vows for scientists (i.e. statement of values); mashup papers

[gusl]

Here is a list of vows I would encourage academics and scientists to take. Feel free to modify and adapt them to your needs. As I've stated them, they can be arbitrarily burdensome. So: be reasonable! The point of publishing one's values is to enable others to remind you when you fall off the horse. So you can add and retract them as you please.

Vow #1: I will do my best to make my code and data available, so that my experiments are easily reproducible. (optional: I will also publish the seeds used in my random number generators)

Vow #2: In writings meant to expose novel results, I will do my best to credit the sources of all my ideas and results, regardless of where or whether they have been published.

Vow #3: I will do my best to explicitly publish my papers as free content (in the sense of libre), and I will encourage readers to annotate and republish them, as long as the original heading is left clear. (optional: I will publish my LaTeX source)

Vow #4: I will do my best to use standard and accessible language, so that as many people as possible outside my immediate specialty can come to understand my papers with reasonable effort.

Vow #5: I will do my best to structure and formalize my arguments, whether mathematical or not. (optional: using argument mapping; using lambda calculus)

One can distinguish between writing that is meant to introduce novel results and ideas, and writing that is meant to educate readers about previously published ideas. In the latter, I would argue that crediting original results is less important, the purpose of the bibliography being to give useful pointers to the intended reader, rather than to give credit where it is due.

I have not taken these vows, at least not yet.

---

Ok.
I hereby take vows #1 (not including the random numbers clause), #2 (may sometimes need to compromise, due to space limitations) , #3 (when publishers and co-authors don't mind).

#4 may be too burdensome. I would love to take #5, but I wonder if it would be a good use of my time. I love formalizing other people's arguments; perhaps it's best left as an exercise to the reader.

---

I would like to start creating mashup papers. For this purpose, it would be useful if authors took Vow #3, and licensed their papers accordingly. Also, I would like to re-publish papers containing my annotations, especially my lambda-calculus annotations. (e.g. "more precisely, you're minimizing λx.KL(x,p)")

It would be great if I could see previous readers' annotations of a PDF, kinda like Gibeo does for web pages (they even have a friends-network, to help you filter out people who make undesirable annotations).

---

UPDATE: I just discovered Science Commons, but it seems to be more about knowledge management than about about copylefting scientific publications (which they call "Open Access"). The NeuroCommons is a Semantic-Sciencey project.

Comments

[plogon.livejournal.com]

I like the idea of papers becomi ng more interactive/collaborative
Wiki style websites are probablymore convenient than pdfs
This was the first hit on google: http://www.monkeytex.com/

If there was a site like that
with all the citeseer papers on it..

[jcreed.livejournal.com]

I wanted to say that #4 and #5 are sort of at odds with each other, but then I realized they weren't exactly --- they're both good goals that sometimes you might need to achieve with different pieces of text. Like: here's the paragraph that explains what I'm doing at a high level if you don't know the formalism already, and here's the paragraph where I make what I'm saying precise.

[gustavolacerda.livejournal.com]

I find annotations a natural way to provide this parallel structure. Essentially, you have two paragraphs at the same level (it doesn't matter who is the annotator, and who is the annotee)

Conventional text is serial, and thus bad at supporting active reading/browsing.

[mapjunkie.livejournal.com]

I think that with 5, one has to be very careful that it doesn't become "one of a number of already accepted formalisms or structures in my field". I think 5, done really well, allows progress through modularity: "here's the same representational framework, with a new set of statistical interpretations", which is a tactic I've used before to much success.

Having said that, I do like this list for what it says, and although I wouldn't require anyone to agree with it, I would at least expect people be able to discuss it carefully and have their own reasons. I don't like the fact that it doesn't say anything about responsibility for the content of the work, however. I think that's more important than any of the vows listed here.

[gustavolacerda.livejournal.com]

Oh, yeah, responsibility would be good too.
Do you mean responsibility for the truth/accuracy of their results?

I tend to believe in buyer-beware, which would be made easier if authors took vows #1,#4/#5. Vow #3 might enable collective criticism at a new level.

There's always reputation too.

[mapjunkie.livejournal.com]

Responsibility for accuracy and correctness is a good start, although I'd also hope for the responsibility of thinking and writing about consequences of one's work. This would include possibly redirecting one's longer term course if it became clear that even if the path forward was academically feasible, there was no strategic idea being advanced that one was willing to stand behind. At the same time, I realize that doing interesting technical work with clear cut goals and being recognized for them, is rewarding for many, despite whomever's interests those goals might actually serve. I don't have to like it, though.

It does bother me that it is easy to construe these constraints such they are entirely free of any reference to what the content of the science might be, no matter how terrible.

[rdore.livejournal.com]

This post seems pretty self righteous in general, it is commanding other people to do more work so you will have to do less.

At least in mathematics, 4 and 5 strike me as a huge waste of time. For 4, most of the people reading your paper (unless it is something truly monumental) will be specialists in the area you work in. Having to regurgitate all of the details of techniques, arguments, terminology, etc., that your readership already knows well is a waste of your time and theirs. And actually formalizing most mathematical papers would be an enormous undertaking.

Your meaning in 2 is ambiguous. It either means something which is standard, or also strikes me as a waste of time. If you're simplying saying that techniques you're using for elsewhere should be cited, I think this is generally agreed upon ettiquette. If you're demanding something for more intricate and laborious, it again adds a huge burden of work to the author(s).

The first part of 3: Does this just mean putting them on the arXiv, your website, etc.? Or is there something more you're demanding? The republish part of the second clause seems a bit weird. What is the motivation there? I can see lots of reasons why that's not appealing.

For 1: it's the sort of thing that would be nice, but again I think demanding it like your entitled to it is pretty pompous.

[gustavolacerda.livejournal.com]

I haven't taken the vows, so no need to interpret that as self-righteous. And nowhere is it demanding, merely encouraging people to publish their own vows, so that they can be reminded when they fall off the horse.

Yeah, the general idea is to be reader- and community- centric... except for Vow #2, which is to protect people who gave you ideas in non-published form. I know of a severe breach of this: not citing someone because all they had was a blog post.

[gustavolacerda.livejournal.com]

ok, I've taken some vows now.

[rdore.livejournal.com]

Your original post was worded more strongly in terms of how heartily people should commit to them.

Let me ask the following question: Should someone put in 50 hours more work on a publication, when it will only be read by about 50 people, and it will save each of those people about an hour of time? How much do you have to perturb these numbers before the answer changes?

And I've seen in papers or heard in talks enough times statements like "by an unpublished theorem of Smith's", that to the extent that I'd be concerned about #2, it feels like it's already followed. (My earlier objection was that #2 could be read to suggest people needed to explain how they came up with their ideas. Which sometimes is definitely important and useful, but other times not.)

[gustavolacerda.livejournal.com]

I am personally acquainted with someone who, after a famous blog post, exchanged several emails from a certain PhD student, who proceeded to reproduce his experiment exactly, including the same arbitrary parameter settings, and then published it... giving no credit at all to my friend, anywhere! Afterall, who cares about a blog post?

[rdore.livejournal.com]

Ok, well, at least from that description, that sounds like a pretty dishonest move. No arguments there.

[marymcglo.livejournal.com]

See also http://www.acm.org/about/code-of-ethics

#2: I would be interested to see where one draws the line between "Citation," "Acknowledgment," and "Coauthor". It sounds like in the blog post case there should have been an offer for coauthorship, particularly since emails were exchanged.

#3: Publishing LaTeX source runs a higher risk for plagiarism. On one hand, if you're already published in a major conference, who cares if someone steals it and submits it to a fourth-tier conference that nobody reads? On the other hand, I find the practice really irritating. And sometimes it doesn't just go to no-name conferences.

#4: Details typically need to be omitted for page-limit concerns, but citations in preliminaries such as "A tutorial on \topic can be found in \cite{}" are helpful.