The Rules Versus Outcomes Discussion

________________________________________________________________________________________________________________________________________

Exchanges Between Friends

Written By:  Herb Sorrenson

I believe that what I have written below is relevant to the political arena, where the goal is not truth but power, and people are willing to stake their fellow citizens’ lives on obviously broken models.  When I say to hell with all the reasoning and commitment to outcomes , I'm sticking with the hoi polloi.  I hope you can see that it is maybe not as careless an approach to life and reality as it might seem.. I would like to use Somerset Maugham's The Outstation, as an illustration:  http://maugham.classicauthors.net/outstation/

I hope this helps you to see why global atmospheric models are such a morass.  Do you think it makes sense to spend trillions of dollars based on "models" that some "scientists" with an obvious commitment to some point of view, are urging?  I am always looking and hoping to be wrong.  I find confirmation somewhat dull and I have an insatiable curiosity about aq lot of things.  Finding out that I’m wrong about something lets me into a whole new world – that’s true science.

To me, the soul of science is observation, and specifically the recognition of consistency of observations across time, and by any other careful observers.  But observation is only the beginning.  It isn't just reproducibility, but understandability.  It isn't just observing, but bucketing or categorizing the observations.  If a scientist drops to earth from Mars, he will immediately begin to notice that there are animate and inanimate objects here.  The important thing is what goes into what bucket, and how we name the buckets and define what’s in the buckets.

If, for example, we divide ethics into two buckets (deontology or rules) and (teleology or outcomes), this is just a method that helps us think about ethics in a certain way.  As an empirical scientist, I have lots of experience in bucketing and naming things, but you don’t have to be a scientist to be able to follow this procedure, but I believe this procedure is crucial to understanding.

If you have a lot of phenomena, one huge question is, how many distinct buckets do you need, and how many sub-buckets.  This is done mathematically by segmentation; something which the marketing world is choking on.  Suffice it to say that the more buckets you have the better your data will fit in the buckets, hence the foolish assumption that the better the data fits your bucket model, the more closely you are reflecting reality in your model.  This may not necessarily be so.

Years ago someone told me that the world is divided into two kinds of people: those who divide the world into two kinds of people . . . and those who don't!!!  That's a joke of course, but it is not that far off a useful model that involves 2 buckets only.  I pretty much think that you should be able to divide what you know into simply two buckets.  However, I have also learned that things are not usually that clean.  This is why I usually allow myself a third bucket to accommodate what really doesn't seem to fit well in two.  I don't do this because I don't have extensive experience with mathematical segmentation procedures, but because often they are an excuse for lack of insight and understanding.

The next step of science, is counting the population in each bucket.  That is, if I have two buckets, how many are in bucket A and how many in bucket B, and of course, how many in bucket C.  Just knowing the size of the buckets is a massive advance on sorting stuff into buckets and naming them.  Here is where Lord Kelvin's observation, "If you can't express your knowledge with numbers, it is of a meager and unsatisfactory sort."  Just knowing which bucket is biggest is like, WOW - this must be the most important!  And then bucket B is THE major alternative.  Examining bucket C, which should be minor in size, often brings to stark attention just why the other two buckets are inadequate, but often helps in understanding the next step.

The next step of science we have already begun by counting the buckets and seeing at least one relationship, which is biggest and which is least.  This third step is all about the relationship among the members within the buckets and from bucket to bucket.  If you are sticking with Lord Kelvin, you will actually express those relations by mathematical formulas.  This is important, because mathematical formulas are inherently predictive. That is, the formulas allow you to compute the properties of the buckets, and the inter-relation of the buckets, based on variables you have teased out of your observations.  You are now getting to the point of having models that ostensibly reflect actual understanding of the phenomena you began observing in the first place.

True scientists validate their models by computing how the buckets will behave when conditions not yet observed occur - extrapolating or interpolating from the observations you have already measured - reducing to the numbers you use in creating the model.  Of course, interpolating is inherently more reliable, because we see that in our observations, a number of variables have varied across a range of values.  Just because we have not observed  every value in that range, doesn't mean that we can't reasonably infer what every  value in the range would be, even if we haven't actually observed every single possibility -that's what interpolation means - between the poles - ends of the data.  Of course we can also reasonably infer, to an extent, what the data outside the range will be. However, this extrapolation - outside the poles - is necessarily less reliable than the other. There are ways to mathematically determine how likely you are to make mistakes with all this. 

Interpolation is ordinarily much easier, and more reliable than extrapolation.  Interpolation doesn't even absolutely require mathematical skills, because it covers terrain we are familiar with by direct observation    Interpolation is essentially the province of RULES, which actually are readily observable.  Extrapolation, on the other hand, relies on projecting outcomes not yet observed, but expected because of our mathematical models.

This is the scientific rationale behind something I wrote about deontology and teleology. It also explains why non-scientific people don't even have to think about the rules to understand them.  Most people know that when the days get shorter, it's going to be colder.  As you can see, most of what you really know about the world, yourself and your life, you simply absorb subconsciously, on an observation basis, without articulating the rules which are "common sense" to you.  Actually, you'll get through life better using your native observation process, than you will by attempting to use the cognitive process for everything.

Just to give you a trivial example, I never have to think about which socks I'm going to wear in the morning.  Because I have about 20 pairs of black socks, they are never mismatched, and I don't even decide each day which to wear.  My rule is, black socks, ever and always.  What difficulty do you think I would have in predicting exactly what socks I will be wearing on any given day ten years from now?

I hope this helps you to see why global atmospheric models are such a morass.  Do you think it makes sense to spend trillions of dollars based on "models" that some "scientists" with an obvious commitment to some point of view, are urging?