Treat Data As Code
Locked
Treat Data As Code
 
|
Re: Treat Data As Code
|
|
Re: Treat Data As Code
|
|
Re: Treat Data As Code
|
|
I agree as well. We all bring different skills to the table. If we can leverage the lessons learned from open source, great. Librarians, lawyers, educators ... great. They can all help us understand and utilize the opportunities. In terms of software and coders, I think it's the more recent open source lessons we can learn the most from though. I remember some of mechanisms used by open source communities back then. They were good but I think productivity has really been accelerated in recent years with tools like Git and GitHub. There have been several attempts to provide GitHub like services for data (including using GitHub itself for data!) but so far not much success. I think it we will see something like that eventually.
H On Tue, Oct 8, 2013 at 8:46 PM, Michael Lenczner <[hidden email]> wrote:
_______________________________________________ CivicAccess-discuss mailing list [hidden email] http://lists.pwd.ca/mailman/listinfo/civicaccess-discuss |
Re: Treat Data As Code
|
|
thanks folks!
Michael, is what not about relevance, it is more about not understanding. I see different patterns here then I saw in Canada / North America. The ckan'nization of data here is a big, the open data census process and bugs in git'hub that I am learning to follow, and I am observing what I can characterize now, as very different and strong data and technology ideologies, that seem to be serving some communities more than others, least of which seems to be the users. I am also witnessing, I think, a very qualitatively based culture, strongly resisting evidence informed planning and decision making. I am really thankful to the responses and will try to understand the significance of git-hub more. it has shown up on the periphery of my work at Carleton and changed the direction of our software development at the geomatics lab, where, a code focused approach replaced or subsumed cartography, yet it is the cartography which communicates to people, and which has a,function and an aesthetic that is being lost.
on the open data census, I am struggling with a great appreciation for the hacker/bootstrap/agile approach/ethos that is driving it, and at the other end questions about data quality, consistent results across jurisdictions, subjective understanding of questions and varying degrees of knowledge and expertise at the table which make me question reliability. the social construction of open data facts vs the scientific method is messing with my mind.
basically, I am seeing old things in new ways, and trying to figure it out. Thanks again! Herb and Glenn, your framing was really helpful. On Wednesday, October 9, 2013, Herb Lainchbury wrote:
-- _______________________________________________ CivicAccess-discuss mailing list [hidden email] http://lists.pwd.ca/mailman/listinfo/civicaccess-discuss |
Re: Treat Data As Code
|
|
I enjoyed the article, and generally agree with the importance of OD to learn the lessons of open source, and adapt their tools and practices. My own project is an attempt to build a web-based data tool along these lines.
What I have learned is that the solution which eventually becomes the "github" for data sets will have some important differences from github.com because
data and code are different, in the following ways: 1. Data represents facts about the world, so there are issues of provenance, authenticity, and accuracy that are irrelevant in the code world. In many cases,
forking a dataset and then modifying one record of it (analogous to modifying one line of code) makes the data set worthless, even fraudulent. 2. For most computer programs on github, once it is deployed as an application,
the outputs of the program on each user's machine are de-coupled from the code itself. Example, if I make an open source graphics editor, there is no reason to include the graphics file with the code repository. In the case of data, it could
be argued that what people do with the dataset, e.g. analyses, visualizations, apps, should be connected somehow to the source data set. Figuring out how to connect the source data with a variety of outputs is a challenge that most OD tool-builders
have to think very hard about. 3. Interpretation. A given record of data can be interpreted by humans in any number of ways. It could be argued that getting the best interpretation out of data is the top
priority for any data set. Consensus building is a social process, that must be supported by socially-enabled tools. Computer code has an entirely different status. It's unambiguous, and it's created to run a program, not to represent facts about
the world. So the hypothetical "github" of data should accomodate consensus building about data, similar to how a wikipedia article about a controversial topic is realized.
4. As Michael and Glenn mentioned, data sets are used by a diversity of different groups, and people with widely varying technical expertise. Each tool for working with OD probably needs to decide what level of expertise it is designed for, or find a way
to accomodate multiple levels. To broaden the accessibility of OD to civil society groups or other low-tech users, probably the social layer (alluded to in #3) is the most likely place to work on that. By comparison, github is aimed exclusively at techies, and
it's not clear who else would have any interest. Ted Strauss
On Wed, Oct 9, 2013 at 1:01 AM, Tracey P. Lauriault <[hidden email]> wrote: thanks folks! _______________________________________________ CivicAccess-discuss mailing list [hidden email] http://lists.pwd.ca/mailman/listinfo/civicaccess-discuss |
Re: Treat Data As Code
|
|
Ted; It is this last part of your email that I am refering to!github is aimed exclusively at techies, and This is the part i wonder about. On Wed, Oct 9, 2013 at 8:22 PM, Ted Strauss <[hidden email]> wrote:
-- _______________________________________________ CivicAccess-discuss mailing list [hidden email] http://lists.pwd.ca/mailman/listinfo/civicaccess-discuss |
| Free forum by Nabble | Edit this page |