that YOU can provide this
warranty, and YOU will be sued, if the patient gets burned
during an irradiation based on YOUR calculations.
You will still need a separate license from NRC to use
the package for fee-based calculations. But that license
will be most likely free, unless you bought a bunch of
CPU's and started offering your services worldwide over
the Internet.
Finally, I would like to stress that EGSnrc IS FREE for
the type of applications defined in the License. Anyone
afraid to be violating the free License should contact
NRC and either make sure that there is no violation of
the free License or obtain a separate license for his/her
use. I think, I can assure you that in the vast majority of
cases this license will come for FREE. And there will be
no strings attached other than, perhaps as Professor Bielajew
suggests, featuring "EGSnrc-inside".
Sincerely,
Iwan Kawrakow, a code developer accused of being against
the sense of scientific brotherhood.
--
---------------------
I. Kawrakow, PhD
Ionizing Radiation Standards
Institute for National Measurement Standards
National Research Council of Canada
Ottawa, ON, K1A 0R6
Canada
Tel. (613) 993 2197
fax (613) 952 9865
e-mail iw...@irs.phy.nrc.ca
Boucek, Jan
I hope I didn't say in my last postings that any person is unreasonable I
was rather talking about NRC or SLAC as institutions. But let me just
summarize why I feel very much on 'thin ice' as far as licensing issues are
concerned. In my opinion SLAC and NRC did change their policy on licensing.
Unfortunately from no licence whatsoever to licence covering everything
(apart from pure research - if there is such thing at all).
There are a few points I feel strongly about:
- there are a lot of people in EGS community who contributed great
deal in enhancement of EGS4 (lot of it in their free time because that's
when you do most of the research) and now they may feel their work is being
sell by someone else,
- even though that all source codes were copyrighted there was no
licence conditions mentioned whatsoever so my feeling at the time was that
obviously I cannot sell any EGS4 code but I would be free to use it in any
(even commercial) application,
- in my opinion EGS4 would never become EGS4 without being freely
distributed and not having restrictions on use included (when it got
popularized and improved so much by you guys here comes SLAC and NRC and
want royalties).
I'm pretty sure everyone agrees with Dr Bielajew that 'EGS4/5/nrc 's got
great potential for commercial applications. But SLAC and NRC approach left
me in grey area only lawyers would be happy in. Seems to me that Dr
Bielajew is right when he is saying: "This would apply to ... the
convolution kernels calculated by Mackie, myself, Rogers and Battista many
years ago. Heck, even TG-21 and
TG-51 uses EGS-derived data! A clarification from Stanford and NRCC on this
issue should be made as soon as possible."
I'm not a lawyer but I think we may be mixing up licence agreement and
copyright. Every book is copyrighted but that does not mean I have to pay a
licence fees if I quote it in any context, read it to my friends or use it
in my dose assessments. However I'm not a lawyer and I'm worried because
I'm kinda hooked on EGS4 (it's a great code). But I will certainly research
other options and regret I could not make informed decision 5 years ago.
In my opinion:
- SLAC or NRC (institutions) should state 'all conditions and
strings attached' in the code or at the point of download so every new user
would be told whole story,
- it would be great if SLAC or NRC 's clarify their positions for
example on a Web page (Basically EGS4 is all about probability: As Iwan
Kawrakow wrote: 'But that license will be most likely free, ' -what about
if not and what rules they would apply , no-one knows what fees are
demanded.) .
Regards,
Jan Boucek
Physicist
Nuclear Medicine Department, SCGH
Nedlands 6009
WA
Tel:(+618) 9346 2322
Fax:(+618) 9346 3610