May 12, 2005

The Density of 'Smartness' Theory | What happens when you get too many smart people in the same room at the same time?

DISCLAIMER: This is in no way a slam against anyone directly or indirectly. In fact this is a very general statement in regards to any committee tasked with any project of any type. And its really quite a simple and complimentary theory at that. While I will be using an example from a recent post from Dr. Kay to XSL-List this is nothing that should be seen as anything other than a random sample of a likely response gathered together by somebody who's speaking from obvious experience (and more than likely frustration in some regards) of working in the committee trenches with a perfect random sampling for this theory.

The theory is simply this:

As the density of "smartness" contained in any one committee increases the efficiency of this same committee decreases at an equal and opposite rate.

Take for example the XQuery Working Group...

Really smart people have been working on XQuery. And yet technically after what I believe is nearly 4 years there is still no recommended working draft. (sorry guys and girls... at least you get the benefit of all being really smart, right? Please don't hate me :D)

For further proof on this theory listen to what Dr. Kay had to say in response to a proposal from Dimitre Novatchev that maybe its not too late to get some simple little adjustments to the XSLT 2.0 WD and as such bring things to near perfect completion, at least in regards to a 2.0 release. His response is as follows:

The debate with memo-function would be about whether it actually has any semantics, or is merely a performance hint. Could a conformant processor ignore it? What is the effect on a "creative" function, one that constructs new nodes each time it is called?

The debate about environment variables would be whether there is any meaning to the term that's independent of particular computing platforms, and about whether anything we said in this area would have any more substance than saying nothing. If we made a general statement that allowed one implementation to interpret it as meaning operating system variables and another as meaning Java system properties, then we wouldn't have achieved much.

Need I say more? :)

I think I've got a solution though. How about for version 3.0 of the XSLT Working Draft (I would use XQuery 2.0 as an example but to be allowed the luxury of being assumed a 2.0 working draft will even ever exist you first need to produce a 1.0 recommended draft... I'm just saying... Ok, I'll shutup ;) we get one REAALLLLLLLLYYYYYYYYYY smart person and lock them away for 2 weeks and whatever she or he (hows that for a politically correct/motivated switch of positions in the normal ordered sequence of gender found in the English language? Think I should run for office??? Yeahhh, ok... ;) comes up with at the end of that 2 weeks we simply call "Recommended" and be done with it.

Now I realize that someone who is REAALLLLLLLLYYYYYYYYYY smart is more than likely going to have a similar density of smartness to that of the entire preceding committee (ummm... ok, maybe not... but lets just play along and see what happens :) so maybe another level needs to be added to this theory... something to the effect that every additional member added to a commitee brings a smartness density that increases the overall density in geometric proportion instead of a linear progression... so if their smartness density, on a scale from 1 to 10 is say 8.7, then instead of just adding it to the total we instead multiply it by the existing total to then establish the difficulty to get anything accomplished rating. We can then establish the odds of getting anything accomplished by simply inversing the final difficulty rating. So, take for example someone like Dimitre... lets say he has a 9.7 smartness density rating, the odds of coming out in two weeks with a finished specification would be 9.7 to 1. Now you might think those are horrible odds but if you calculate even two people who have a density rating of 9.1 then the odds of accomplishing anything have now risen to 82.81 to 1, obviously not a crew you would want to take with you to Comdex or CES the next go round...

Actually, come to think of it if you were to put me in a room for two weeks that would give you like 1.5/1 odds of getting a specification in two weeks. Obviously we need to then add a quality of specification factor but with what should be more obvious, with a 1.5 smartness density I'm probably not the right person to be figuring out how to factor this into the overall picture... ;)

So to conclude, I guess all I'm saying is maybe this is worth thinking about in the spirit of progress? Maybe it takes 5 or 6 two week sessions to tilt the odds in the favor of more possible than not but even 12 weeks would be a pretty spectacular achievement.(Watch how I pull the XQuery advocates back into my favor/clutches...) Think of it this way... if we already had XQuery as part our daily lives (we being the other 45% not already using it on a regular basis that is.. come on, I cant go this long without even one little jab at DD. Something tells me maybe its time to let things be though... how sad... oh, well... who's my next XBitch gonna be... Im watching and waiting... it could be you... ;) how much more efficient would we be in our programming? If we had XQuery two years ago how much further along would we be with other things? I'm guessing a pretty decent clip ;) See, something to think about for sure...

Now imagine if we had XSLT 2.0 at that same time... the possibility of progress would be astronomical! :D

Bye! <RUN speed="REAALLLLLLLLYYYYYYYYYY Fast!!!"/>