COMMENT ⊗   VALID 00013 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	.begin center select 5
C00004 00003	.cb Preface
C00007 00004	.s Introduction
C00012 00005	.s General Policies
C00022 00006	.s Computer Use
C00023 00007	.s Other Support Systems
C00024 00008	.s External Communication
C00025 00009	.s Personnel Policies
C00029 00010	.s Procurement
C00030 00011	.s Accounting
C00031 00012	.s Individual Responsibilities
C00032 00013	.begin ref
C00033 ENDMK
C⊗;
.begin center select 5;
SAIL NAVIGATION
%1Les Earnest
%2Associate Director, SAIL
.end
.skip 6
.cb Abstract

Standard operating procedures to be used in the Stanford
Artificial Intelligence Laboratory and policies for making
additional rules within the Lab are discussed.
It is well known that written rules are bureaucacy's way of preventing
useful work from being done.
This material is presented in the spirit that you should know your enemy.
.skip 4
%2This material reperesents the personal views of the author and has
not been reviewed or approved by anyone, as will be obvious.%1

.cb Preface
Rulemaking is an activity indulged in by all humans, almost from the
moment they come into the world.
In fact, it can be regarded as a definitively human function.
Consider what has happened to earlier anthropocentric views.
"Man as a unique lifeform" was finished off by Darwin and friends.
"Man as a toolmaking animal" has been demolished by a number of furry creatures.
"Man as a cultural being" must consider the many species of birds that
require a proper upbringing in order to sing socially acceptable arias.

"Man as a communicator" must now deal with an increasing number of
gorillas and chimpanzees who know sign language.
Dolphins are so smart they won't talk to us, just to each other.

Given that apes are now learning to use keyboards, it is just a
matter of time until some of them pick up FORTRAN.
Of course, we will still be able to sneer that
"Their programs work alright, but their documentation is atrocious" or
"They monkey around too much".

So how about the slogan "Man is a rulemaking animal"?
This too may not hold up in the long run,
but it may permit the humanists and the bureaucrats to make their
last stand together.

.s Introduction

Many people underrate the importance of regulations.
Good rules, like good algorithms, permit a wide range of problems to
be dealt with routinely and efficiently.
Bad rules waste time and effort.

Rules should serve the objectives of the organization,
but should also deal with certain personal needs as well.
Since SAIL is primarily a research establishment, it is run mainly
from the bottom.
That is, the direction the work takes is controlled largely by
the interests, abilities, and discoveries of the people doing research.
This is not to say that the management is without influence --
allocation decisions must be made whenever there are substantial competing
demands for resources such as money, people, computers, or space -- and
this in turn affects the direction of the work.

The rules discussed here are a small cross-section of those that affect the
Lab.
The set of federal, state, and local laws, contract and grant requirements,
and University regulations under which we operate is truly amazing--and
appalling.  And then there are ethical and moral principles . . .

While "ignorance of the law is no excuse", it is a practical reality.
To further complicate matters, the rules discussed below may conflict
with external regulations in some cases.
The traditional hierarchical view holds that in case of conflict, the
"higher level" regulations prevail.
But things don't always work out that way.

When faced with a dilemma, you are urged to consider first the Negotiation
Rule (see "SAIL General Policies" below).
Bureaucrats should generally be treated as adversaries, but if you
can establish common interests, you may find that they make excellent
co-conspirators in dealing with other agencies.

.ss Appropriate People
In the sections that follow, people who have particular administrative and
service responsibilities in the Lab will be referred to by their
programmer IDs (2 or 3 letters).
Here are some of the more frequently referenced people and their
principal responsibilities at this writing.
.begin turn on "\" tabs 5,20; indent 0,19; preface 0;crbreak; nojust
%3ID \  Name \  Responsibilities%1
CJS \Connie Stanley \Report mailings, Prancing Pony services.
CR \Carol Rosebrough \Petty cash, Pony billing, purchase requisitions.
HVA \Hersche Allen \Personnel admin., building services, financial monitoring.
JBR \Jeff Rubin \System programming.
JMC \John McCarthy \Overall direction, but don't bug him (low tolerance).
LES \Les Earnest \Furniture moving, general management, but don't try him first.
PAT \Patte Wood \Report publishing, travel expense reports, typing coordination.
TED \Ted Panofsky \Computer equipment.
.end
There will also be occasional references to "your supervisor".
If you don't know who that is, you are either lucky or confused.
.s General Policies
One important set of regulations under which SAIL operates is published by
the University administration [1].
This is a 2 kilogram red looseleaf notebook full of personnel, procurement,
and accounting procedures, organization charts, standard forms, and other
light reading.  There are several other published collections of Stanford
regulations as well.

It is interesting to note that in recent years,
University regulations have been doubling by weight every three years or
less.  If this continues for another decade, it will be safe to
ignore them because the administration will be fully occupied
in publishing and will have no time for enforcement.

This section describes how rules should be made, modified, and evaded
within SAIL.  It is formulated partially in response to external trends.

.ss Negotiation Rule
The primary rule is that all rules are negotiable.
"Negotiable" is not the same as "breakable" or even "flexible".
Rather, it means that if the rules don't seem to fit a given situation,
then you should complain to the appropriate person and give them a chance
to deal with it (see "Appropriate Persons" in Section 1).

After-the-fact "negotiation" will be treated as nothing more than a
plea for mercy.

The negotiation is not necessarily one-way.  Rules discussed below are intended
to deal with fairly standard situations and do not necessarily describe
God-given rights.
Rulemakers have a right to improvise in very non-standard situations or
in cases where conventional privileges are being abused.
But in case of apparent misinterpretation, the culprit should be given the
benefit of the doubt . . . the %2first%1 time.

.ss Rule Avoidance
A rule should be formalized only if the expected benefits outweigh the
costs of communicating and enforcing the rule.
While this seems obvious, a remarkable number of
regulations ignore this principle.

In particular, a rule should not be created where common sense is
a better guide.
The fact that some people can screw up anything they
attempt is not compelling evidence that additional rules are needed.

Rules should not be created simply to call attention to the rulemaker's
power, or to entrap and punish his enemies.
Of course, it is OK if a rule catches bad guys as a side effect (heh-heh).

In observation of the Rule Avoidance Rule, the author has avoided
writing this note for twelve years.
It may be too early still.

.ss Insecurity Rule
Resources should not be locked up unless the savings realized
will exceed the cost of their inaccessibility.
This is actually the Rule Avoidance Rule applied to property.
Rephrased: "Don't lock the barn door if horses are being smuggled
%2in%1".

For example, the cost of leaving the Xerox machine unlocked is substantial
and it is also clear that this privilege is abused to some extent.  The
benefits in reduced frustration and increased productivity are harder to
measure but are substantial in a place where people work around the clock.
In the current situation, there is fairly clearly a net gain.
It might be different in a more public place.

Similar principles apply to locking up information, such as computer files.
Many people seem to be excessively protective of their programs and
documents.
While this practice may be justified for this for work in
progress, programs and files associated with completed work should be
made freely available, lest others be forced to repeat the effort.
One of the substantial benefits of the Arpanet is the ease with which
information moves, in many cases through friendly "theft".

There is an exception to the Insecurity Rule in the case of marketable
software (see "Computer Operations" below).  We must also protect
information for which a non-disclosure agreement has been signed as
a condition of access.

.ss Evasion Rule
A person who plans to evade a rule should be sure that the expected gains
outweigh the risks %2and should be prepared to accept the penalties.%1
Evasion is sometimes the correct choice, especially for bad rules, but one
should not do it routinely and without analysis.

Under the Negotiation Rule, you should not attempt to evade
a bad rule until after you have tried to get it changed.
While evasion may appear simpler than getting the rule amended,
it leaves the same mess for the next person.

.ss Appeals
If you think that you are being mistreated by someone or by the System,
there are several alternative directions in which you can appeal for
help.

First, you can go as far as you wish up the administrative chain.  For
most people in the Lab it is as follows:  your supervisor, LES, JMC,
Chairman of Computer Science Department, Dean of School of Humanities and
Sciences, Provost, President, Board of Trustees.  For music people,
the branch leads through the Music Department Chairman then joins the other
path at the Dean of H. & S.

A second approach is to work through the Stanford Ombudsman, who is listed
in the classified section of the Faculty-Staff Telephone Directory.

A third approach is to file a grievence, as described in Guide Memo 22.10 [1].

Or you can hire a lawyer and go at it through the courts . . .

.s Computer Use
.ss Account Authorization
.ss Conditions on Use
.ss Privacy
.ss Reservations and Service Level
.ss Computer Communications
system messages
gripes
dist. lists
documentation
.s Other Support Systems
low overhead
.ss Offices
bicycles and pets
.ss Office Supplies
.ss Building Services
heat, A/C, cleaning
.ss Mail
.ss Telephones
busy WATS--dial 9, #, give charge #
.ss Bulletin Boards
.ss Library
.ss Typing
.ss Copying
.ss Reproduction
.ss Filming and Videotaping
.ss Food
Pony, Yumyum
.ss Recreation
volleyball, sauna, showers
Orgies and other social events
University services
.s External Communication
.ss Internal Documentation
.ss A.I. Memos
.ss Films and Videotapes
.ss Lectures
.ss Conferences
.ss Journal Articles
.ss Media Representatives
.ss Visitors and Tourists

.s Personnel Policies
The University has a very strong caste system that is intermediate in
complexity between those of military organizations and India.
Whereas India has about 3,000 castes and the military get by with just two,
Stanford has four -- faculty, students, exempt staff, and non-exempt staff.
Each caste has distinct privileges and obligations, as discussed below.
Transitions between castes are both difficult and rare.

Some might argue that the top-heavy University administration constitutes
a fifth caste (the Brahmans?), but that refinement is not necessary to
the current discussion.

appointments, fringe benefits, consulting
.ss Faculty and Student Research Assistant Appointments
Faculty and student appointments in the Laboratory usually come about
through informal discovery and negotiation between the individual and
the Principal Investigator (the person who obtained the research funds)
or his representative.

Salaries for faculty members are set by their departments and schools.
The fraction of such appointments that is supported by the Lab is negotiable
and is usually higher in the summer than during the academic year.

Student Research Assistant salaries are the same for everyone in the
Computer Science Department and are currently as shown below.
Student appointments are normally half time during the academic year and
full time during the summer.

.ss Staff Appointments
Another regulatory masterpiece deals with job classifications and
salaries [2].  There is a copy on HVA's shelf.
One measure of the importance of these regulations is that we spend
more time evading them than any others.
If these rules were clearly written, we would be in even deeper trouble
(and so would the rest of the University).

best person for job vs. job for best person
inform on funding source
.ss Visiting Scholars
.ss Working Hours
.ss Honorariums
.ss Benefits
.ss Vacation
.s Procurement
.ss Expendable Materials
.ss Capital Equipment
.s Accounting
.ss Travel Expenses
expense advances
no food on day trips
foreign travel, flag carriers
.ss Petty Cash
.ss Property Accounting
.s Individual Responsibilities
.ss Invention Disclosure
.ss External Publications
page charges, reprints
.ss Professional Activities
conferences
computer OK
no dues
.ss Consulting
.ss Conflict of Interest
disclosure
.ss Privacy of Others

.begin ref;
⊗"GUIDE - Administrative Organization, Policies and Procedures",
Stanford University.
There are two copies at SAIL on the shelves in the offices of HVA and
LES.  Available for browsing, not borrowing.

⊗"Stanford University Job Classification and Pay Plan", Personnel Department.
A copy lives on HVA's shelf.

.end