[BlueObelisk-discuss] ODOSOS text

[Andrew Dalke]

Perhaps the Blue Obelisk "Open Data" page could describe what one should do in order to make their datasets open, or to disclaim any legal protections to data sets?

Here is text which I think helps fill in the "Open Data", "Open Source",
"Open Specification" pages of the Blue Obelisk pages.

Please note that I have changed some of the Blue Obelisk terms
to be more acceptable to my views and understanding. For example, I
do not believe that an Open Standard essentially requires "community process."

Cheers,

Andrew
***@dalkescientific.com




===============
http://sourceforge.net/apps/mediawiki/blueobelisk/index.php?title=Open_Source

We use "Open Source" to encompass
free [http://www.gnu.org/philosophy/free-sw.html]
and
open source [http://opensource.org/docs/definition.php] software
as well as software in the
public domain [http://en.wikipedia.org/wiki/Public_domain].
(ODFOSOS doesn't have the same simplicity to it). We believe
that open source is an essential prerequisite for useful peer
review, we believe that open source research software is
the best way to advance and disseminate knowledge, and we
encourage people to release their own software under an open
license.


Do you want to release your software as Open Source?

The principles of free and open source software is based in
copyright law. Software is essentially text, and in most countries
you, or perhaps your employer, automatically have the right to
prevent others from copying that text. You must actively either
grant others a license to make copies or give up your copyright
and put it into the public domain.

The book "Producing Open Source Software" has an excellent
summary of how to
choose a license [http://producingoss.com/en/license-quickstart.html].
Our recommendation is to use the
GNU GPL [http://www.gnu.org/licenses/#GPL],
the MIT / X Window System License [http://opensource.org/licenses/mit-license.php]
if you are interested in a simple license, and the
Apache License 2.0 [http://www.apache.org/licenses/LICENSE-2.0]
for a permissive license which also explicitly grants a patent license.

If you want to put your software into the public domain then we
recommend you either use the text from the
SQLite public domain dedication [http://www.sqlite.org/copyright.html]
or use the
CC0 public domain dedication [http://creativecommons.org/about/cc0].
Be aware that "public domain" doesn't have a well-defined internationally
recognized meaning, so many people prefer instead the certainty of
a license.


===============
http://sourceforge.net/apps/mediawiki/blueobelisk/index.php?title=Open_Data

We use "Open Data" to apply to data sets which are part
of the open scientific literature, including data sets
which are published on research web sites. We believe
that facts and collections of facts must be in the public
domain, or at the very least distributed by a data license
which adheres to the
Panton Principles [http://pantonprinciples.org/].
We believe all published data sets, including those which
contain material covered under copyright, should be
released under a permissive data license which allows
anyone to copy, process, analyze, modify, and redistribute
that data, for any purpose.

We specifically exclude personally identifying information from
this belief, and have no position about data which might be
seen a posing a public hazard.

Do you want to release your data as Open Data?

A fact is something like "the triple point of water is 273.16 K"
or "the SMILES for methane is [CH4]". While gene patents
come close, facts concerning the natural world are not,
by themselves, protectable. There must be some transformative
or creative step to make a fact protectable under patent or
copyright law.

The difficulty is that a collection of facts might be protected.
For example, if you've used creative thought and expert opinion
in order to select the elements in the data set, then you may
have a specific database right. If you've arranged the data in
a novel and creative fashion, then that is also protected.
The specific details depends on your country's legal system;
in the UK, a database right exists if there is a "substantial
investment in obtaining [or] verifying" the data in the database.
You can easily see how that might apply to scientific data.

You can make your data Open Data by putting it in the public
domain. This is called a public domain dedication or waiver,
and it is not a license. The easiest and best solution is to use the
CC0 public domain dedication [http://creativecommons.org/about/cc0].
You may think it's okay to use a simpler version, like a
variation of the
SQLite [http://www.sqlite.org/copyright.html]
or PDB
PDB [http://www.rcsb.org/pdb/static.do?p=general_information/about_pdb/policies_references.html]
dedications. The problem is that these only cover copyright
and not the database, moral, or other related rights which might
also be in the data set.

You need to be careful. Just because you collected a bunch of
data doesn't mean that you have the right to dedicate it to
the public domain. For one, your employer or university might
be the legal rights holder, so check with them first. You need
to be careful that you don't violate the legal rights of others.
For example, suppose you develop a crowed-sourced collection
of chemical protocols, where commentary on each of the protocols
was contributed by others. Those people have a copyright interest
in the result, so you can't distribute your data set without their
permission. For this case we recommend that you require that
contributors license their contribution under the
Attribution-ShareAlike 3.0 Unported License [http://creativecommons.org/licenses/by-sa/3.0/].

And importantly, don't take someone else's proprietary data set,
extract a lot of the of records, and release it. Even if you
add a lot of new data, that's just plain illegal under copyright law.


===============
http://sourceforge.net/apps/mediawiki/blueobelisk/index.php?title=Open_Standards

We believe that Open Standards are necessary to promote scientific
data exchange, analysis services, and data archiving.

By Open Standards we mean that data formats, and control and exchange protocols
must be documented in such a way that person
"skilled in the art" [http://en.wikipedia.org/wiki/Person_having_ordinary_skill_in_the_art]
can produce software which can read, write, and exchange data with other
software which implements those Open Standard. Open Standards must be available
on non-discriminatory terms, must not require patent, trademark, or other
licensing, must not require a royalty fee of any sort, must not be protected
using DRM or other technical means, and must not prohibit reverse engineering.

For historical reasons, we acknowledge that older standards published in
the open literature and available through public research libraries and
the publishers, may be included as an Open Standard. Otherwise, Open
Standards must not require an access fee, registration, or license agreement
in order to access and use the documentation, and must not prevent redistribution
of copies of the standard, including by DRM or copyright restrictions.

We take no position on if Open Standards must allow derivative works. We
specifically exclude trademark issues so long as software may use the
Open Standard without trademark permission. We take no position on if
an Open Standard requires public involvement or feedback.


Do you want to release your data as an Open Standard?

A standard is just documentation. The easiest way to make it an
Open Standard is to release the document under one of the open
content licenses. Do you want others to be able to develop
variations of the standard? Then we suggest using the
Attribution 3.0 (CC BY 3.0) [http://creativecommons.org/licenses/by/3.0/]
or
Attribution-ShareAlike 3.0 (CC BY-SA 3.0) [http://creativecommons.org/licenses/by-sa/3.0/]
licenses. Otherwise, if you want to control the standards document
(which is not the same as controlling the standard) then we suggest
the
Attribution-NoDerivs 3.0 (CC BY-ND 3.0) [http://creativecommons.org/licenses/by-nd/3.0/].

Unfortunately, standards may depend on patents, and patent issues are
not part of the above licenses. You need to be aware if your standard
requires any patent. You may also need to require that participants
in the standards development waive their patent claims before joining,
but so far a lack of patent waiver has not caused problems.

You may also release a reference implementation or validation dataset
as part of the specification. These should be released as
Open Source [http://sourceforge.net/apps/mediawiki/blueobelisk/index.php?title=Open_Source]
and
Open Data [http://sourceforge.net/apps/mediawiki/blueobelisk/index.php?title=Open_Data],
respectively.

[Peter Murray-Rust]

Thank you Andrew for this extensive and constructive suggestion.
[Please note that I personally do not have write access to the BO pages].

I would be happy to forward it to the OKF open-science group for comment.

[Craig James]

Coincidentally, this was on SlashDot today:

"Five leading Internet standards bodies have joined together to articulate
a set of guidelines for the creation of open standards that they
say will foster
continued innovation, competition and interoperability in the
Internet industry. ..."

http://www.networkworld.com/news/2012/082912-openstand-262026.html?hpg1=bn

The guidelines they arrived at are here. Good stuff, but very broad.

http://www.networkworld.com/news/2012/082912-openstand-262026.html?hpg1=bn

Craig

[Peter Murray-Rust]

I am part of a group of Open advocates and there is very considerable
concern over this.The policy addresses FRAND (
http://en.wikipedia.org/wiki/Reasonable_and_non-discriminatory_licensing) .
One member writes: "The basic problem is that it seems to bless frand
standards, which allow patents". There is particular concern that W3C has
agreed to this.

I am hoping to get a formal indication as to why they are unhappy with
FRAND. So I can only flag a warning at present.

[Andrew Dalke]

Post by Craig James
"Five leading Internet standards bodies have joined together to articulate
a set of guidelines for the creation of open standards ...

Interesting. The principles are laid out at http://open-stand.org/principles/

A difficulty in applying those principles to Blue Obelisk's
use of "Open Standards" is:

Due process: Decisions are made with equity and fairness among
participants. No one party dominates or guides standards development.
...
Balance. Standards activities are not exclusively dominated
by any particular person, company or interest group.

Consider PMR's CML or my chemfp. One party dominated/guided those
respected standards.


I am influenced by this interview:
http://blogs.computerworlduk.com/open-enterprise/2012/04/interview-with-charles-h-schulz-on-open-standards/index.htm

An open standard is a standard whose development and
distribution have not fallen prey to vendor capture.
In other words an open standard is the result of an
open, inclusive, participative development and
standardisation, and a standard that can be used without
any restriction.

There are many standards that do not belong to this category.
There are standards that are developed in close cooperation
by a set of coopted partners for instance, standards that
were directly or indirectly developed under the dominance
and exclusive control of one party; and then there are of
course standards whose distribution and usage are conditioned
upon the explicit or implicit acknowledgement of specific
rights from “upstream parties”, such as patents.

I struggle to come up with a standard in this field which
is an open standard by this definition. I know that 22 years
ago there was the SMD structure format as an attempt to replace
MDL's connection table formats, but it failed.

Otherwise, everything I know of was done primarily under the
exclusive control of one party.


This is why I excluded community involvement and lack of
vendor capture in the writeup I did which kicked off this thread.



Actually, very little of that list applies to any chemistry "Open
Standard", if only because that list refers to actual standards
bodies.

I am also suspicious about:

Given market diversity, fair terms may vary from royalty-free
to fair, reasonable, and non-discriminatory terms (FRAND).

The FSF says:

Standards bodies that promulgate patent-restricted standards
that prohibit free software typically have a policy of
obtaining patent licenses that require a fixed fee per
copy of a conforming program.
http://www.gnu.org/philosophy/words-to-avoid.html#RAND

and Simon Phipps commentary on "Why RAND Is Bad For Open Source" is at:
http://blogs.computerworlduk.com/simon-says/2012/04/open-standards-consultation-guide/index.htm


I do not believe the "Open Stand" principles are appropriate
for the Blue Obelisk.


Andrew
***@dalkescientific.com

[Peter Murray-Rust]

Thanks,


A difficulty in applying those principles to Blue Obelisk's

Due process: Decisions are made with equity and fairness amongleast
participants. No one party dominates or guides standards development.
...
Balance. Standards activities are not exclusively dominated
by any particular person, company or interest group.
Consider PMR's CML or my chemfp. One party dominated/guided those
respected standards.
For CML Henry Rzepa was an equal partner throughout, of course. Note also

that several other people from several labs have been involved in writing
the various peer-reviewed sub-parts of CML. (Stefan Kuhn, Egon Willghagen,
Robert Lancashire, for example). The process was informal, and anyone that
really wanted to be involved could be. The reality is that it has taken 18
years and I suspect strongly that if votes had been taken all the way
through it wouldn't have lasted 2 years.

I *have* participated in a completely open meritocratic process when I
catalysed the creation of the SAX protocol:
http://www.saxproject.org/sax1-history.html
This was remarkable in that it lasted one month and in David Megginson's
words:

[SAX] was released on Monday 12 January 1998, one month less a day after
the beginning of the discussion. This could be a record for an industry
initiative

This is very infrequent, of course!! It is commoner for standards to run
into mud and stall.

This is a good time to announce on this list that the continued development
of CML will be taken on board by a group run by PNNL US and CSIRO AU. I
shall continue to be involved but IMO National laboratories are the natural
place for the development of community activities and their maintenance and
regulation and that is why Henry and I approached them. CML is stable -
there have been no significant alterations to the spec for several years,
it is widely used, there is a large code body and body of examples.

http://blogs.computerworlduk.com/open-enterprise/2012/04/interview-with-charles-h-schulz-on-open-standards/index.htm
An open standard is a standard whose development and
distribution have not fallen prey to vendor capture.
In other words an open standard is the result of an
open, inclusive, participative development and
standardisation, and a standard that can be used without
any restriction.
There are many standards that do not belong to this category.
There are standards that are developed in close cooperation
by a set of coopted partners for instance, standards that
were directly or indirectly developed under the dominance
and exclusive control of one party; and then there are of
course standards whose distribution and usage are conditioned
upon the explicit or implicit acknowledgement of specific
rights from “upstream parties”, such as patents.
I struggle to come up with a standard in this field which
is an open standard by this definition. I know that 22 years
ago there was the SMD structure format as an attempt to replace
MDL's connection table formats, but it failed.

I think SAX qualifies, and is typical of the difference of approach in ICT.
Speed of development in other domains is slower and generally much more has
to be built.

I think the CIF standard is an example of an excellent process - fully
controlled and run by the community for the community. InChI is also in
this category.Most of the rest relies on individuals or groups getting
something going, proving it and building a significant proof of concept.

Otherwise, everything I know of was done primarily under the
exclusive control of one party.
This is why I excluded community involvement and lack of
vendor capture in the writeup I did which kicked off this thread.
Actually, very little of that list applies to any chemistry "Open
Standard", if only because that list refers to actual standards
bodies.
Given market diversity, fair terms may vary from royalty-free
to fair, reasonable, and non-discriminatory terms (FRAND).
Standards bodies that promulgate patent-restricted standards
that prohibit free software typically have a policy of
obtaining patent licenses that require a fixed fee per
copy of a conforming program.
http://www.gnu.org/philosophy/words-to-avoid.html#RAND
http://blogs.computerworlduk.com/simon-says/2012/04/open-standards-consultation-guide/index.htm
I do not believe the "Open Stand" principles are appropriate
for the Blue Obelisk.

Nor do I

[Andrew Dalke]

Post by Peter Murray-Rust
[Please note that I personally do not have write access to the BO pages].

So, umm, who does?

Post by Peter Murray-Rust
For CML Henry Rzepa was an equal partner throughout, of course.

We have had this discussion before. My view was and is that
CML, SMILES, MDL's format, InChI, and every other format
in this field [qualification: cheminformatics] that I can
think of was done through something different that what
Schultz calls an open standard.

Well, maybe mmCIF is an exception, but it's not a format
I've had to deal with for 15 years. I guess the PDB format
falls in the same category.

(I put InChI on the list because the main development was
done under NIST and IUPAC, and as far as I know there is
still no standard for it other than 'what the code does',
and there is no goal for interoperability or data exchange;
it's a one-way algorithm. In any case, it's on the border
because since it's done with a very different
and more formal approach to standards development.)


The fact that two people were equal partners seems little
different than the "standards that are developed in close
cooperation by a set of coopted partners, for instance".

As Craig rightly points out, this field is simply too small
for Schultz's definition of open specification to apply.

I did work on the DAS2 spec in bioinformatics and even
an OMG/CORBA spec for sequence objects. Bioinformatics
seems big enough that open specifications are possible.
There are many more people doing bioinformatics work,
more data and more analysis software, and more data
sharing in that field than cheminformatics.

In that field, the "DDBJ/EMBL/GenBank" feature table
is an open specification, I believe.

Post by Peter Murray-Rust
This is a good time to announce on this list that the continued development of CML will be taken on board by a group run by PNNL US and CSIRO AU.

Congratulations.

Post by Peter Murray-Rust
I struggle to come up with a standard in this field which
is an open standard by this definition. I know that 22 years
ago there was the SMD structure format as an attempt to replace
MDL's connection table formats, but it failed.
I think SAX qualifies, and is typical of the difference of approach in ICT. Speed of development in other domains is slower and generally much more has to be built.

I don't think of SAX as a standard in this field. I think
of it more as a standard in the XML field. Otherwise I
could say that ASCII, Unicode, C, C++, HTML, HTTP, SMTP, SQL, and
a dozen other common standards fit the definition.

Post by Peter Murray-Rust
I think the CIF standard is an example of an excellent process -
fully controlled and run by the community for the community.

I'm showing my provinciality. When I say "this field" I wasn't
thinking about xtal structures at all. I've never worked with
CIF data. BO of course includes more than cheminformatics software.


Andrew
***@dalkescientific.com

[Peter Murray-Rust]

Post by Andrew Dalke

Post by Peter Murray-Rust
[Please note that I personally do not have write access to the BO pages].

So, umm, who does?

I guess Egon at least. (It's possible I have a login and forgotten the pw)

Post by Andrew Dalke

Post by Peter Murray-Rust
For CML Henry Rzepa was an equal partner throughout, of course.

We have had this discussion before. My view was and is that
CML, SMILES, MDL's format, InChI, and every other format
in this field [qualification: cheminformatics] that I can
think of was done through something different that what
Schultz calls an open standard.

This wasn't to argue the point - it was to make sure that Henry's
contribution was recognized.

I suggested the "mantra"- ODOSOS as a phrase to concentrate around. It has
neatly encapsulated the efforts of the BO. Nothkng in the BO absolutely
defines anything - there is a zen-like quality to it and the community is
the most important thing. OS(t****) fitted into the rhythm. The key thing
is interoperability and community - which we have achieved.We have no
membership, no budget, no agenda - just a mantra and dinners.

Post by Andrew Dalke
Well, maybe mmCIF is an exception, but it's not a format
I've had to deal with for 15 years. I guess the PDB format
falls in the same category.

No, PDF was developed at Brookhaven. mmCIF really came out of RCSB. But the
point is that CIF is driven by the community with a community process.
There is a committee - COMCIFS - I'm on it by invitation. There's a strong
sense of meritocracy - anyone working in the field - writing code,
specifications, etc. is likely to be coopted. IMO it's the sort of process
that IUPAC should adopt more often. (IUPAC groups are also mainly bottom-up
- IUPAC responds asynchronously to calls for projects and gives modest
amounts of support).

Post by Andrew Dalke
(I put InChI on the list because the main development was
done under NIST and IUPAC, and as far as I know there is
still no standard for it other than 'what the code does',
and there is no goal for interoperability or data exchange;
it's a one-way algorithm. In any case, it's on the border
because since it's done with a very different
and more formal approach to standards development.)
It *is* a IUPAC-owned process and there is now an InChI Trust run by Steve

Heller

Post by Andrew Dalke
The fact that two people were equal partners seems little
different than the "standards that are developed in close
cooperation by a set of coopted partners, for instance".

See above

Post by Andrew Dalke
As Craig rightly points out, this field is simply too small
for Schultz's definition of open specification to apply.
Agreed.
I did work on the DAS2 spec in bioinformatics and even
an OMG/CORBA spec for sequence objects. Bioinformatics
seems big enough that open specifications are possible.
There are many more people doing bioinformatics work,
more data and more analysis software, and more data
sharing in that field than cheminformatics.

I am an author (among 95 others of MIABE). It's as close as wecan get to
community-driven and owned.

Post by Andrew Dalke

Post by Peter Murray-Rust
This is a good time to announce on this list that the continued

development of CML will be taken on board by a group run by PNNL US and
CSIRO AU.
Congratulations.

I will write more.

Post by Andrew Dalke

Post by Peter Murray-Rust
I struggle to come up with a standard in this field which
is an open standard by this definition. I know that 22 years
ago there was the SMD structure format as an attempt to replace
MDL's connection table formats, but it failed.

Most things fail!

There was also MIF, based on CIF for molecules.

Post by Andrew Dalke

Post by Peter Murray-Rust
I think SAX qualifies, and is typical of the difference of approach in

ICT. Speed of development in other domains is slower and generally much
more has to be built.
I don't think of SAX as a standard in this field. I think
of it more as a standard in the XML field.

Agreed. I had thought the discussion was general when I wrote that.

Post by Andrew Dalke

Post by Peter Murray-Rust
fully controlled and run by the community for the community.
I think the CIF standard is an example of an excellent process -

I'm showing my provinciality. When I say "this field" I wasn't
thinking about xtal structures at all. I've never worked with
CIF data. BO of course includes more than cheminformatics software.
There is a symbiosis between community - specifications - software. If any

of the three components is missing then it is difficult for ODOSOS to
emerge. Where, for example, is NIST's UNITsML - at least 15 years old or
ANiML for analytical. Theye may be open, but they are not yet used. AFAIK.
Most BO material gets used eventually or mutates into something else.

Post by Andrew Dalke
Andrew
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Blueobelisk-discuss mailing list
https://lists.sourceforge.net/lists/listinfo/blueobelisk-discuss

[Peter Murray-Rust]

Here's a reference from my colleagues ...

Lundell, B. (2012) Why do we need Open Standards?, In Orviska, M. and
Jakobs, K. (Eds.) Proceedings 17th EURAS Annual Standardisation Conference
'Standards and Innovation', The EURAS Board Series, Aachen, ISBN:
978-3-86130-337-4, pp. 227-240.

[Andrew Dalke]

This wasn't to argue the point - it was to make sure that Henry's contribution was recognized.

Thank you for the clarification. I did indeed omit recognizing him.

I suggested the "mantra"- ODOSOS as a phrase to concentrate around. It has neatly encapsulated the efforts of the BO. Nothkng in the BO absolutely defines anything - there is a zen-like quality to it and the community is the most important thing. OS(t****) fitted into the rhythm. The key thing is interoperability and community - which we have achieved. We have no membership, no budget, no agenda - just a mantra and dinners.

That "wishy-washy be-in" quality is one of the main reasons I don't
consider myself aligned with the BO. You'll note that in my proposed
text I made specific statements of "we believe" and "we recommend"
and even several "musts." Without at least a good approximation to
a decent definition, how does one decide if a project is ODOSOS?

As a case in point, the original CIF definition notoriously used
patent law in order to "protect STAR File and/or CIF from use for
commercial gain without prior permission." Was the CIF format
during that time an Open Specification according to Blue Obelisk ODOSOS
principles? If there are no principles, then the answer is "sure,
why not? Their zen is in the right place."

I say that the CIF format at that time was not open. I also said
it during that time.

No, PDB was developed at Brookhaven.

Based on what I read in http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2143743/pdf/9232661.pdf :

In order to establish the PDB, acceptance by the crystallographic
community was necessary, requiring a pilgrimage in 1970 to the Medical
Research Council (MRC) laboratory and Crystal Data Centre (CDC) in
Cambridge. One result of this exchange was a concession that coordinates
of protein structures would be stored in the same format as the small
molecule CDC database (with a redundant ATOM label at the beginning of
each card), retaining the now-arcane counting number at the end. But the
idea of a PDB was accepted by Professors Pemtz, Blow, Kennard, Diamond,
and colleagues in Cambridge.
...
Earlier in 1971, the meeting of the American Crystallographic Association
in North Carolina had an ad hoc session where the need for a PDB and some
of the requirements of the user community were discussed. The second
Alpbach meeting in early 1972 gave me the first opportunity to present
the PDB informally to the European crystallographic community.

I can't help but interpret this as saying that the original PDB
format was actually developed at the CDC, not Brookhaven! And that
there was a lot of community involvement.


I read from the November 1976 newsletter that they upgraded to the new 80
column format in 1976, and there was some sort of review by depositors:

After approval by depositors, new-style entries for old parameter sets
replace the original ones.
...
As depositors and users who have examined data in the new format will
be aware, this new file structure can accommodate much additional information
not previously included in the Bank. ... Full details of the file structure
would take too much space here but a complete description is available
by filling out and returning the request form.


The PDB format included changes recommended by users. In October 1981, I see

Over the course of the next few years we anticipate making some changes
in the format of our atomic coordinate entries. ... Because of user
requests we are reserving a class of records for user definition and use.
All records beginning with the four letters USER are hereby reserved ...


I also see many places in the newsletters where they solicited comments and
suggestions. An extensive set of changes occurred with the first NMR
submissions, described in January 1989.

To accomodate differences between these data and those derived from
diffraction measurements, a number of new PDB record types will be
instituted. ... While developing the scheme outlined above to
accomodate NMR-based structures to the PDB format, we are also in
the process of evaluating a number of format upgrades or extensions.
Suggestions along these lines will be welcome, and should be sent to ...

leading up to January 1993,

The initial meeting of the PDB Format Upgrade Committee was held
... to help the PDB implement a comprehensive upgrade of its
interchange format in order to better serve the futures needs of
the community.
...
An outcome of the Committee's discussion is the implementation of
a number of changes to the current PDB format.
...
PDB staff members will be discussing issues related to this project
at forthcoming scientific meetings. Suggestions and comments are
welcome at all times...

The members of the Upgrade Committee were listed in the April 1992
issue. I list here only the affiliations:

BIOSYM, Rutgers, Merck Sharp and Dohme Research, NIH, Univ. of Uppsala,
Osaka Univ., UCSF, Duke, Welch Medical Library, Sterling Winthrop,
Univ. of London, Upjohn, Univ. of Brussels, NLM/NCBI

Then of course the mmCIF development started in 1993 out of that
upgrade effort, but the traditional 80 column format of the PDB
continued to be developed in collaboration with others.


So I don't see how the "PDB was developed at Brookhaven" is less
open than "But the point is that CIF is driven by the community
with a community process." Looking at the CIF history, described
in "The Crystallographic Information File (CIF): a New Standard
Archive File for Crystallography", Acta Cryst. (1991). A47, 655-685
at http://www.iucr.org/iucr-top/cif/standard/cifstd1.html :

1) CIF is "based on the Self-Defining Text Archive and Retrieval
(STAR) procedure of Hall", and
2) "At a meeting of the IUCr WPCI, held in conjunction with the
XI European Crystallographic Meeting (1988) in Vienna, it
was decided to develop a universal file based on the
Self-Defining Text Archive and Retrieval (STAR) procedure
of Hall (1991a)"
3) "The WPCI commissioned the authors [Hall, Allen, and Brown] to
develop a universal exchange file to be called the
Crystallographic Information File (CIF)."

Based on that, it appears that the syntax of CIF was developed
at or before 1988 by Hall, and the data dictionaries of CIF
were then developed by three people commissioned by the IUCr
Working Party on Crystallographic Information. These people
received much input from a large number of people, including
people at the PDB. The work was then approved by the IUCr.

I guess the only difference I see is that the IUCr was, as
Björn Lundell's "Why do we need Open Standards?" slides state:

The standard is adopted and will be maintained by a
not-for-profit organisation, and its ongoing development
occurs on the basis of an open decision-making procedure
available to all interested parties (consensus or majority
decision etc.).

The "not-for-profit organization" sets a high bar, so none
of the Blue Obelisk efforts can be an open standard. The
Brookhaven PDB work was non-for-profit, but of course it
did not have an "open decision-making procedure available
to all interested parties." I know nothing about the current
RCSB efforts.

But the point is that CIF is driven by the community with a community process. There is a committee - COMCIFS - I'm on it by invitation. There's a strong sense of meritocracy - anyone working in the field - writing code, specifications, etc. is likely to be coopted. IMO it's the sort of process that IUPAC should adopt more often. (IUPAC groups are also mainly bottom-up - IUPAC responds asynchronously to calls for projects and gives modest amounts of support).

The governance of the COMCIFS is that working groups have autonomy,
drafts are submitted to COMCIFS for approval. So while the working group
has a "community with a community process", it's these five people

• J. R. Hester (Chair)
• H. J. Bernstein
• R. W. Grosse-Kunstleve, Lawrence Berkeley National Laboratory
• B. McMahon (Coordinating Secretary), IUCr )
• J. Westbrook, Research Collaboratory for Structural Bioinformatics/Rutgers Univ.
- http://www.iucr.org/resources/cif/comcifs/members

who get to decide what CIF is. How does one get to be a member of
this voting group? Is voting membership 'available to all interested
parties'? No, because "COMCIFS should consist of about 6 voting members
appointed by the Executive Committee of the IUCr on the recommendation
of COMCIFS."


Let me stress that I do not think that CIF needs to be more transparent
or open. I am pointing out that it does not fall under the same
definition of open standard as others give, including in your reference to
Björn Lundell's "Why do we need Open Standards?", and I don't agree
with the suggestion that the PDB format was any less under the ODOSOS banner
than CIF.

It *is* a IUPAC-owned process and there is now an InChI Trust run by Steve Heller

Yes. I was describing the history, not the current situation.

My comment that InChI does not provide a open specification still
stands. I am unable to write a replacement for it except by reading
though the source code.

(I am realizing too that there is a distinction in my head between
an "open standard" and an "open specification", and that I am more
concerned about open specifications. In that case, I can accept that
InChI is an open standard, but it is not an open specification. I
will have to think about that some more. Perhaps the mantra should be
ODOSOSOS? :)


Cheers,

Andrew
***@dalkescientific.com

[Peter Murray-Rust]

Thanks,
The useful thing to take from this is that there is a dialectic - I and you
learn from each other.

Post by Andrew Dalke
That "wishy-washy be-in" quality is one of the main reasons I don't
consider myself aligned with the BO. You'll note that in my proposed
text I made specific statements of "we believe" and "we recommend"
and even several "musts." Without at least a good approximation to
a decent definition, how does one decide if a project is ODOSOS?

No one is committed to be involved if they don't want to. It's primarily
driven by people who want to create Open resources and want to work
together.

The good thing is that it has created a lot of valuable resources which, I
would like to believe, would have been less without the community.

There are no absolutes.If ODOSOS runs out of rail, so be it. I don't think
it has yet. We aren't committed to (say) GPL-virality or passionately
anti-GPL. We embrace commercial organizations where they want to be
involved and leave them alone if they don't. Our software is used by many
of them, happily as far as we know.

How do we decide if something is ODOSOS? the issue hasn't come up - in that
we haven't rejected anything or anyone. Anyone can come to a BO dinner,
post on this list. That's the extent of the activities. Blue Obelisks have
been given to commercial people.

I agree it's not formal. What have we lost? My guess - which you have also
mentioned in the last 2-3 days - is that too much striving will simply
drive people away. Our products are limited by the energy that individual
bring to the process. If someone wants to makes things more formal, no one
is stopping that, but they would need to convince us that it's worth our
while to spend time on that as opposed to writing code

Post by Andrew Dalke
As a case in point, the original CIF definition notoriously used
patent law in order to "protect STAR File and/or CIF from use for
commercial gain without prior permission." Was the CIF format
during that time an Open Specification according to Blue Obelisk ODOSOS
principles? If there are no principles, then the answer is "sure,
why not? Their zen is in the right place."

Time changes things. I was influenced at the time to think that trade marks
and patents were a useful idea. Now I am against them. I learn from
experience and change my mind - it's a positive thing to do. When I was
heavily involved in InChI I argued for a restriction in the licence. Now I
would criticize that opinion. It's clear that in all the complexities very
few people get things right first time. Remmeber the fuss we had with LGPL
and InChI and CMl and... If we had foreseen that I think I would have
argued for a BSD or even weaker. But you cannot work all this out at the
start.

Post by Andrew Dalke
I say that the CIF format at that time was not open. I also said
it during that time.

No, PDB was developed at Brookhaven.

Based on what I read in

I'll accept your account - it's very detailed and jogs my memory.

Post by Andrew Dalke
I guess the only difference I see is that the IUCr was, as
The standard is adopted and will be maintained by a
not-for-profit organisation, and its ongoing development
occurs on the basis of an open decision-making procedure
available to all interested parties (consensus or majority
decision etc.).
The "not-for-profit organization" sets a high bar, so none
of the Blue Obelisk efforts can be an open standard. The
Brookhaven PDB work was non-for-profit, but of course it
did not have an "open decision-making procedure available
to all interested parties." I know nothing about the current
RCSB efforts.

But the point is that CIF is driven by the community with a community

process. There is a committee - COMCIFS - I'm on it by invitation. There's
a strong sense of meritocracy - anyone working in the field - writing code,
specifications, etc. is likely to be coopted. IMO it's the sort of process
that IUPAC should adopt more often. (IUPAC groups are also mainly bottom-up
- IUPAC responds asynchronously to calls for projects and gives modest
amounts of support).
The governance of the COMCIFS is that working groups have autonomy,
drafts are submitted to COMCIFS for approval. So while the working group
has a "community with a community process", it's these five people
• J. R. Hester (Chair)
• H. J. Bernstein
• R. W. Grosse-Kunstleve, Lawrence Berkeley National Laboratory
• B. McMahon (Coordinating Secretary), IUCr )
• J. Westbrook, Research Collaboratory for Structural
Bioinformatics/Rutgers Univ.
- http://www.iucr.org/resources/cif/comcifs/members
who get to decide what CIF is. How does one get to be a member of
this voting group? Is voting membership 'available to all interested
parties'? No, because "COMCIFS should consist of about 6 voting members
appointed by the Executive Committee of the IUCr on the recommendation
of COMCIFS."

The skill of most successful processes is that votes are avoided unless
necessary. Brian McMahon has been very skilful in running the CIF process
to avoid conflicts. But yes, if votes are required, there is a limited
number of people involved (I don't think I am a voting member).

Post by Andrew Dalke
Let me stress that I do not think that CIF needs to be more transparent
or open. I am pointing out that it does not fall under the same
definition of open standard as others give, including in your reference to
Björn Lundell's "Why do we need Open Standards?", and I don't agree
with the suggestion that the PDB format was any less under the ODOSOS banner
than CIF.

It *is* a IUPAC-owned process and there is now an InChI Trust run by

Steve Heller
Yes. I was describing the history, not the current situation.
Agreed,

many of these things have history. The early stages are usually ad hoc and
sometimes messy.

Post by Andrew Dalke
My comment that InChI does not provide a open specification still
stands. I am unable to write a replacement for it except by reading
though the source code.

I have been challenging this as well.

Post by Andrew Dalke
(I am realizing too that there is a distinction in my head between
an "open standard" and an "open specification", and that I am more
concerned about open specifications. In that case, I can accept that
InChI is an open standard, but it is not an open specification. I
will have to think about that some more. Perhaps the mantra should be
ODOSOSOS? :)

Some of these questions come down to basic democracy - which no-one has
solved or will.

*Churchill: "Democracy is the worst form of government, except for all
those other forms that have been tried from time to time." (from a House of
Commons speech on Nov. 11, 1947)*

We have Open Standards which are formally democratic but in fact influenced
by major lobbying (cf Microsoft) , we have BDFLs (Linux, Python, etc), we
have stalled community processes with formal committees and no software (I
spent 6 wasted months in a OMG project for open software for molecules -
complete waste of time other than I realised how the software needed
refactoring, The Blue Obelisk has a Doctor Who model of Open Source - some
Doctors regenerate, others are still going.

If you have software which can co-exist in BO, fine - it doesn't have to be
labelled as BO. (Note that we haven't been very active in branding although
I think it would be a good idea).

On the very positive front I now believe that we have offerings in nearly
all important areas. That means that it is relatively easy to construct and
Open solution *and potentially get communal involvement*. Few people want
to build add-ons for proprietary systems unless they are paid.

I think the value of the BO is increasing - it works in fits and starts and
there are several things I can see happening within the next year.


P.

Post by Andrew Dalke
Cheers,
Andrew
------------------------------ ------------------------------
------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Blueobelisk-discuss mailing list
https://lists.sourceforge.net/lists/listinfo/blueobelisk-discuss

[Peter Murray-Rust]

More informed and useful comment on OpenStand:
http://www.consortiuminfo.org/standardsblog/article.php?story=20120830102530600

[Peter Murray-Rust]

More on OpenStand
http://www.consortiuminfo.org/standardsblog/article.php?story=20120830102530600

[Andrew Dalke]

How do we decide if something is ODOSOS? the issue hasn't come up - in that we haven't rejected anything or anyone. Anyone can come to a BO dinner, post on this list. That's the extent of the activities. Blue Obelisks have been given to commercial people.

I am beginning to believe that the Blue Obelisk is little more
than a dinner club, which I expected more of a social movement.

If someone wants to makes things more formal, no one is stopping that, but they would need to convince us that it's worth our while to spend time on that as opposed to writing code

I have proposed concrete changes to three wiki pages. I believe
these capture the views of those who consider themselves aligned
with the ideals of the Blue Obelisk. I believe these help better
clarify what these ideals are and more likely that I will contribute
more support to the Blue Obelisk, and perhaps someday consider
myself a member.

I have heard precisely no complaints about the content of that
original post. Nor have I heard confirmation that anyone agrees
with any of the points, or the need for my effort. The only
statement has been:

Thank you Andrew for this extensive and constructive suggestion.
I would be happy to forward it to the OKF open-science group for comment.

Instead, the remainder of the thread has been commentary and
follow-up on the "Open Stand" principles which Craig pointed out.

I don't need people to stop me. I need people to say that they
are or are not convinced, and if not, to provide guidance on
what is lacking, or to say that this approach is inappropriate
to the goals of the Blue Obelisk.


Andrew
***@dalkescientific.com

[Peter Murray-Rust]

Post by Peter Murray-Rust
How do we decide if something is ODOSOS? the issue hasn't come up - in

that we haven't rejected anything or anyone. Anyone can come to a BO
dinner, post on this list. That's the extent of the activities. Blue
Obelisks have been given to commercial people.
I am beginning to believe that the Blue Obelisk is little more
than a dinner club, which I expected more of a social movement.

No - the dinners could disappear tomorrow and it wouldn't affect the BO.
It's simply that when we are at chemistry meetings we want to meet up with
the people we have corresponded with. It's fun, but not essential.

Remember this is a voluntary organization. People contribute what they can
when they can. There is no way we can compel anyone to do anything -
everything is meritocratic. There is no BDFL and no committee.

So what do we contribute? This is my own opinion and I'd welcome others.
* we assert, collectively, that ODOSOS matters. People have taken us
seriously. We are beginning to see people outside the grouo using "Blue
Obelisk" as a reference, either within chemistry or from other disciplines
* we communicate, just as we are doing now. There would be no natural
mailing list if we had not created it. The mailing list both helps with its
information and grows a sense of community
* we create shared resources - Data, the Shapdao Q/A system, reference pages
* we keep in touch with each others software. The Blue Obelisk legitimizes
anyone to approach anyone else. Of course we could do that anyway but O
find it makes it easier
* It brings in more people. There's a visible community in a way that there
isn't in (say) computation chemistry software
* When necessary we try to solve problems communally. Just as we may have
to do on the JavaVM issue. Most of the time it isnt necessary.
* we keep up-to-date more easily - we look at what we are all doing.
* we are create added value. The BO has published two papers, the last with
20 groups. I don't believe we could possibly have got that paper published
without an organisation.

That's not to say that there aren't other Open source chemistry programs
and they are valuable. They probably get on quite OK by themselves. But if
they want to interact it is clear that we exist and will be helpful where
possible.

"a social movement"? That's up to the individuals and groups - I can't
speak for them. Most are involved in either writing software to help their
own research or teaching, some run this as a commercial venture. Yes, I PMR
want to change the world, but I'm not foisting my ideas on the other BO
people and anyway it wouldn't work.

And remember that everyone other than me has day jobs and they normally
take precedence. I suspect that they have marginal time and find the way
things are at present works for them when they need it.

And yes,
It's valuable to have some self-examination occasionally.

If someone wants to makes things more formal, no one is stopping that,

but they would need to convince us that it's worth our while to spend time
on that as opposed to writing code


I have proposed concrete changes to three wiki pages. I believe

these capture the views of those who consider themselves aligned
with the ideals of the Blue Obelisk. I believe these help better
clarify what these ideals are and more likely that I will contribute
more support to the Blue Obelisk, and perhaps someday consider
myself a member.
I have heard precisely no complaints about the content of that
original post. Nor have I heard confirmation that anyone agrees
with any of the points, or the need for my effort. The only
Thank you Andrew for this extensive and constructive suggestion.
I would be happy to forward it to the OKF open-science group for comment.
Instead, the remainder of the thread has been commentary and
follow-up on the "Open Stand" principles which Craig pointed out.

As I said, the BO is composed of people :-)

I don't need people to stop me. I need people to say that they
are or are not convinced, and if not, to provide guidance on
what is lacking, or to say that this approach is inappropriate
to the goals of the Blue Obelisk.

I've written my thoughts above. I hope some others will contribute

P.

Andrew
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Blueobelisk-discuss mailing list
https://lists.sourceforge.net/lists/listinfo/blueobelisk-discuss

[Noel O'Boyle]

Post by Andrew Dalke

How do we decide if something is ODOSOS? the issue hasn't come up - in that we haven't rejected anything or anyone. Anyone can come to a BO dinner, post on this list. That's the extent of the activities. Blue Obelisks have been given to commercial people.

I am beginning to believe that the Blue Obelisk is little more
than a dinner club, which I expected more of a social movement.

If someone wants to makes things more formal, no one is stopping that, but they would need to convince us that it's worth our while to spend time on that as opposed to writing code

I have proposed concrete changes to three wiki pages. I believe
these capture the views of those who consider themselves aligned
with the ideals of the Blue Obelisk. I believe these help better
clarify what these ideals are and more likely that I will contribute
more support to the Blue Obelisk, and perhaps someday consider
myself a member.
I have heard precisely no complaints about the content of that
original post. Nor have I heard confirmation that anyone agrees
with any of the points, or the need for my effort. The only
Thank you Andrew for this extensive and constructive suggestion.
I would be happy to forward it to the OKF open-science group for comment.

Andrew, I think you should just go ahead and modify the pages. I
looked at the current text and yours seems like an improvement. I
think I explained on a separate thread how to do it. We can modify it
further once it's up on the wiki if there is disagreement.

Post by Andrew Dalke
Instead, the remainder of the thread has been commentary and
follow-up on the "Open Stand" principles which Craig pointed out.
I don't need people to stop me. I need people to say that they
are or are not convinced, and if not, to provide guidance on
what is lacking, or to say that this approach is inappropriate
to the goals of the Blue Obelisk.
Andrew
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Blueobelisk-discuss mailing list
https://lists.sourceforge.net/lists/listinfo/blueobelisk-discuss

[Peter Murray-Rust]

Post by Peter Murray-Rust

Post by Peter Murray-Rust
Thank you Andrew for this extensive and constructive suggestion.
I would be happy to forward it to the OKF open-science group for

comment.
Andrew, I think you should just go ahead and modify the pages. I
looked at the current text and yours seems like an improvement. I
think I explained on a separate thread how to do it. We can modify it
further once it's up on the wiki if there is disagreement.
I agree. I got distracted immediately after your post by the OpenStand

correspondence and only returned just now. Just go ahead

[Andrew Dalke]

The skill of most successful processes is that votes are avoided unless necessary. Brian McMahon has been very skilful in running the CIF process to avoid conflicts. But yes, if votes are required, there is a limited number of people involved (I don't think I am a voting member).

That is absolutely my point. The informal structure for CIF
is based on consensus building and community feedback, even
though the formal structure is based on voting by a small and
restricted number of people. (I listed all five voting members
and provided a link to the full list of members. You are
not a voting member.)

Just like the continued development of the PDB format was,
historically speaking, formally controlled by Brookhaven but
informally controlled by consensus building and community
feedback.

Just like the continued development of CML, until the near
future, was and is controlled by two people who had help and
feedback from the community.

Just like the development of SMILES by Daylight was
controlled by one company, but with the feedback and support
of the cheminformatics community.

And not at all like the historical MDL format, where the format
specification was proprietary until the user base rose up
and demanded a published definition or they would make and use
a new format of their own.


For better or worse, the Blue Obelisk movement cannot look
towards the formal structure of a open standard development
process to decide if something is an "Open Standard." It
must look instead primarily to what restrictions are placed
on its use, followed by how well the standards developer
supports the public and responds to or incorporates feedback.

(I place that second because I believe a standard remains
a standard even if there is no active development. I do
not want to obligate people to support their standards
forever.)

My view is why, years ago, I railed against the then-Blue-Obelisk
position that Daylight's SMILES was not an example of an Open
Standard. I managed to convince you all then. The text
I gave in the "ODOSOS text" thread gives more rigor to my
view, so the argument doesn't need to be reraised for every
individual project.

Andrew
***@dalkescientific.com

[Egon Willighagen]

Post by Peter Murray-Rust

Post by Andrew Dalke

Post by Peter Murray-Rust
[Please note that I personally do not have write access to the BO pages].

So, umm, who does?

I guess Egon at least. (It's possible I have a login and forgotten the pw)

These are the project admins: baoilleach, dleidert, egonw, ghutchis,
petermr, rajarshi, steinbeck

All have the possibility to create accounts, (re)set passwords, etc...

Here's the list of registered wiki users:
http://sourceforge.net/apps/mediawiki/blueobelisk/index.php?title=Special:ListUsers

IIRC, the procedure is like this:

- the new users gets a SourceForge account and lets this know to one
of the above admins
- the users uses that to go to the wiki and logs in
- that admin goes to
https://sourceforge.net/apps/mediawiki/blueobelisk/index.php?title=Special:UserRights
and adds this new SF account as 'editor'

Something like that...

Egon

[Craig James]

Post by Andrew Dalke

Post by Craig James
"Five leading Internet standards bodies have joined together to articulate
a set of guidelines for the creation of open standards ...

Interesting. The principles are laid out at http://open-stand.org/principles/
A difficulty in applying those principles to Blue Obelisk's
Due process: Decisions are made with equity and fairness among
participants. No one party dominates or guides standards development.
...
Balance. Standards activities are not exclusively dominated
by any particular person, company or interest group.
Consider PMR's CML or my chemfp. One party dominated/guided those
respected standards.

One party dominated or exclusively developed almost everything in
cheminformatics. It's mostly because we're a tiny group of people so
any one endeavour almost always is pushed by some particular person.

Our criterion should be along the lines of control rather than
participation. It's fine if only one party participates, but if
someone else wants to help and is qualified, they should be welcomed.

Post by Andrew Dalke
http://blogs.computerworlduk.com/open-enterprise/2012/04/interview-with-charles-h-schulz-on-open-standards/index.htm
An open standard is a standard whose development and
distribution have not fallen prey to vendor capture.

Exactly. Another way of saying "participation, but not control."

Post by Andrew Dalke
I struggle to come up with a standard in this field which
is an open standard by this definition. I know that 22 years
ago there was the SMD structure format as an attempt to replace
MDL's connection table formats, but it failed.
Otherwise, everything I know of was done primarily under the
exclusive control of one party.

Was it really exclusive control in all cases, or were there some where
others could have participated but didn't?

The original SMILES was controlled by Daylight, but OpenSMILES was
participatory. CML was controlled by Peter, but I suspect he would
have welcomed input. It doesn't really matter ... the past is the
past. It's just an interesting question.

Post by Andrew Dalke
I do not believe the "Open Stand" principles are appropriate
for the Blue Obelisk.

No, and I hope it didn't look like I was suggesting otherwise. "Open
Stand" is just an interesting document to review to see if there's
something OB can use or overlooked.

Craig