The Quantum Gravity Problem

From: Kenneth Downs <firstinit.lastname_at_lastnameplusfam.net>
Date: Fri, 15 Oct 2004 13:43:19 -0400
Message-ID: <n72pkc.f8h.ln_at_mercury.downsfam.net>



Here is a little its-been-a-good-week Friday musing.

I wonder if database theory is suffering from a version of what is going on in Physics. In Physics for the past few decades they have had to struggle with the fact that the two fundamental theories of the twentieth century do not play nice together. Relativity describes gravity well, but it is not a quantum theory. Quantum theory is considered the most successful theory in history, but does not describe gravity. Since most physicists believe that the underlying truths are quantum in nature, everyone is searching for a quantum theory of gravity, instead of searching for the relativistic theory of E & M and nuclear forces.

So can we draw any useful analogy here, with perhaps the RDM being quantum and Hierarchies being Relativity? This choice is not arbitrary, it implies that we can find a way to add hierarchies to the RDM before we will get RDM into a hierarchical form.

I'd suggest the analogy is useful more as a motivational tool than an instructional one. For instance, Einstein's disdain for quantum theory gave him the freedom to pursue and develop General Relativity, while the excitement that others felt for quantum led them to ignore gravity completely and pursue quantum. However, sooner or later the us-vs-them camps began to fade and the question became, how do we get these two great but incompatible theories reconciled?

Still it seems the analogy goes further. Physics has progressed from a study of what was obvious and readily observable through to the more subtle and yet more comprehensive truths. Perhaps the RDM and hierarchical models stand as that which we could readily observe (Remember: "People understand tables just fine"), but we need to look for models that unify the two in a way that may not be so readily obvious. Hmmm, what would that be?

So to wrap us this idle musing, is the suggestion that we could put relations into columns an effort to put hierarchies into the RDM, as in:

CREATE TABLE hier_test (

  hier_id char(10), 
  hier_id_parent char(10) ,
  hier_id_kids relation(hier_test as
           (select * from hierarchy where hier_id_parent = hier_id)),
  ...other columns...
)

I could then pull out a hierarchy with this command:

SELECT * from hierarchy where hier_id_parent is null

which would do the same as the WITH RECURSE join modifier.

-- 
Kenneth Downs
Use first initial plus last name at last name plus literal "fam.net" to
email me
Received on Fri Oct 15 2004 - 19:43:19 CEST

Original text of this message