Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[Davide Del Vento]

>> When I used version
>> control for my solo research project, I used mercurial.  Mercurial is not
>> just for small projects.  Python uses mercurial for their open source
>> project.  Facebook recently started using Mercurial instead of Git [1].
>
> Note, Python is moving to github --
>
> http://www.snarky.ca/the-history-behind-the-decision-to-move-python-to-github

As Timothy mentioned in a subsequent message, and the link you provide
specifies, that has simply to do with the fact that "GitHub has
basically built a social network of open source contributors", which
bitbucket has failed to do.

___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[Terri Yu]

Dealing with git is a huge pain.  When I screw something up in git, I have
to take a deep breath, look things up on StackOverflow, and double check
all my commands, so that I don't break something.  The only reason I
learned it was because I had to learn it to contribute to open source
projects.  And when I try to help newbies make their first open source
contribution, the biggest roadblock is always git.  When I used version
control for my solo research project, I used mercurial.  Mercurial is not
just for small projects.  Python uses mercurial for their open source
project.  Facebook recently started using Mercurial instead of Git [1].

Terri

[1]
https://code.facebook.com/posts/218678814984400/scaling-mercurial-at-facebook/

On Mon, Feb 29, 2016 at 3:51 PM, David Martin (Staff) <
d.m.a.mar...@dundee.ac.uk> wrote:

> We spend about 50 contact hours teaching our undergraduates the basics of
> R. Even that is not enough. It has been said that you need 100 hours to
> reach competency, and 1000 hours to master a subject. And the next stage is
> 10,000 hours to be an expert..
>
> How much time has he invested in actually learning those skills? I was
> totting up the time we spend teaching X versus the amount of complaints we
> get that the students don't know X. There is a strong inverse correlation.
> Folk want a cheap easy fix and have been promised that with computers. It
> does exist but it has to be earned. You don't get cheap and easy for free.
>
> Dr David Martin
> Lecturer in Bioinformatics
> College of Life Sciences
> University of Dundee
>
>
> 
> From: Discuss  on behalf of
> Lex Nederbragt 
> Sent: 29 February 2016 20:43
> To: Software Carpentry Discussion
> Subject: Re: [Discuss] RajLab: From reproducibility to over-reproducibility
>
> Hi all and thanks for the many responses.
>
> My feeling reading this post was about tools (partly echoing Greg): 'we'
> know 'all of us' should use the appropriate tool, e.g. version control (is
> that what you call the moral high-ground?). But for the novice, these
> tools/methods have steep learning curves, thus high upfront time
> investment, and no immediate benefit (!!!). There is a burden on 'us' to
> convince 'others' of the need to invest time to adopt these tools.
>
> I am not sure whether more convincing (how? Research-based evidence?) or
> training is the answer, versus much easier to learn and use tools.
>
>Lex
> ___
> Discuss mailing list
> Discuss@lists.software-carpentry.org
>
> http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org
>
> The University of Dundee is a registered Scottish Charity, No: SC015096
>
> ___
> Discuss mailing list
> Discuss@lists.software-carpentry.org
>
> http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org
>
___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[David Martin (Staff)]

We spend about 50 contact hours teaching our undergraduates the basics of R. 
Even that is not enough. It has been said that you need 100 hours to reach 
competency, and 1000 hours to master a subject. And the next stage is 10,000 
hours to be an expert..

How much time has he invested in actually learning those skills? I was totting 
up the time we spend teaching X versus the amount of complaints we get that the 
students don't know X. There is a strong inverse correlation. Folk want a cheap 
easy fix and have been promised that with computers. It does exist but it has 
to be earned. You don't get cheap and easy for free.

Dr David Martin
Lecturer in Bioinformatics
College of Life Sciences
University of Dundee



From: Discuss  on behalf of Lex 
Nederbragt 
Sent: 29 February 2016 20:43
To: Software Carpentry Discussion
Subject: Re: [Discuss] RajLab: From reproducibility to over-reproducibility

Hi all and thanks for the many responses.

My feeling reading this post was about tools (partly echoing Greg): 'we' know 
'all of us' should use the appropriate tool, e.g. version control (is that what 
you call the moral high-ground?). But for the novice, these tools/methods have 
steep learning curves, thus high upfront time investment, and no immediate 
benefit (!!!). There is a burden on 'us' to convince 'others' of the need to 
invest time to adopt these tools.

I am not sure whether more convincing (how? Research-based evidence?) or 
training is the answer, versus much easier to learn and use tools.

   Lex
___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

The University of Dundee is a registered Scottish Charity, No: SC015096

___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[Lex Nederbragt]

Hi all and thanks for the many responses. 

My feeling reading this post was about tools (partly echoing Greg): 'we' know 
'all of us' should use the appropriate tool, e.g. version control (is that what 
you call the moral high-ground?). But for the novice, these tools/methods have 
steep learning curves, thus high upfront time investment, and no immediate 
benefit (!!!). There is a burden on 'us' to convince 'others' of the need to 
invest time to adopt these tools.

I am not sure whether more convincing (how? Research-based evidence?) or 
training is the answer, versus much easier to learn and use tools.

   Lex
___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[Davide Del Vento]

> if you’re like me, you will screw up at some point, leading to some problem,
> potentially catastrophic, that you will spend hours trying to figure out.
> I’m clearly not alone... “Abort: remote heads forked” anyone? :) At that
> point, we all just call over the one person in lab who knows how to deal
> with all this crap and hope for the best. And look, I’m relatively computer
> savvy, so I can only imagine how intimidating all this is for people who are
> less computer savvy.
>
>
> I heard fewer complaints of this kind, and believe that I saw higher
> adoption rates, when we were teaching Subversion.  I'm not going to
> recommend that we switch back, but I do miss it whenever I have to help
> someone deal with a detached head...
>

I fully agree with that. I can't understand why git has become so
popular, compared to mercurial which in my opinion is as easy as
subversion. To people thinking "github!" the answer is "bitbucket!"
Same features as github, plus being free for non-public repos (which
are paid on github). It's VHS vs Betamax once again, however in that
case you can google a stupid reason why the former became more popular
than the latter. For git-vs-hg I'm at a loss.

To answer Lex, the response I give is the following:

If you feel "pain" in using a tool (e.g. git), you are either doing it
wrong, or not proficient enough. You have to figure out which case it
is, and act accordingly.

Take running. If you feel pain in your legs while running, you are
either using them in the wrong way, or need more/different exercise to
strengthen them. It's ok to walk home upside down on your hands, if
they hurt that badly, but that cannot be your routine after each
workout. Talk to your coach. It's ok to walk upside down to meet that
deadline for that paper, but you ain't thinking of keep doing that way
forever, are you?

I am convinced that any of us can be that coach, given enough personal
contact with the team affected by these problems.

Cheers,
Davide

___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[Erik Bray]

On Mon, Feb 29, 2016 at 5:38 PM, Bennet Fauber  wrote:
> Should we not mistake the tool for the task?  A hammer and a
> screwdriver are different tools for different tasks, but if a
> screwdriver with a 2" shaft and one with a 4" shaft will both tighten
> and loosen the same screw, who's to say that one is the 'right' one to
> use?  As Greg's pointed out, we teach the tools, but really we're
> trying to teach the practices.

I try my best to emphasize that these are just tools, and that there
are other tools that do the same things (and even try to name them).
I also try to point out, when teaching git for example, what concepts
apply to version control in general, and which ones are specific to
git.  Same for Python vs. other programming languages, etc.

The problem is, depending on the level of the workshop, one can get
too bogged down the details of one particular tool and lose the forest
through the trees.  This has always been a difficult balance to
strike.

But using version control--*any* version control, and I include
Dropbox in that, is vastly better than Thesis_version1.doc,
Thesis_version2.doc, Thesis_version2(revised).doc, etc...  It needs to
be clear, even within the rubric of "each to their own ability", that
that is NOT version control at all (even if it bares some
similarities).  Though now that I've said that I feel like I need to
define my terms here :)

Erik

___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[C. Titus Brown]

On Mon, Feb 29, 2016 at 04:24:42PM +0100, Lex Nederbragt wrote:
> Hi,
> 
> This blog post: 
> http://rajlaboratory.blogspot.no/2016/02/from-reproducibility-to-over.html 
> seems like a perfect example of how many people will think. And I can???t 
> really come up with some compelling evidence to convince this person he is 
> (completely) wrong. 
> 
> On the plus side, they are very aware of the issues and are doing things a 
> lot better than many others...

In response to Arjun's request, I sent these three articles:

http://ivory.idyll.org/blog/replication-i.html
http://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1001745
http://www.nature.com/news/interactive-notebooks-sharing-the-code-1.16261

that provide various responses, source links, and a vision for some sort
of future.

---

Arjun is probably responding to our everpub proposal,

https://github.com/betatim/openscienceprize/blob/master/proposal.md

---

Some robust twitter commentary and vaguely interesting discussion follow
from here:

https://twitter.com/ctitusbrown/status/704295499176620033
https://twitter.com/ctitusbrown/status/704291963692982272

cheers,
--titus

___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[Bennet Fauber]

Should we not mistake the tool for the task?  A hammer and a
screwdriver are different tools for different tasks, but if a
screwdriver with a 2" shaft and one with a 4" shaft will both tighten
and loosen the same screw, who's to say that one is the 'right' one to
use?  As Greg's pointed out, we teach the tools, but really we're
trying to teach the practices.

Greg, how far off 'good enough' practices does that blog post sound?

-- bennet



On Mon, Feb 29, 2016 at 11:24 AM, Greg Wilson
 wrote:
> The biggest thing I'm taking away from this blog post is that lousy
> interfaces and obscure failure modes really are inhibiting adoption of
> better computing practices.  From the post:
>
> if you’re like me, you will screw up at some point, leading to some problem,
> potentially catastrophic, that you will spend hours trying to figure out.
> I’m clearly not alone... “Abort: remote heads forked” anyone? :) At that
> point, we all just call over the one person in lab who knows how to deal
> with all this crap and hope for the best. And look, I’m relatively computer
> savvy, so I can only imagine how intimidating all this is for people who are
> less computer savvy.
>
>
> I heard fewer complaints of this kind, and believe that I saw higher
> adoption rates, when we were teaching Subversion.  I'm not going to
> recommend that we switch back, but I do miss it whenever I have to help
> someone deal with a detached head...
>
> Thanks,
> Greg
>
> --
> Dr Greg Wilson
> Director of Instructor Training
> Software Carpentry Foundation
>
>
> ___
> Discuss mailing list
> Discuss@lists.software-carpentry.org
> http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[David Martin (Staff)]

One reason why I like to use graphical clients such as SourceTree for Git. It 
hides much of the pain. Part of it is training and experience. It takes some 
time to be familiar with appropriate record keeping in a lab book. The same is 
true of using any software. I don't know why there is such an entitlement 
culture of it being easy, or not something that has to be practiced till it is 
got right.

..d

From: Discuss [mailto:discuss-boun...@lists.software-carpentry.org] On Behalf 
Of Greg Wilson
Sent: 29 February 2016 16:25
To: Software Carpentry Discussion 
Subject: Re: [Discuss] RajLab: From reproducibility to over-reproducibility

The biggest thing I'm taking away from this blog post is that lousy interfaces 
and obscure failure modes really are inhibiting adoption of better computing 
practices.  From the post:


if you're like me, you will screw up at some point, leading to some problem, 
potentially catastrophic, that you will spend hours trying to figure out. I'm 
clearly not alone... "Abort: remote heads forked" anyone? :) At that point, we 
all just call over the one person in lab who knows how to deal with all this 
crap and hope for the best. And look, I'm relatively computer savvy, so I can 
only imagine how intimidating all this is for people who are less computer 
savvy.

I heard fewer complaints of this kind, and believe that I saw higher adoption 
rates, when we were teaching Subversion.  I'm not going to recommend that we 
switch back, but I do miss it whenever I have to help someone deal with a 
detached head...

Thanks,
Greg



--

Dr Greg Wilson

Director of Instructor Training

Software Carpentry Foundation

The University of Dundee is a registered Scottish Charity, No: SC015096
___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[Greg Wilson]

The biggest thing I'm taking away from this blog post is that lousy 
interfaces and obscure failure modes really are inhibiting adoption of 
better computing practices.  From the post:


if you’re like me, you will screw up at some point, leading to some 
problem, potentially catastrophic, that you will spend hours trying to 
figure out. I’m clearly not alone... “Abort: remote heads forked” 
anyone? :) At that point, we all just call over the one person in lab 
who knows how to deal with all this crap and hope for the best. And 
look, I’m relatively computer savvy, so I can only imagine how 
intimidating all this is for people who are less computer savvy. 


I heard fewer complaints of this kind, and believe that I saw higher 
adoption rates, when we were teaching Subversion.  I'm not going to 
recommend that we switch back, but I do miss it whenever I have to help 
someone deal with a detached head...


Thanks,
Greg

--
Dr Greg Wilson
Director of Instructor Training
Software Carpentry Foundation

___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[Joanna Jedrzejewska-Szmek]

I practically disagree with every word of that blog post. Wow.

If you use version control properly it can also serve as a lab notebook. If 
there are more people developing a project, it can also tell who to ask for a 
reason for some specific change.

I use latex, because shuffling figures, tables (oh, and making tables look 
better than  my cat's puke) and bibliographies is less work. I had to recently 
write a paper in openoffice and it was an ordeal, mostly because we tracked 
changes. And we could have used latex and version control and it would have 
been so much easier.  Of course I write scripts to generate my figures, 
especially when my boss asks me to move that trace to the other panel and add 
another here, and by the way, let's change all plots to red and black, instead 
of red and green, think of the colorblind!, and yes, this heatmap's pretty 
cool, but I'd like a different color map. It would take me ages, otherwise.

I use version control mostly because I tend to make changes that make scripts 
and models stop working properly. I used to use an equivalent of dropbox and it 
made me cry all the time. 

Joanna

From: Discuss  on behalf of Lex 
Nederbragt 
Sent: Monday, February 29, 2016 9:24 AM
To: Software Carpentry Discussion
Subject: [Discuss] RajLab: From reproducibility to over-reproducibility

Hi,

This blog post: 
http://rajlaboratory.blogspot.no/2016/02/from-reproducibility-to-over.html 
seems like a perfect example of how many people will think. And I can’t really 
come up with some compelling evidence to convince this person he is 
(completely) wrong.

On the plus side, they are very aware of the issues and are doing things a lot 
better than many others...

Lex
___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] RajLab: From reproducibility to over-reproducibility

[Karin Lagesen]

On 29.02.2016 16:24, Lex Nederbragt wrote:

Hi,

This blog post:
http://rajlaboratory.blogspot.no/2016/02/from-reproducibility-to-over.html
seems like a perfect example of how many people will think. And I
can’t really come up with some compelling evidence to convince this
person he is (completely) wrong.

On the plus side, they are very aware of the issues and are doing
things a lot better than many others...


Regarding git, they are forgetting the main reason I use it for my 
stuff, and that's to save myself from the random typo. It is so easy to 
accidentally introduce something that is a logical change which may 
change the conditions under which you are running your analysis. With 
version control, I am able to see all changes, and hence I can actually 
_catch_ stuff like this.


Karin

___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

[Discuss] RajLab: From reproducibility to over-reproducibility

[Lex Nederbragt]

Hi,

This blog post: 
http://rajlaboratory.blogspot.no/2016/02/from-reproducibility-to-over.html 
seems like a perfect example of how many people will think. And I can’t really 
come up with some compelling evidence to convince this person he is 
(completely) wrong. 

On the plus side, they are very aware of the issues and are doing things a lot 
better than many others...

Lex
___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] Versatile PhD

[Joshua Ainsley]

As one of the people who have gone through Insight Data Science, I can
share my experiences to this interesting discussion.

For someone looking to learn development skills, Insight really isn't the
best option. You are expected to be able to code fairly well before the
program. In addition, the work load is intense. I generally worked 10-12
hour days (including weekends) for most of the unpaid program. With my wife
working full time, the only way we made it work with two kids was taking
advantage of some very generous grandparents. That said, the program was
fantastic and it was a great way to transition out of academia. I learned a
lot and had a built in network for my new career as a data scientist.

During Insight, you create a data product which is simply a website that
solves some sort of problem using data. These problems should be something
that people can relate to, which makes explaining what the code does much
easier. I find that scientific code or contributions to FOSS projects are
very useful, but they can definitely sometimes be hard to explain to a
recruiter or interviewer when you are looking for a job outside of
academia. My academic research coding was pretty technical and involved a
lot of neuroscience and molecular biology background before I could even
begin explaining the code. During Insight, I made a website that ranked
addresses in NYC based on how good they were for kids. Both projects
involved a lot of coding, but my Insight project was much easier to talk
about during interviews to both technical and non-technical people.

I think scientific coding and FOSS contributions are great and important,
but good software that solves a more approachable problem that you can tell
a good story around is better when looking for an industry position outside
of your academic field.


On Sun, Feb 28, 2016 at 4:08 PM, Greg Wilson <
gvwil...@software-carpentry.org> wrote:

> Hi Terri; thanks for your mail.
>
> On 2016-02-28 1:46 PM, Terri Yu wrote:
>
>> I was just trying to suggest that for academics who don't have the CS
>> background and software internships, getting involved in the open source
>> community would be a good way to develop software skills and find a
>> community that will support someone with an atypical background.  Also,
>> it's not that easy to jump into an open source project and start making
>> contributions as someone who is new to the community and new to software.
>> Suggesting to someone that they should go make open source contributions to
>> pad their resume, without giving them any guidance or even explaining to
>> them the FOSS ethos -- that seems a tad irresponsible.  That's why I put
>> the emphasis on joining the FOSS community rather than focusing on making
>> contributions.  No open source project maintainer likes getting random pull
>> requests.
>>
>
> I agree, and that's good advice -I hope that Software and Data Carpentry
> are good communities to start with. We may not have much in the way of
> software for people to hone their skills on, but I'd like to think that
> doing pull requests on lessons is a gentle introduction to the workflow.
> I'd be very grateful to hear from anyone whose first encounter with GitHub
> (or other open source machinery) was through DC/SWC, and who then went on
> to contribute to other things.
>
> I'd also be grateful for pointers to research development communities that
> are particularly welcoming to, and have tasks accessible to, newcomers -
> astro.py (http://www.astropy.org/) and rOpenSci (http://ropensci.org/)
> are two that I've heard mentioned before, and there are others on our
> website at http://software-carpentry.org/join/projects/.  (Additions to
> that list would be very welcome.)
>
> Finally, some of our instructors have gone through the 7-week program
> offered by Insight Data Science (http://insightdatascience.com/), which
> is a way to learn development skills without having to load it on top of
> everything else.  Stories from people who've taken part in it or similar
> programs would be very welcome as well.
>
>
> Cheers,
> Greg
>
> --
> Dr Greg Wilson
> Director of Instructor Training
> Software Carpentry Foundation
>
>
___
Discuss mailing list
Discuss@lists.software-carpentry.org
http://lists.software-carpentry.org/mailman/listinfo/discuss_lists.software-carpentry.org

Re: [Discuss] Versatile PhD

[Brendan Smithyman]

Hi All,

I’ve been lurking this conversation for a bit and it’s been really interesting 
to hear different perspectives. I’m a geophysics PhD who is also employable / 
sometimes employed as a programmer (including some closed source work right 
now). This all resonates a lot with me.

I guess that I wonder if some of the recommendations are putting the cart 
before the horse. There are some characteristics that are often associated with 
open-source developers that might be genuinely beneficial. They often (1) are 
community-minded, and have a strong social conscience, and (2) have the 
technical competency, problem-solving skills and persistence to not only use, 
but also contribute to projects that are often not very well-organized or 
approachable. To make meaningful contributions to open source one needs to be 
pretty driven. This is often—but not necessarily—a characteristic of PhD 
graduates.

Personally, I also don’t see open-source contributions as something that should 
necessarily go in a cost/benefit analysis the same way as hourly wages should. 
Sometimes I get involved for xkcd #386  reasons, but 
more often it’s because I love the project and I want to help. I know how hard 
development can be, and these people are giving away for free something from 
which I derive substantial value. I also don’t pirate software for much the 
same reasons, and I think there’s a sort of implicit social contract with open 
source projects to “help as you are able”.

I think some of those same characteristics are what can make good open source 
developers into good employees. However, I think these are effects, not 
necessarily causes. Committing software to an open source project on GitHub is 
often correlated with these behaviours, but that doesn’t mean that these 
behaviours magically manifest themselves just because someone signs up on 
GitHub. Telling PhD candidates to go contribute to open source might be missing 
the point. Telling PhD candidates to find something that they are passionate 
about and lend a hand might be closer to the source of the matter. For many of 
us (or, at least, for me) that’s programming, open source, SC, etc.

Of course, at the end of the day we all have to eat, so maybe emulating those 
behaviours is still a worthwhile goal.

Cheers,
Brendan

—

Brendan Smithyman
Postdoctoral Fellow

Western University, Earth Sciences
Biological & Geological Sciences, Rm. 1045
London, ON, Canada N6A 5B7
c. 778.990.5957

> On Feb 29, 2016, at 7:59 AM, Erik Bray  wrote:
> 
> On Sun, Feb 28, 2016 at 7:46 PM, Terri Yu  > wrote:
>> The panelists at Versatile PhD did say that contributing to open source
>> projects was a good way to develop an online portfolio.  But it sounded more
>> like they were just repeating conventional industry advice.  It didn't sound
>> like any of them had worked on anything open source themselves.  I was just
>> trying to suggest that for academics who don't have the CS background and
>> software internships, getting involved in the open source community would be
>> a good way to develop software skills and find a community that will support
>> someone with an atypical background.  Also, it's not that easy to jump into
>> an open source project and start making contributions as someone who is new
>> to the community and new to software.  Suggesting to someone that they
>> should go make open source contributions to pad their resume, without giving
>> them any guidance or even explaining to them the FOSS ethos -- that seems a
>> tad irresponsible.  That's why I put the emphasis on joining the FOSS
>> community rather than focusing on making contributions.  No open source
>> project maintainer likes getting random pull requests.
> 
> +1 to all of this.  I've been having a rough time with Google Summer
> of Code lately, actually, because of this.  Students are told they
> need to get into FOSS--and they are jumping more and more quickly to
> get into GSoC; getting involved in projects months before GSoC is even
> announced for the year.  A few do a good job at this; they lurk for a
> while, try to learn how some project is developed, and when the time
> is right make good, useful pull requests.
> 
> There are increasingly many, however, who are taking a sort of spray
> approach where they just hammer with random pull requests that take
> more time to review than to actually do the work ourselves, and often
> are not very good.  It depends though.  Even the authors of the "not
> very good" ones are sometimes responsive to guidance and learn
> quickly.  Others require more hand-holding than it's worth.
> 
> But I agree with Terri that it *is* still sound advice to get involved
> with the open source community (and, as it always needs to be
> reminded, this doesn't just mean coding!)  For those who can find the
> time and the will, it's an invaluable way to learn and demonstrate CS