diff options
author | kra3 <kra3@kra3.in> | 2012-07-01 20:12:57 +0530 |
---|---|---|
committer | kra3 <kra3@kra3.in> | 2012-07-01 20:12:57 +0530 |
commit | 7749530d926180931d3eb415f6f2eb6f06607280 (patch) | |
tree | f8916c3d489593e14697e59162e55a5f9de7162e /evaluatingfosscontributions.tex | |
parent | 5299386e023f2d501dff6fd2efb19569c2cb8e53 (diff) | |
download | logbook-of-an-observer-7749530d926180931d3eb415f6f2eb6f06607280.tar.gz logbook-of-an-observer-7749530d926180931d3eb415f6f2eb6f06607280.tar.xz logbook-of-an-observer-7749530d926180931d3eb415f6f2eb6f06607280.zip |
added post evaluating foss contributions
Diffstat (limited to 'evaluatingfosscontributions.tex')
-rw-r--r-- | evaluatingfosscontributions.tex | 91 |
1 files changed, 91 insertions, 0 deletions
diff --git a/evaluatingfosscontributions.tex b/evaluatingfosscontributions.tex new file mode 100644 index 0000000..cc9496f --- /dev/null +++ b/evaluatingfosscontributions.tex @@ -0,0 +1,91 @@ +\section*{Evaluating FOSS Contributions} +\vskip 2pt + +Counting FOSS contributions towards research grants actually threw open +a new area of investigation as such I guess. Its about evaluation of novelty +factor of contribution to FOSS project. As we know, in any field of research, +evaluation metrics is a big area of investigation and people come up with new +distances and measures every now and then(even we are in the middle of +such an effort for OCR). Normally novelty factor is decided based on where +the related paper is published, how good it is explored and how good is the +theoretical foundation for the papers. The interesting thing is, many FOSS +projects cite papers published(and some regularly publish papers) in main +stream journals to get acknowledged for their novelty and to ensure the novelty of +the algorithms they use. Many times authors note use in FOSS projects to +report on real time usage, easier adaptation etc.(many plugins in GIMP are +developments of European univ. PhDs, famous one is, a resynthesizer plugin). +But the idea of evaluating the novelty factor solely considering the contribution +to the project require a new metric to evaluate it too. The normal factors +required to assert novelty is all present in a collaborative project very +much in line with the normal lab, professor, conferences, peer reviews mechanism. +The only difference and a crucial missing factor will a published(i would +say inaccessible too, a paper costs 5-10 USD depending on the publisher, +conference and journal) covering paper. Peer review of the technique, +its implementation in real time projects and required documentations as +part of the project documentation and various blog and log entries are +available. Peer review of subject experts happen very well in discussions +over IRC and mailing lists(most of which are archived). Different perspectives +from theoretical foundations to practical implementation issues were discussed +in a single go there(depends on the project too). But this varies from project +to project. A project or a contribution which generates a bigger discussion +and criticized and evaluated rigorously should get more points(very similar +to classification of conferences and journals to A +,A,etc.). We can even +have a FOSS project classification depending on how much discussions, scrutiny +and perspectives are evaluated before new features are incorporated into the +existing system. Also another thing we can draw parallel is the criterion used +by conferences and journals for accepting a paper and the peer review system +of the projects. These are some ideas got into my mind, when thought about a +systematic evaluation metric for novelty in FOSS contribution. A metric +and system like this will help to counter so many software patents too i guess. +There is Special Interest Group of ACM for Computer Science Education. +They have a special section on FOSS. I haven't seen the proceedings so +don't know what all they discussed. But it will be good to check these +formal forums and their proceedings to look for prior ideas on the subject. +I don't have access to ACM libraries here. If we can put some time and +thought into this, we can develop a draft and then may be start an open +discussion too. This will help FOSS projects to avoid depending on non +free published items for claiming the novelty factor due for them(since +Santhosh is not interested in publishing, he is not recognized by anyone +in academia of Indian Language Research though his works are very popular). + + +I think what ever i wrote was taking it apriori that acceptance of ideas +to a project is enough for validation. My problem was how do we evaluate +the novelty factor(we know there is novelty factor, but how to scale it). +Then later on turning this novelty factor itself to rate the projects. +Now projects interact with academia in a weird way, we should find a middle +ground, where even academic contributions like submitting a paper to A+ +journal is counted similar to adding the same algorithm with all its detail +to a project with A+ novelty factor. People might not accept it as such, +and at first, there will be double contributions, but with enough campaigning +and ensuring that the evaluation framework is strong enough and thus reliable, +we can make some progress. It will also work as a counter measure to now +monopolistic attitude of IEEE,ACM etc. in case of academic publishing. +Only thing i worry about is the arguments against the review of documentation +(like how implementation of something in one project will ensure its +reimplementation capability in different scenario if the documentation +is not aimed at that). Capability to reimplement and produce results for +a different set of users for a different set of purposes should also carry +weightage(like how much help does this implementation give on doing that). +That usually doesn't come under the aims of project and they don't care, +but the ones who are doing contribution and waiting for it to be counted +towards their Degree or Salary should be aware and do it. Collaborative +publishing can be very well used and example of wikipedia can support the claim. +Acceptance by user community is a validation of novelty. But how the detail +or contribution is accepted may not always be a measure of novelty +(some contributions, very novel, might not trigger much response, some +trivial ones might trigger huge response). So to evaluate novelty and the +original contribution, there should be a mechanism which in turn projects +can use to count or evaluate their innovativeness or novelty factor. +This along with a must do documentation of the contribution in a collaborative +peer reviewed wiki kind of system should ensure freedom of the knowledge +generated out of the process. It is a matter of not just accepting FOSS to +mainstream academic research, but more or less bringing back the idea of +freedom to the academia. Should prepare a draft framework(i don't have +much idea on how to prepare it). Then should try evaluating some recent +contributions to some projects on basis of this framework(we can use SILPA +as one of the sample projects). Then present this to the world as a method +to count novelty in collaborative projects without using the normal way of +stat of publication(all FOSS projects maintained by universities or research +organizations cite their publications to show novelty). +\newpage |