Peter,

Setting aside issues of bivalent vs. multivalent logic, let me try to
respond to the question of how a Bayesian might reason. It seems to me that
how the Bayesian reasoning would go depends on who is doing the reasoning.
Take three possibilities:

(1) The Judge - a case comes before a Judge in which the defendant is
accused of roller skating in the park on Sunday in violation of city
ordinance. The defendant's attorney argues that roller skates are not motor
vehicles so the ordinance doesn't apply. How should the Judge decide the
case? Suppose there are no precedents and the ordinance was enacted so long
ago that no one is really sure what the original purpose was, and there is
no agreement today about what the purpose of the ordinance should be.

The central concept in Bayesian analysis is _risk_. The Judge's
responsibility is to determine to the best of his or her ability whether or
not the defendant violated the "spirit of the law", and the Judge can't make
exceptions for individual cases. There should be only one consitent
interpretation of the ordinance that applies to everyone. So the Judge has
two problems: decide what interpretation to give the law, and decide whether
the defendant violated the ordinance according to that interpretation.
Assume that the main problem is the former, and that the decision will be
relatively straightforward once the interpretation is made.

The Judge might then enumerate the possible interpretations, much as you
have done, and assign (relative) risk values to each along with a likelihood
of each risk. Then the Judge makes the decision with the least expected
risk.

(2) The Council - an item comes before the city council concerning a
possible amendment to the "no motor vehicles in the park on Sunday"
ordinance, given the many complaints the council has been receiving from
persecuted roller skaters. Assume that the council reasons as a whole
(through discussion) and in the interest of the city (personal risks and
motivations aside).

The council must determine the risk involved in allowing or diallowing
roller skating in the park on Sunday. Possible risks in allowing skating
might be: (i) people are disturbed by the noise (ii) religious people are
upset by all the activity on Sunday (iii) people are injured by roller
skaters bumping into them (the park is more crowded on Sunday). Possible
risks in disallowing skating might be: (i) widespread depression sets in
among the roller skaters (ii) a precedent is set that moves the city further
down the slippery slope fascist conservatism.

Again, risk values might be set for each of the identified risks, and
probabilities assigned to them. Then the council decides to allow or
disallow roller skating according to which decision gives the least expected
risk.

(3) The Cop - a police officer walking through the park is accosted by an
elderly man who complains that people are roller skating, in explicit
disobediance of the ordinance prohibiting the use of motor vehicles in the
park on Sunday. The man expects the officer to tell the eject the skaters
from the park with a warning, or arrest them if they refuse to comply.

The officer knows of the ordinance, but he is not sure whether roller
skating is included in the category of motor vehicles. He must decide
whether or not to oblige the elderly complainant. If he doesn't, he risks
failing to enforce the law. If he does, he risks improperly enforcing the
law. Again, a calculation can be made as to which action has the least
expected risk.

...

One issue I have glossed over, especially in the case of the Judge, is the
distinction between what might be called "rule-Bayesianism" and
"act-Bayesianism", in analogy to corresponding types of utilitarianism. In
evaluating the risk associated with a decision, should one evaluate the risk
resulting from the outcome of this particular decision, or should one
consider the risk involved in everyone making this same decision when faced
with the same circumstances? But I think that gets too far off the point.

I'm not proposing to give a definitive Bayesian treatment here, just to
indicate what one might look like. I definitely _don't_ think that one ever
needs to consider things like "to what degree is are roller skates motor
vehicles?". Roller skates may have various characteristics in common with
motor vehicles, along with various differences. But these don't combine to
give a "net" value of motor-vehicleness. Rather, each characteristic is
examined individually according to how it affects the question at hand. And
what one ends up with is a _risk of classifying roller skates as motor
vehicles_, rather than a value representing the extent to which roller
skates are motor vehicles, or a probability or likelihood that roller skates
are motor vehicles.

Hope this is helpful.

Regards,
J.P.

-----Original Message-----
From: Peter Tillers [mailto:[EMAIL PROTECTED]
Sent: Thursday, January 01, 2004 10:18 PM
To: 'Jason Palmer'; [EMAIL PROTECTED]
Subject: RE: [UAI] causal_ vs_ functional models


The problem I posed was how one might or should talk about the uncertainty
of the following proposition: roller skates are motor vehicles for purposes
of an ordinance that prohibits the operation of a motor vehicle in a park on
Sunday.

Jason Palmer wrote (in part):

>There was presumably a reason the ordinance was enacted, for example it
might have been thought that motor vehicles disturb the peace that a park is
supposed to possess. In that case to determine whether or not roller skates
are motor vehicles, we should determine whether or not skating disturbs the
peace. The difficulty in deciding the roller skate question presumably comes
from the difficulty in deciding whether or not roller skating is a
disturbing activity. <

Alas, Jason Palmer's suggestion makes the business of interpretation of
legal rules such as ordinances and statutes much more simple than it
actually is. Consider: there is often only sparse "legislative history," the
legislative history that exists often points in different directions, it is
often not clear what sorts of phenomena count as evidence of "legislative
intent," some observers believe that it is inappropriate or impossible to
look behind the words of an ordinance or statute for the actual or true
intent of the legislator or legislators, some observers believe that
interpreters should assume that the legislator had reasonable intentions or
would not have wished to produce invidious or harmful or irrational results,
some observers believe that the common law background (in the U.S.) must be
taken into account (at least sometimes) when the words of a statute or
ordinance are construed, some observers simply assert that words in a
statute or ordinance should be given a reasonable construction, etc., etc.,
.... And the validity of all of these and many other approaches to the
interpretation of legal material such as ordinances and statutes is the
subject of much debate ... and the empirical benchmarks (if any) that any
one of these approaches (incoclusively!) suggests or implies are often or
always inconclusive and, worse yet, they are often non-existent, unknown, or
unavailable.

***

To be a bit more concrete:

imagine the possible interpretive complications & issues if any one of the
following conditions holds:


1. Some city council members ("CCMs") on the city council that enacted the
"motor vehicle in the park on Sunday" ordinance said (at one point, in any
event) that the purpose of the ordinance was to keep noise levels down

while some other CCMs said (at one point) that the purpose of the ordinance
was to limit air pollution

while other CCMs said that the purpose was to minimize the risk of physical
harm to users of the park on Sunday

whereas other CCMs entertained (at one point but not at anther point) rather
catholic views of the purposes of the ordinance and said that the ordinance
had all of the purposes mentioned at one time or another by other CCMs.


2. CCM-1 said in city council chambers that the purpose of the ordinance was
to keep noise levels down in the park on Sunday but in a news conference
CCM-1 said that she voted for the ordinance only because the religious
sensibilities of her constituents are offended by motorized traffic in the
park on Sundays.


3. Steel roller skates make much more noise than plastic roller skates but
Honda Accords make less noise than a pair of either type of roller skate.

Everyone agrees (let us suppose) that the operation of Honda Accords in the
park on Sunday violates the ordinance -- unless, of course, a Honda Accord
is operated by a police officer on duty, by an EMT, or, _possibly_, by an
ordinary citizen rushing his pregnant wife to the hospital with the definite
but mistaken belief that birth is imminent.

Municipal garbage trucks can be operated in the park on Sundays. No city
council member mentioned garbage trucks before the ordinance was enacted. No
one knows if any city council members even thought about garbage removal in
parks on Sunday.


4. Police offers have not ticketed users of motorized wheelchairs in the
park on Sundays. The wheelchairs thus far operated in the park on Sundays
have electric motors. It is not clear how police would react to wheelchairs
powered by loud gasoline motors or to wheelchairs that can travel 35 miles
per hour.


5. Police officers have not ticketed drivers of vehicles with "handicapped"
plates who drive slowly through the park on Sunday.


6. Vehicles powered by electric motors emit almost no pollutants during
their operation and some of them are extraordinarily quiet. But some
vehicles powered by electric motors make loud clattering noises, some of
them travel very quickly, and some of them -- for example, city buses -- are
very large and occupy large swaths of space that might otherwise be used by
pedestrians, prams, and pram pushers. (N.B. Is a pram a motor vehicle? Does
it matter whether the pram has a motor? Does it matter whether a pram is
carrying sand rather than a baby?)


7. The state's motor vehicle registration statute does not require the
registration of any vehicle whose maximum motorized speed (on level ground)
is below 20 m.p.h. (There are ambiguities here as well. Let me count the
ways. No, let me not!)


8. The ordinance was enacted after newspaper articles reported that several
small children were hit in the park on Sundays by Yellow Cabs.


9. When asked -- and even when not asked --, some citizens say, "I am
disturbed by roller skaters in parks." Other citizens say, "Roller skaters?
What they do makes no difference to me." Other citizens say, "Roller skaters
make parks so romantic. I love them on any occasion and anywhere." Yet other
citizens say other things about roller skating in response to various kinds
of questions. Some (irascible? malevolent? demented?) citizens have been
seen throwing gravel on the park paths that roller skaters use.


10. People over the age of 40 routinely frown and grimace at the sight or
sound of skateboarders in the park on Sunday.


11. Etc., etc., ...

Sincerely,

   Peter T



------- End of Forwarded Message

Reply via email to