Sunday, 3 November 2013

Oracle and releasing documentation: Do it, or don't!

Something is fundamentally wrong here!

I was watching a great presentation by Carlos Sierra on Adaptive Cursor Sharing, similar to what he presented in OakTable 2013 that I was present in (Yes, I chose Carlos' talk over Jonathan Lewis' on 12c Histograms! What are you gonna do about it?!), just to remind myself of the critical points made about the subject.

If you have seen the presentation (around 01:05:45+) or have been present at his talk, you'd know that there is a certain point in the presentation that EVERYBODY seems to be confused.

The confusion is about how the ranges of row counts in three different buckets -that define the buckets, are calculated. Carlos has explicitly admitted to the issue in his article"Observations on a test case built on seem to indicate that boundaries for these 3 buckets are: less than 1,000; between 1,000 and 1,000,000; and more than 1,000,000."

Carlos is a person who rightly praises OakTable members using a definition that members -rightly, again- attribute to themselves; i.e. Oracle Scientists. But isn't it a bit strange to call people who study the behaviour of a man-made entity, scientists?! Would you call the greatest mechanics you know, (for example) Ford Scientists?!

And that brings me to my point: I've been privileged to be among the members of this group on the Summer of 2013, and I know that "Scientist" is only one of many titles you could choose to describe them with. But the absurdity of the title arises from the unprofessional behaviour of Orcale itself.

I have absolutely no problem with Oracle keeping its documentation secret; after all they've spent millions of man-hours to develop a capable piece of software, but you'd have to make a choice here: either -for whatever reason you choose to do so- you release ALL your documentation, (likes of which you can find here), or you don't do any of it.

The fact that many professional like OakTable members have to do research to find out about the behaviour of a man-made product (as opposed to a natural phenomena/being) whose creator possesses the information about its inner-workings, can only make an independent observer more intent on judging the creator as "yet another corporation" or -in a best case scenario, "not able to provide an association between the size of the company and the quality of its product," while it can be regarded as an actual player in the field of data science.

The ambiguities such as the ones about the ACS could only bolster the position of those who might ask: Is Oracle releasing its documentation -the way it has- merely a PR move?

No comments:

Post a comment