Ten simple rules of credible practice

[Jacob Barhak]

Hi Bill,

Alas, if everyone would follow these rules then there was no need for this committee.

You raise version control as an example of an obvious. Actually it is a good example of a rule not being implemented in many environments.

If for you some of these tools seem deprecated, then you yourself are in a good spot, yet many others are not there yet - I fear the majority. Yet I cannot prove or disprove this general statement - this is my general feeling from many experiences and interactions I had. Other committee members can testify here. Together we should give a good overview.

Even modern guidelines seem to overlook this versioning rule. See my response to recent modeling guidelines:

http://www.linkedin.com/groups/Public-R ... mp_4158822

I got a deprecating reaction, similar to yours, when I tried to send this response to the guidelines to the journal that published those guidelines.

Not all is dark and there are those who use version control and good practices, yet parts of the modeling community do not use those tools. For example a lot of the modeling work is done on spreadsheets. And although online tools such as Google spreadsheets automatically manage version control, there is resistance from these communities to use Google tools or even for opening a free Google account - spreadsheet files are still used to pass data through email. If everyone of those users had a alternative version control system - great. Yet how many actually do? I would love to get user testimonies here in response.

Note that versioning is important for a next stage of evolution in modeling - model sharing. I can tell you from recent experience that we are not at the sharing level yet. The modeling community still has to digest versioning and learn to open up.

I have the habit if keeping all versions of all my work in different enumerated files. This habit was initially sparked by unstable systems and loss of work. I was doing this for years even without version control tools - I still do. Many times my practice seemed peculiar to my surroundings and I had questions asked about it. Some of those were of dismissive nature. Those negative attitudes have to go away. Once this will become obvious to all then this rule can loose relevancy and become deprecated in the next version of our rules. Yes, our rules should will have different versions and it is good we have those on the wiki where versions are automatically kept without effort to the user.

Some of the rules you had in your short list as software management have matured and should be practiced by other disciplines now.

Yet there is always a technological and conceptual gap among practices between disciplines. I can give some recent examples - some looking forward. Hopefully keeping an up to date version of our rules will help close those gaps.

[Jacob Barhak]

You will find this recording loosely related to this discussion thread since Lealem and I are talking here on reproducibility and open source. Here is a link to the podcast:

Lealem Mulugeta, Anthony Scopatz, Jacob Barhak, inSCIght, The Scientific Computing Podcast, Episode 32: Code Reproducibility and Open Source, Posted on 2013/10/19, Online: http://inscight.org/2013/10/19/episode- ... en-source/

Note that this podcast has been recorded prior to the forming of the committee. You will find many clarifications here to our points of view as they appear in the discussions and the rules.

Hopefully you find this fitting this discussion.