does anyone care about mail list threads organized by topic?

LAMMPS LAMMPS Mailing List Mirror
1

HI all - looking for some feedback …

For a few months we attempted to organize mail postings by topic:

http://lammps.sandia.gov/threads/topics.html

in addition to the auto-generated lists by thread and date:

http://lammps.sandia.gov/threads/threads.html
http://lammps.sandia.gov/threads/maillist.html

The problem is that it’s too labor intensive to assign each thread
to a topic, and I realize we’ll never keep up. The page is already

months behind.

So I plan to remove that page, hoping that Google and the SourceForge
maillist search is good enough to help users find old maillist threads
related to their Qs.

Does anyone have push-back, or want to volunteer to topic-ize
all the threads (good luck!) ?

Steve

2

Hey Steve,

I tend to use google when hunting the mailer list for specific topics with my search query in the format: [search query] site:lammps.sandia.gov. This seems to fill my needs fairly well. Although I wasn’t aware of the attempt to sort threads by topic.

As for if you wanted to continue trying to perform sorting, here are my two initial thoughts:
Would it be feasible to have some kind of filter that catches predefined keywords for autosorting? This may have issues when topics are more complex or encompass more than one topic if you want each thread listed under only one topic. Also, I imagine it would still require some individual to verify the topics are sorted appropriately and there will undoubtedly be emails that do not use any of the keywords that may be predefined.

Or, in the same line of thought, rather than one volunteer sorting new topics, asking individuals to provide some special keyword from a given list in the mailer-list submission guidelines. When they ask the mailer list something, they would put the keyword(s) at the top, possibly with some special character (thinking of the way many sites now have #[keyword]). But then you encounter the issues of actually getting people to include said keyword(s) or the case where their project doesn’t fall into any of the keywords.

Hope this helps,
Michael

3

Your second idea basically suggests a PAC-like code scheme for the mailing list. Not too hopeful on its success. The bulk of the Qs tend to come from people who in many cases don’t even understand the problem they are facing. I find hard to believe they will spend time looking down a list of keywords. Hey, but don’t listen to me, I’ve been called pessimistic sometimes in the past :wink:
Carlos

4

Haha. I admitted it would be troublesome getting people to use them. I was recalling people going “This is a [insert topic] question, not a LAMMPS question” when I wrote it, so people often don’t follow the guidelines already.

I was just trying to contribute ideas that may lead to a feasible solution without knowing what exactly it would take to implement them other than it would likely be troublesome to some degree but hopefully less troublesome than doing it manually.

5

Haha. I admitted it would be troublesome getting people to use them. I was
recalling people going "This is a [insert topic] question, not a LAMMPS
question" when I wrote it, so people often don't follow the guidelines
already.

I was just trying to contribute ideas that may lead to a feasible solution
without knowing what exactly it would take to implement them other than it
would likely be troublesome to some degree but hopefully less troublesome
than doing it manually.

there have been discussions on similar topics in the past.

here is a summary of my recollection of the pros and cons:

- several people would rather see a webforum than a mailing list.
those are mostly people *seeking* help. the benefit would be that
questions can be easily appended to related threads and some forums
have a scoring system where the visibility of individually discussions
is raised or lowered based on the number of views and thumbs up/down
from users. the flip side of that
coin is that for people providing answers, mailing lists are much less
effort since you don't need to log into a specific web page and search
around.

- also the topic of having a wiki or other community maintained
repository has come up a few times. the concern here is that LAMMPS
does not have a large enough community to make it work (just consider
the ratio of viewers/consumers vs. providers/writers on wikipedia).

- another discussion topic is whether questions on general MD should
be directed to a more general forum (shared with users from other MD
packages). that would be particularly considered helpful, if it had a
focus on MD beginners. however, the concern is again, who would be
answering questions. the same consideration and argument could be made
for people with an interest in software development with or on top of
LAMMPS.

my personal take is that the existing mailing list format has the
lowest maintenance overhead and anything else would require some time
commitment from a few volunteers. to make a webforum a success, it
needs a (small) group of attentive moderators that take care of the
quality of questions and answers. it also seems that an up/down voting
capability can be very helpful (see stackoverflow and related sites,
for example) to make it easier to find *good* answers.

a "hybrid" solution that i have made very good experiences personally
is to have a "Questions and Answers" document/webpage/forum. this
could be maintained by a small number of volunteers (best people that
are still in the process of learning, since they have a *much* better
sense for what is helpful to somebody seeking help) through having a
topical structure (similar to a wiki) where they "harvest" questions
and answers posted to the mailing list and edit them to contain only
the essential information for question and answer. most of the time,
people asking for help do not provide all necessary information to
give a good answer and then also the answers are often scattered
across multiple e-mails, e.g. when multiple people provide different
pieces of an answer or different perspectives. a rather simple
editorial process could boil this down to just the essential parts. if
a question is repeatedly asked, only the aspects that are new would
need to be added to a given topic. with this procedure we would slowly
have a fairly elaborate so-called knowledge base where everybody
benefits: people who need help get concise and complete information,
people who edit this get to learn a lot about MD and LAMMPS, and
experts only need to answer questions and can do this with little
effort via e-mail.

since there are limitations to what can be hosted on
lammps.sandia.gov, we just registered the lammps.org domain and could
set up a suitable webforum/wiki/webpage under that domain where it
will be easy to incorporate contributions by the LAMMPS community
outside of SNL.

with the growth of the LAMMPS user base and the change in how people
communicate, some adjustments *should* be made, but almost all of them
require somebody to take the initiative.

axel.

6

On the topic at hand, you can count me as no push back. Google does a fantastic job of indexing the list and I always find what I am searching for there.

-s-

7

I have talked about this in the past but will bring it up here once again. Putting some extra effort into convincing researchers to post their scripts into a properly organized database (has to be created first) has the potential of changing how science is done as well. I am not talking about shortcuts for the newbies to avoid learning. I am talking about active collaboration at a different level. Look for example at journal publishers; aren’t they clever? One does the work/research, writes it as good as possible, then gives away all copyrights to have it published in the publisher sites and on top of they charge one a fee $$$ to access the contents later!!! The system only work because one is ultimately paying to access other peoples ideas yet, the original workers get no compensation. Now, if I had access to all lammps scripts other researchers are willing to contribute I will be saving time putting together some systems, would probably be citing those people more often and eventually would care more abut this site than the journal where the original work ended up at. Sure all this require a critical mass of people willing to contribute. Yet, look at the openkim project, an analog idea along a different line but even funded by the US gov. Anyone?
Carlos

8

I have talked about this in the past but will bring it up here once again.
Putting some extra effort into convincing researchers to post their scripts

how?? we have a hard time as it is to get a complete input deck from
people that need help but don't do anything special.

nevertheless, i recently saw somebody posting this link to an
interesting discussion on the subject:
https://www.siam.org/news/news.php?id=2131

into a properly organized database (has to be created first) has the
potential of changing how science is done as well. I am not talking about
shortcuts for the newbies to avoid learning. I am talking about active
collaboration at a different level. Look for example at journal publishers;
aren't they clever? One does the work/research, writes it as good as
possible, then gives away all copyrights to have it published in the
publisher sites and on top of they charge one a fee $$$ to access the
contents later!!! The system only work because one is ultimately paying to
access other peoples ideas yet, the original workers get no compensation.
Now, if I had access to all lammps scripts other researchers are willing to
contribute I will be saving time putting together some systems, would
probably be citing those people more often and eventually would care more
abut this site than the journal where the original work ended up at. Sure
all this require a critical mass of people willing to contribute. Yet, look
at the openkim project, an analog idea along a different line but even
funded by the US gov. Anyone?

if you can "sell" things the right way and have a good "gimmick", it
is not so difficult to get funding. turning this funding into
something useful is a totally different thing, and i have seen a lot
of funding been wasted or just turned into a way of paying salaries
and then later doing some patching up to make it look as if the
project actually did something useful. if you want something with
serious staying power, you need more: you need to introduce a
different culture of doing things. and that is one of the most
difficult things to do.

axel.

9

What you describe has been funded as part of the Materials Genome Initiative and will be deployed at NIST. It is my understanding that this will to be visible to the public soon.

10

Are you talking about the openkim? That I am aware of.
Carlos

11

> I have talked about this in the past but will bring it up here once
again.
> Putting some extra effort into convincing researchers to post their
scripts

how?? we have a hard time as it is to get a complete input deck from
people that need help but don't do anything special.

That I am not sure. Maybe one can start by putting into this
lammps.orgwiki/forum a section on MD simulations with corresponding
practice scripts.
Kind of like an online book with example calculations of liquid densities,
elastic constants, phonon dispersion relations, the kind of stuff that
people simulate the most and can be achieved via lammps. Professors
teaching computational materials could set up their courses such as that
assignments on the topic are done via lammps and scripts are later
collected to upload to the site. The lammps team at sandia could have
contests with some $ prices asking contestants to compute properties such
as the one done in the "fuid properties contest" (
http://fluidproperties.org/). This could be extended to academics as well.
Yes, these might not be the best ideas but notice that I am giving you
some facts and no fluff.

nevertheless, i recently saw somebody posting this link to an
interesting discussion on the subject:
https://www.siam.org/news/news.php?id=2131

> into a properly organized database (has to be created first) has the
> potential of changing how science is done as well. I am not talking about
> shortcuts for the newbies to avoid learning. I am talking about active
> collaboration at a different level. Look for example at journal
publishers;
> aren't they clever? One does the work/research, writes it as good as
> possible, then gives away all copyrights to have it published in the
> publisher sites and on top of they charge one a fee $$$ to access the
> contents later!!! The system only work because one is ultimately paying
to
> access other peoples ideas yet, the original workers get no compensation.
> Now, if I had access to all lammps scripts other researchers are willing
to
> contribute I will be saving time putting together some systems, would
> probably be citing those people more often and eventually would care more
> abut this site than the journal where the original work ended up at. Sure
> all this require a critical mass of people willing to contribute. Yet,
look
> at the openkim project, an analog idea along a different line but even
> funded by the US gov. Anyone?

if you can "sell" things the right way and have a good "gimmick", it
is not so difficult to get funding.

ditto.

turning this funding into

something useful is a totally different thing, and i have seen a lot
of funding been wasted or just turned into a way of paying salaries
and then later doing some patching up to make it look as if the
project actually did something useful.

So, ever heard of venture capitalists? I am just one guy pushing forward
one idea.
Your negative/lack of enthusiasm is no reason to assume the outcome won't
be useful. You are just another guy pushing another idea.

if you want something with
serious staying power, you need more: you need to introduce a
different culture of doing things. and that is one of the most
difficult things to do.

That I know. Not like you, but I freely volunteer my little time in this
very forum trying to change the culture and the status quo :wink:
Carlos

12

actually, i would *love* to such projects succeed, but i am also a
dude that has heard the phrase "'somebody' should do... " and "the XXX
developers should ..." a few times too often. so before getting
invested into something myself, i want to see a bit more than just a
good pitch.

axel.

13

I totally get it. I am very aware of all your efforts on many different fronts regarding writing software for many different purposes and teaching workshops. My comments are in part oriented to testing some grounds. But, yes you are correct, I haven’t said “I do” thus I can see why my messages may be coming across only as pitches (was the "good pitch even related to my post :wink: ?.
Carlos

14

No, this is not related to openkim. As part of the Materials Genome Initiative, NIST is establishing essential data exchange protocols and the means to ensure the quality of materials data and models. A key aspect of this is the deployment of repositories. It is my understanding that the first repository will be visible to the public soon.

15

Thanks for the various comments. Two of my own:

I am not opposed if others (not me, no time) want to put
effort into starting up a forum or Q&A section
that answers common questions in a more
organized fashion. It could be on the new lammps.org

site that Axel mentioned and we’d link to it from
the main web site. You can use the database

of mail messages if you like, for content.

There are 2 links on the current web page (Scripts, Howto)
where people can post their scripts and tricks-of-the-trade.

But no one does. I agree it could be a good communicty
resource if lots of stuff was there, but it probably simply
isn’t how people like to work …

Steve