Can
computers cope with human races?
by Les Earnest <les at cs.stanford.edu>
© 1989 by the Association for Computing and
Machinery
Published in Communications of the ACM, February
1989. Copying without fee is permitted provided that
the copies are not made or distributed for direct commercial advantage and
credit to the source is given. Abstracting with credit is permitted.
In trying to apply a computer to a task that humans
do, we often discover that it doesn't work. One common problem is that humans are able to deal with fuzzy concepts but computers are not
-- they need precise representations and it is hard to represent a fuzzy
concept in a precise way. However, if we look closer at such tasks, we often
discover that the weakness actually lies not in the
computer but in ourselves -- we didn't understand what we were doing in the
first place.
When faced with a problem of this sort, some people
refuse to recognize the conceptual failure. Instead of seeking a better
representation for the task, they thrash away at making the fuzzy scheme work,
insisting that there is nothing wrong with the conceptual base. I will
illustrate one such problem with a true story. The central theme is the fuzzy
concept of racial and ethnic classification, as used by the U.S. government and
a horde of other bureaucracies. These organizations have been carrying out
elaborate statistical computations and making major policy decisions based on
this concept for many years and are still doing it, with problematical results.
I begin with my first major encounter with this scheme, some 25 years ago.
White faces in new places. In 1963, after
living in Lexington, Massachusetts for 7 years, my family and I moved to the
Washington D.C. area where I helped set up a new office for Mitre
Corporation. After three days of searching, we bought a house then under
construction in a pleasant new suburb called Mantua Hills, near Fairfax,
Virginia. I hadn't noticed it during our search, but it soon became evident
that there were nothing but white faces in that area.
In fact, there were nothing but white faces for miles around.
We expected to find some cultural differences and
did. For example, people drove much less aggressively than in Massachusetts.
The first time that I did a Boston-style bluff at a traffic circle, the other
cars yielded! This took all the fun out of it and I was embarrassed into
driving more conservatively.
When I applied for a Virginia driver's license, I
noticed that the second question on the application, just after ÒName,Ó was
ÒRace.Ó When filling out forms, I have always made it a practice to omit
information that I think is irrelevant. It seemed to me that my race had nothing
to do with driving a car, so I left it blank. When I handed the application to
the clerk along with the fee, he just looked at me, marked ÒWÓ in the blank
field and threw it on a stack. I guess that he had learned that this was the
easiest way to deal with outlanders.
It shortly became apparent that on all forms in
Virginia, the second question was ÒRace,Ó right after ÒName.Ó Someone informed
me that as far as the Commonwealth of Virginia was concerned, there were just
two races: ÒwhiteÓ and Òcolored.Ó Included in ÒcoloredÓ were all
dark-skinned people, including both kinds of Indians. I felt uncomfortable with
this system, knowing that it was part of a scheme of legal discrimination that
still pervaded the laws of many states. For example, it was still illegal in
Virginia for a ÒwhiteÓ and a ÒcoloredÓ person to marry.
Our contractor was a bit slow in finishing the
house. We knew that there was mail headed our way that was probably
accumulating in the post office, so we put up the mailbox even before the house
was finished. The first day we got just two letters -- from the American Civil
Liberties Union and Martin Luther King's organization, SCLC. We figured that
this was the postman's way of letting us know that he was on to us. Sure enough, the next day we got the rest of our accumulated
mail, a large stack.
When our kids brought forms home from school, I
started putting a ÒCÓ after the second question, leaving it to the authorities
to figure out whether that meant ÒColoredÓ or ÒCaucasian.Ó I doubt that this actually confused anyone -- the entire school was lily
white.
Racing clearance. About this time, my boss
and I and another colleague applied for a special security clearance that we
needed. There are certain clearances that can't be named in public -- it was
one of those. I had held an ordinary Top Secret clearance for a number of years and had held the un-namable clearance a
short time before, so I did not anticipate any problems. When I filled out my
personal history form, I noticed that question #5 was ÒRace.Ó In the past I had not paid attention to this question; I just
thoughtlessly wrote ÒCaucasian.Ó Having been sensitized by my new environment,
I reexamined it.
All of my known forebears came from Europe, mostly from Bavaria and Bohemia,
with a few from England, Ireland, and Scotland. A glance in the mirror,
however, indicated that there was Middle Eastern blood in my veins. I have a
Semitic nose and skin that tans so easily that I am often darker than many
people who pass for Black. Did I inherit this from a Hebrew, an Arab, a Gypsy
or perhaps one of the Turks who periodically pillaged Central Europe?
Maybe it was from a Blackfoot Indian that an imaginative aunt thinks was in our
family tree. I will probably never know.
As an arrogant young computer scientist, I believed
that if there is any decision that you can't figure out how to program, the
question is wrong. I couldn't figure out how to program racial classification,
so I concluded that there isn't such a thing. I subsequently reviewed some scientific
literature that confirmed this impression. ÒRaceÓ is, at best, a fuzzy concept
about typical physical characteristics of certain populations. At worst, of
course, it provides a basis for more contemptible conduct than any concept
other than religion. In answer to the race question on the security form, I
decided to put Òmongrel.Ó It would have been slightly less provocative had I
said Òhuman,Ó but I've always enjoyed diddling forms a bit.
Shortly after I handed in the form, I received a
call from a secretary in the security office of the Defense Communications
Agency. She said she had noticed a typographical error in the fifth question
where it said ÒMongrel.Ó She asked if I didn't mean ÒMongol.Ó ÒNo thanks,Ó I
said, ÒI really meant mongrel.Ó She ended the conversation rather quickly.
A few hours later I received a call from the chief
security officer of D.C.A., who I happened to know. ÒHey, Les,Ó he said in a
friendly way, ÒI'd like to talk to you the next time you're over here.Ó I
agreed to see him later that week. When I got there, he tried to talk me out of
answering the race question Òincorrectly.Ó I asked him what he thought was the
right answer. ÒYou know, Caucasian,Ó he replied. ÒOh, you mean someone from the
Caucasus Mountains of the U.S.S.R.?Ó I asked pointedly. ÒNo, you know, white.Ó
ÒActually, I don't know,Ó I said.
We got into a lengthy discussion in which he
informed me that as far as the Defense Department was concerned there were five
races: Caucasian, Negro, Oriental, American Indian, and Pacific Islander. I
asked him how he would classify someone who was, by his definition, 7/8
Caucasian and 1/8 Negro. He said he wasn't sure. I asked how he classified
Egyptians and Ethiopians. He wasn't sure. I said that I wasn't sure either and
that ÒmongrelÓ seemed like the best answer for me. He finally agreed to forward
my form to the security authorities but warned that I was asking for trouble.
A question of stability. I knew what to
expect from a security background investigation: neighbors and former
acquaintances let you know it is going on by asking ÒWhat are they trying to
get you for?Ó and kidding you about what they told the investigators. Within a
week after my application for the new clearance was submitted, it became
apparent that the investigation was already underway and that the agents were
hammering everyone they talked to about my Òmental stability.Ó
Gale, the personnel manager where I worked, was
interviewed quite early and came to me saying ÒMy God! They think you're crazy!
What did you do, rape a polo pony?Ó He also remarked that they had asked him if
he knew me socially and that he had answered ÒYes, we just celebrated Guy
Fawkes Day together.Ó When the investigator wanted to know ÒWhat is Guy Fawkes
Day?Ó he started to explain the gun-powder plot but thought better of it. He
settled for the explanation that ÒIt's a British holiday.Ó
An artist friend named Linda, who lived two houses
away from us, told my wife that she had no trouble answering the investigator's
questions about my stability. She said that she recalled our party the week
before when we had formed two teams to ÒWalk the plank.Ó In this game,
participants take turns walking the length of a 2 x 4 set on edge and drinking
a small amount of beer. Anyone who steps off is eliminated and the team with
the most total crossings after some number of rounds wins. Linda said that she
remembered I was one of the more stable participants.
I was glad that she had not remembered my
instability at an earlier party of hers when I broke my watch and bruised my
ribs in a fall off a skateboard. The embarrassing cause of the accident was
that I had run over the bottom of my own toga!
Meanwhile, the investigation continued full tilt
everywhere I had lived. After about three months it stopped and a short time
later I learned that the clearance had been granted. The other two people whose
investigations were begun at the same time did not receive their clearances
until several months later. In comparing notes, it appeared that the
investigators did the background checks on my colleagues in a much more
leisurely manner. We concluded that my application had received priority
treatment. The investigators had done their best to pin something on me and,
having failed, gave me the clearance.
The lesson was clear: if you want a clearance in a
hurry, put something on your history form that will make the investigators
suspicious but that is not damning. They get so many dull backgrounds to check
that they relish the possibility of actually nailing
someone. By being a bit provocative, you draw priority attention and quicker
service.
After I received the clearance, I expected no
further effects from my provocative answer. As it turned out, there was an
unexpected repercussion a year later and an unexpected victory the year after
that. The repercussion turned out to be an odd side effect of a new computer
application.
Mongrel in a star-chamber. About a year after
I had been granted the supplementary security clearance, I received a certified
letter directing me to report to the Air Force Office of Special Investigations
at Suitland, Maryland very early in the morning on a certain day a month later.
To one whose brain seldom functions before 10 AM, this was a singularly
unappealing trip request.
My wife somehow got me up early on the appointed day
and I drove off in my TR-3 with the top down, even though it was a cold winter
morning. I hoped that the air would stimulate my transition to an awakened
state.
When I arrived and
identified myself, I was immediately ushered into a long narrow room with venetian
blinds on one side turned to block the meager morning light. I was seated on
one side of a table on which there were two goose-neck lamps directed into my
eyes. There was no other light in the room, so I could barely see the three
inquisitors who took positions on the opposite side of the table.
Someone punched on a tape recorder and the trio
began taking turns at poking into my past. They appeared to be trying to
convince me that I was in deep trouble. While the pace and tone of their
questions were clearly aimed at intimidation, they showed surprisingly little
interest in my answers. I managed to stay relaxed, partly because I was not yet
fully awake.
They asked whether I had any association with a
certain professor at San Diego State College, which I had attended for one
year. I recognized his name as being one who was harassed by the House
Un-American Activities Committee during the McCarthy Era. He was an alleged
Communist. I answered that I did not know him but that I might have met him
socially since he and my mother were on the faculty concurrently. They wanted
to know with certainty whether I had taken any classes from him. I said that I
had not.
They next wanted to know how well I knew Linus
Pauling, who they knew was a professor at Caltech when I was a student there. I
acknowledged that he was my freshman chemistry professor and that I had visited
his home once or twice. I did not mention that Pauling's lectures had so
inspired me that I decided to become a chemist. It was not until I took a sophomore
course in physical chemistry that I realized I wasn't cut out for it.
I recalled that Pauling had been regularly harassed
by certain government agencies during the McCarthy Era because of his leftist
ÒpeacenikÓ views. He was barred from overseas travel on occasion and the
harassment continued even after he won the Nobel Prize in chemistry, but seemed
to diminish after the second one, the Peace Prize.
The inquisitors wanted to know how often I got
together with one of my uncles who lived nearby. I acknowledged that we met
occasionally, the last time being a short time earlier when our families dined
together. It sounded as though they thought they had something on him. I knew
him to be a very able person with a distinguished career in public service. He
had been City Manager of Fort Lauderdale and several other cities and had held a number of positions in the State Department. It occurred
to me that they might be planning to nail him for associating with a known
mongrel.
The questions continued in this vein for hours
without a break. I kept waiting for them to bring up a Caltech acquaintance
named Bernon Mitchell, who had lived in the same
student house as me. Mitchell had later taken a position at the National
Security Agency, working in cryptography, then defected to the Soviet Union
with a fellow employee. They were apparently closet gays. In fact, the
inquisitors never mentioned Mitchell. This suggested that they may not have
done a very thorough investigation. A more likely explanation was that Mitchell
and his boyfriend represented a serious failure of the security clearance
establishment -- one that they would rather not talk about.
After about three and a half hours of non-stop
questioning I was beginning to wake up. I was also beginning to get riled over
their seemingly endless fishing expedition. At this point
there was a short pause and a rustling of papers. I sensed that they were
finally getting around to the main course. ÒWe note that on your history form
you claim to be a mongrel,Ó said the man in the middle. ÒWhat makes you think
you are a mongrel?Ó
ÒThat seems to be the best available answer to an
ill-defined question,Ó I responded.
We began an exchange that was very much like my
earlier discussion with the security officer in the Defense Communications
Agency. As before, I asked how they identified various racial groups and how
they classified people who were mixtures of these Òraces.Ó The interrogators
seemed to be taken aback at my asking them questions. They asked why I was
trying to make trouble. I asked them why they would not answer my questions.
When no answers were forthcoming, I finally pointed out that ÒIt is clear that you do not know how to determine the race
of any given person, so it is unreasonable for you to expect me to. I would now
like to know what you want from me.Ó
The interrogators began whispering among themselves.
They had apparently planned to force me to admit my true race and were not
prepared for an alternative outcome. Finally, the man in the center spoke up
saying, ÒAre you willing to sign a sworn statement about your race?Ó
ÒCertainly,Ó I said. They then turned up the lights and called for a
stenographer. She appeared with notebook in hand and I dictated a statement: ÒI
declare that to the best of my knowledge I am a mongrel.Ó ÒDon't you think you
should say more than that,Ó said the chief interrogator. ÒI think that covers
it,Ó I replied. The stenographer shrugged and went off to type the statement.
With the main business out
of the way, things lightened up -- literally. They opened the venetian blinds
to let in some sunlight and offered me a cup of coffee, which I accepted. We
had some friendly conversation, then I signed the typed statement, which was
duly notarized.
Punch line. My former tormentors now seemed
slightly apologetic about the whole affair. I asked them what had prompted this
investigation. After some glances back and forth, one
of them admitted that ÒWe were putting our clearance data base on IBM cards and
found that there was no punch for `mongrel'.Ó I thought about this for a
moment, then asked ÒWhy didn't you add a new punch?Ó ÒWe don't have any
programmers hereÓ was the answer. ÒWe got the program from another agency.Ó
I said, ÒSurely I am not the only person to give a
non-standard answer. With all the civil rights activists now in government
service, some of them must have at least refused to answer the race
question.Ó The atmosphere became noticeably chillier as one of them
answered, with clinched teeth, ÒYou're the only one. The rest of those people
seem to know their race.Ó
I was surprised to learn that nearly everyone
believed in the concept of racial classification. It appeared that even people
who were victims of discrimination acknowledged it as part of their identity.
It was clear that the security people believed I had caused this problem, but I
felt that it was the result of a stupid question and the common programmer's
blunder of creating a categorization that does not include ÒOtherÓ as an
option. They apparently found it impractical to obtain the hour or two of a
programmer's time that would have been needed to fix the computer program, so
they chose instead to work with their standard tools. This led to an
expenditure of hundreds of man-hours of effort in gathering information to try
to intimidate me into changing my answer.
Some important political developments occurred
during the period between my rapid security clearance and the later
inquisition. Civil rights workers from all over the country worked on voter
registration in the South. Three of them disappeared near Philadelphia,
Mississippi on June 22, 1964, and were later found to have been murdered by
local officials. A federal omnibus civil rights act happened to be signed into
law one week later. It banned discrimination in voting, jobs, and public
accommodations and generally removed the last vestiges of legal support for
racial discrimination at the national level.
I never did find out how the security investigators
coped with the fact that I remained a mongrel, but in 1966 I discovered that
something very good had happened: the race question had disappeared from the
security clearance form. In fact, this question disappeared from nearly all
government forms then. I liked to think I helped that change along.
Feeling naked without chains. For a short
period in the mid-'60s, just after the race question disappeared from the
forms, it became socially unacceptable in certain circles to talk about a
person's race, but then an odd thing happened. Those who had been discriminated
against for so long began to think of their racial identity as something to be
proud of and those who wished to end discrimination decided that they needed to
classify people into racial groups in order to be able
to statistically measure compliance with anti-discrimination laws and to
actively right earlier wrongs.
In support of the latter goals, government
bureaucrats invented an ÒethnicÓ classification system that identified the
minorities that they felt might be discriminated against. They never bothered
to define their terms because, like the earlier racists, they had only a hazy
notion of where the boundaries were. Thus, the fuzzy old concept of racial
classification that had been a tool of racists for so long came to be embraced
by their former victims and those who believed that in order
to combat discrimination, you had to classify everyone and compile
statistics.
Beginning in 1965, I helped John McCarthy organize
the Stanford Artificial Intelligence Laboratory and remained as principal
bureaucrat of that organization for 15 years. My first encounter with the new
classification system was in the late '60s, when I received a form that called
for a matrix of numbers to be filled in for the lab, with job levels in one
dimension and ethnic and sex classifications the other way. The classifications
made no more sense to me than those used by the armed services earlier, but
being at a place like Stanford gave me access to expert advice on a wide range
of subjects, so I decided to seek help.
I happened to know Joshua Lederberg, who had
received a Nobel Prize for his work in genetics. This seemed to qualify him as
an expert, so I asked him how he determined the ethnic classes of his staff
members. Josh laughed and said, ÒThat classification is nonsensical. I just let
them choose whatever they would like to be.Ó
This sounded like good advice, but a problem arose
when I applied it. Among the listed ethnic classes were ÒSpanish surnameÓ and
ÒBlack,Ó but one of my secretaries happened to be black and had a Spanish
surname and she felt that she should be listed in both places. I did as she
requested, which meant that the rows and columns of the matrix didn't add up. I
left it that way just to see what would happen. I never heard a word about it.
Ethnic emperor's new clothes. After a few
years of having us fill out ethnic matrices, somebody in the Stanford
administration figured out that they could save a lot of fuss by simply adding
ethnic classifications to their personnel database, so that the computer could
generate all the statistics that the government might want. Furthermore, the
rows and columns of computer-synthesized matrices would always add up correctly
-- truly a conceptual breakthrough! Thus was born the
idea of adding individual ethnic codes to the IBM cards that contained our
personal data. Sound familiar?
The Stanford administration managed to avoid one
mistake that the Defense Department security folks had made: they didn't ask
anyone to classify themselves, thus bypassing troublemakers like me. Everyone
was secretly classified by certain key administrators. These administrators
must have been remarkably well trained, because they did their classifying just
by looking at people -- I never found anyone who was interviewed to determine
their ethnic classification.
I later obtained a copy of the instructions for
determining new ethnic code. Here is the full text.
ETHNIC CODE Required for all employees. The codes
are:
1 = Black, not of Hispanic origin
2 = Asian or Pacific Islander (persons having
origins in any of the original peoples of the Far East, Southeast Asia, the
Indian Subcontinent, or the Pacific Islands.)
3 = American Indian or Alaskan Native.
4 = Hispanic (persons of Mexican, Puerto Rican,
Cuban, Central or South American or other Spanish culture or origin, regardless
of race.)
5 = Non-Minority (persons having origins in any of
the original peoples of Europe, North Africa, or the Middle East.)
I learned that this classification scheme came from
the federal government. I believe that it is still in use throughout the United
States today. Interestingly, there were still five categories, just as the
security folks claimed in the early '60s, though the categories had shifted. I
was glad to see that the problem with Hispanic blacks had been solved, but
there were still some mysteries. According to the
definitions, people from Spain were both Code 4 and code 5. Were we to assume
that 4 takes precedence over 5? Not clear.
Once again, there was no code for ÒmongrelÓ or
ÒundefinedÓ or ÒmixedÓ or Òother.Ó Notice that ÒNon-MinorityÓ is defined as
people from certain specific parts of the world and so it does not function as
the ÒotherÓ category. The designers of this system obviously believed that
everyone belonged to some unique ethnic category, though they didn't describe
exactly how to put them there.
In fact, there were no instructions given on how to
classify people who are mixtures of things. For example, where would I place my
three grandchildren who are half Yup'ik Eskimo and
half whatever we are? They were born in Alaska, so I guess that they qualify as
ÒAlaskan NativeÓ (code 3).
I believe that the failure of this ethnic code to
deal with people of mixed origins is not accidental. It is part of a persistent
conspiracy in the United States to deny that there are such people. Instead of
being accurately identified, they are forced to choose membership in one of the
traditional racial groups. Needless to say, there are
no reliable statistics on the number of such people, but it is certainly very
large and getting larger.
ÒBlackÓ and ÒWhiteÓ are relative. We know
that nearly all of the people in the U.S. who call
themselves ÒblackÓ are genetic mixtures of African and European peoples.
Because our culture is predominantly European, anyone who has detectably
African features is called Òblack,Ó even if they are genetically, say, 7/8
European. If we were a predominantly African country, these same people would
likely be called ÒwhiteÓ because they have detectably European features. In
other words, these racial classifications seem to be made relative to the norm,
which makes them intrinsically subjective and rather unreliable.
I understand that South Africa, which has an African
majority and a dominant European culture, distinguishes between ÒBlack,Ó
ÒWhite,Ó ÒAsian,Ó and ÒColored,Ó the last being the catchall for mongrels. I find
their racial policies abominable, but their racial classification system is
slightly more logical than ours; of course, it too is senseless if you look a
bit deeper.
Sometime after Stanford undertook the secret ethnic
classification project, I saw one of my personnel
forms and discovered that I was code 5. I seriously considered protesting.
After all, some of my ancestors were Huns, a fierce nomadic tribe of Asian
warriors. In the 5th Century A.D., under the leadership of Attila, they swept
over most of Asia and Europe as far West as Gaul, raping, pillaging, and
spreading their genes everywhere. Thus I qualify as
code 2.
On the other hand, based on the best available
anthropological evidence, we can all trace our ancestry back to Africa at an
earlier time. Thus if ÒBlackÓ means anyone whose
ancestors came from sub-Sahara Africa, we all qualify as code 1.
While it was clear that logic would be on my side if
I made a fuss over my ethnic code, I had to consider the potential consequences
of such an action. One possibility was that I would have another run-in with
the ethnic police. I figured that I could handle that. On the other hand, I had
become older and, presumably, wiser and did not wish to expend energy on
hopeless crusades. Everyone else seemed to believe in this absurd scheme, so I
finally decided to live with the shame of being a code 5 ÒNon-minority.Ó
What are we doing here? ÒSo what?Ó you may
say, ÒNobody takes this ethnic stuff seriously anyway.Ó Wrong! Ethnic codes
determine eligibility for certain scholarships and which schools our children
are bussed to. Large employers must exhibit ethnic statistics within certain
ranges in order to avoid charges of discrimination.
Under some Affirmative Action programs approved by the courts, people who are
identified as members of certain ethnic groups are given hiring or promotion
preferences, supposedly to rectify past discrimination. If you own a business,
your ethnic code determines your eligibility for certain kinds of low interest
loans and may give you preferential access to many kinds of government
contracts. And so forth.
I have made fun of racial and ethnic classification
systems not because I disagree with the goals of those who have created these
schemes (even though I do disagree in some cases) but because our society as a whole continues to treat these schemes as if they had
substance, somehow ignoring the fact that they are, and have always been,
nonsense. We continue to build ever more elaborate bureaucratic structures atop
this rotten foundation.
Suppose that we continue building, presumably
because it is for a good cause. Do you think that these structures will be
dismantled once the original purpose has vanished or has been forgotten? From
what I know of the way bureaucracies work, I am sure that the answer is ÒNo.Ó
Fighting racism with ÒbenignÓ racism leads to indefinite racism that will end
only when another process intervenes. Something will eventually intervene, as I
discuss later, but it seems silly to wait that long.
Problems and alternative solutions. Great
progress has been made in reducing racial discrimination in my lifetime. I grew
up in a racist society that had laws prohibiting blacks in the South from
eating in the same restaurants, attending the same schools, using the same
restrooms, drinking from the same fountains, riding in the same part of a bus
or train, sitting in the same part of theaters, or entering the same public
parks as whites. I attended public high school in Louisville, Kentucky, where
we were segregated three ways: white boys, white girls, and colored.
Job discrimination pervaded all of
the U.S., not just the South. Almost the only jobs that were available to
blacks, who were then called ÒnegroÓ or ÒcoloredÓ or something more derogatory,
were as cooks, domestic servants, or bootblacks. Professional sports were
closed to them and few could get jobs in the entertainment industry, though
some made their living as singers, dancers, or musicians.
Earlier, when I lived in San Diego, some
acquaintances who happened to be members of a ÒdangerousÓ race were rounded up
by the government and placed in concentration camps, though I was not aware at
the time of what had happened to them -- my parents told me that they had
Òmoved away.Ó Only recently have we learned of atrocities that happened in
those camps, such as the elderly man who went outside the fence to retrieve a
ball that had been thrown there by his grandson and was machine-gunned to death
by a guard.
We can take some pride in the fact that our
government did not systematically murder people in our concentration camps, but
then neither did the Germans in the beginning. I shudder to think what might
have happened if our war in the Pacific had gone badly. U.S. Government
propaganda had already convinced most citizens that Japanese were a sub-human
species who bred themselves to serve the Japanese national purpose by flying
kamikaze missions or otherwise serving as cannon fodder and that one of our
national goals was to ÒKill Japs!Ó
Some of these prejudicial feelings seem to have
resurfaced recently in response to Japanese economic success. Nevertheless, we
seem to have made substantial progress in eradicating prejudice and
discrimination during the last 40 years. Still, it is clear that
this problem is not solved.
My vision of the future is a colorblind society. I
know that we will not reach it in my lifetime because old habits die hard. In
fact, we will never entirely escape this problem because of a peculiarity of
human nature: wherever there are distinguishable groups of people, tribal
instincts can take root and turn it into an Òus versus themÓ situation. This
phenomenon is not just racially based Ð if you have
any doubt, attend a high school basketball game and observe the fans on both
sides.
There was a short time in the mid-'60s when I
thought we were headed in the right direction. Essentially all
of the legal underpinnings of racial discrimination had been knocked out
of the law books. Then certain anti-discrimination forces mobilized, claiming
that it was insufficient to simply outlaw discrimination. They believed that it
was also necessary to measure it statistically by classifying everyone.
I disagree. I think that all the important issues
can be addressed without resorting to the absurd exercise of trying to assign
ethnic codes to everyone. For example, one way to homogenize school composition
is to assign students in some area to schools in that area in accordance with a
random number generator. This would achieve racial balance without resorting to
classification foolishness.
The most direct way to fight discrimination in
housing or employment is to send a well-qualified minority applicant to a
suspect and, if the applicant is refused, send a less well qualified majority
applicant. If that person is accepted, repeat the experiment once or twice to
be sure, then nail them! This scheme has been tested and it works.
Back to basics. Please understand that I do
not claim that the concept of race is totally meaningless. People in certain
parts of the world do bear physical similarities to one another and racial
terms are sometimes useful as labels for those similarities, provided
that we do not pretend that these terms have well-defined meanings.
Ashley Montagu [1] and others have pointed out that most popular racial
concepts are, in fact, myths. What is truly nonsensical is to turn the fuzzy
concept of race around and try to classify all individuals as being members of
some particular race.
Also, I do not claim that there necessarily have to be logical inconsistencies in racial and ethnic
codes, though all that I have seen to date do exhibit such properties. Simply
including a Òmixed bloodÓ or ÒmongrelÓ category would solve a lot of problems,
but for some reason that idea does not seem to occur to most people who design
these codes.
Though many people clearly believe that racial and
ethnic classifications are somehow linked to science, I observe that their
relationship to genetics is a lot like astrology's link to astronomy. The
analogy is imperfect, however; very few government officials are willing to
publicly admit that they plan their lives around astrology (though some
apparently do), but nearly all of them publicly plan their programs around
ethnic classifications. Indeed, the government pours millions of dollars each
year into reaffirmation of this belief and requires that private industry join
in the massive delusion.
The widespread delusion about racial and ethnic
classification has not been confined to the nonscientific world, unfortunately.
As Lancelot Hogben remarked 56 years ago [2]:
ÒGeneticists believe that anthropologists have decided what a race is.
Ethnologists assume that their classifications embody principles which genetic
science has proved to be correct. Politicians believe that their prejudices
have the sanction of genetic laws and the findings of physical anthropology to
sustain them.Ó
While there often are visible differences between
people from areas that are widely separated, these differences are very small
compared with the physical similarities of all humans [3]. Genetic studies
indicate that all modern humans evolved from a single population about 100,000
years ago, possibly less than half that time. On the evolutionary time scale,
this is a very short period.
As these people spread out, the genetic compositions
of widely separated groups moved apart to some degree, but there have always
been gradations and mixtures of characteristics in between them. In other
words, there are no clear boundaries between groups. Europeans and certain
Africans are genetically somewhat closer to each other than to their cousins in
the Far East, but there are all kinds of variations in between.
Some of the geographical variations in human
characteristics appear to be environmental adaptations. For example, having a
relatively large amount of melanin in the skin not only makes it dark but also
protects against intense solar radiation, which reduces the frequency of skin
cancer and other skin disorders. On the other hand, having very fair skin
facilitates the absorption of sunlight and the production of vitamin D, which
inhibits rickets and other diseases. Thus, the Nordic complexion is well suited
to life at the higher latitudes where there is less sunlight available. Some
people's skin has the ability to bleach out if it is
not exposed to much sun or to become very dark if it is. This adaptation would
have been useful to nomadic groups that periodically migrated from one zone to
the other.
Other interpretations of visible differences in
people are more speculative. The Semitic nose, for example, has a larger moist
interior that could be advantageous when breathing hot, arid air. The
relatively small noses and other features of Far Eastern people could have been
an adaptation to an extremely cold environment. Perhaps their ancestors evolved
in one of the nastier parts of Siberia.
Some of the visible differences in widely separated
groups are certainly not environmental adaptations but are the result of
Ògenetic drift.Ó For example, a small population with a chance collection of
genetic characteristics may happen to grow into a very large population that
then further propagates these characteristics.
How do we describe people? In addition to the
use of ethnic classifications as an alleged tool for fighting discrimination,
ethnic terms are also used for visual identification by police and the media,
though with different conventions. While police reports are usually specific,
such as ÒMale Caucasian, 5 feet 10,Ó newspapers usually report only departures
from the racial norm. In other words, if the person's race is not mentioned, it
is presumably Òwhite.Ó Almost never does either group identify people as
being of mixed blood, even though a very large portion of the people they deal
with actually are.
While there are many people in the U.S. who visually
match certain racial stereotypes, there are also a lot who do not and the
proportion in the latter category is increasing year-by-year. People who don't
fit any racial stereotypes can cause serious problems for those who try to
identify them in racial terms. For example, I know a lady with very dark skin
and bright orange-red hair. How do you suppose she should be classified? I saw
some comely ladies in Amsterdam a few years ago with pale skin and bright green
hair. To which racial group would you say they belong?
Obviously, the police and others who use racial
terms for identification have no clearer understanding of these terms than do
the bureaucrats. For the most part, they seem to use racial terms as synonyms
for skin color. Thus, it is reasonable to ask why they don't use a more precise
vocabulary that already exists: artists' terms for skin color. The reason seems
to be rooted in history; racial terms were adopted for individual
identification at a time when our entire society was racist, including police
and newspaper reporters. Old habits die hard.
Fortunately, for those who long for a way out of the
classification morass, help is on the way!
Fuzzy concept made precise. Given that human
genetic codes are now in the process of being unraveled, it will soon be
possible to accurately classify people into racial groups. All we need do is
measure the distance between a given individual's genetic code and those of
various racial standards and assign that person to the nearest racial group.
There are several schemes under development for measuring genetic distance [4].
One of the more straightforward methods uses an adaptation of Hamming distance,
as follows.
The basic genetic material, DNA, is composed of
strings of nucleotides, each of which consists of one of four bases: adenine,
thymine, guanine, or cytosine. In computer terms, then, the genetic information
can be represented as a string of bytes, each having one of four values. The
distance between two such codes can be taken as simply the number of
corresponding positions in which the two codes differ.
The word ÒcorrespondingÓ is a bit tricky.
Considerable analysis may be needed to align elements that have the same or
similar functions, given that there may be gaps or additions in one code or the
other [5].
The principal function of DNA is to control the
fabrication of proteins using various kinds of RNA as intermediaries, but only
about 1% of the DNA sequences represent protein formulas. Some of the remaining
material apparently represents control structures, which determine when the
various fabrication events happen, but there also appears to be a lot of ÒgarbageÓ
-- codes that do nothing. About 10% of the material is so-called satellite DNA,
which consists of the same sequence concatenated over and over up to thousands
of times. Another 20% consists of the same sequence repeated in many scattered
places. The functions of these repeated sequences, if any, are unknown.
If you consider the above scheme to be a byte-level
metric, there is also a word-level distance measure that can be used. Genetic
mechanisms interpret the code in three-byte words, called Òcodons'Ó by
geneticists, each of which specifies one of 20 amino acids or a ÒstopÓ code.
Thus, another plausible distance measure is simply a count of the number of
corresponding codons in which there is a difference in the amino acid
specified. Note that each word could designate one of 43 = 64 actions, but
a number of codes yield the same action, leaving just
21 possibilities. The designer apparently left no undefined codes for future
use.
Unfortunately, it will not be practical to routinely
analyze individual genetic codes in their entirety for the foreseeable future
-- the current practical limit is to determine a few thousand nucleotide
sequences at a time, whereas human DNA contains about 3 billion. There are
plans to map the entire sequence for some individual, but that will be a major
undertaking that could not be done routinely for many people any time soon.
Nevertheless, it is practical to measure genetic
distance based on comparisons of selected genes. In order to
do this, we will need to select the set to be used and define corresponding
code sequences for various racial standards, such as a standard Black, standard
White, standard Chinese, etc. Of course, some people will want to carry this a
step further and define a standard Texan or even a standard South
Philadelphian.
The process of choosing which races will be
ÒstandardÓ will no doubt generate a lot of heat, but suppose that we manage to
do that. Then everyone can be classified as being a member of the racial group
whose standard is closest to their own. With either of the two distance
measures discussed above or one of the others that are under development [4],
it will be possible to assign everyone unambiguously to a racial group except
for the rare individuals who happen to be exactly equidistant from the two
closest standards. To deal with these rare exceptions, we can probably devise
tie-breaking rules.
While this marvel of future science will yield exact
and unambiguous racial classifications, such a scheme clearly will not be
useful for visual identification. In fact, I can't think of anything that it
would be good for, other than to provide a formalized basis for bigotry. For the purpose of identification, the individual's genetic
codes will be far more useful than any racial classification derived from them.
Urge to merge. Whether or
not we solve the problem of racial discrimination through education,
political action, and law enforcement, human biology will apparently solve it
for us in the long run. If there are no more major influxes of foreign
populations into the United States, distinguishable racial groups will
essentially disappear in this country within a few centuries because of the
Òurge to merge.Ó In other words, the U.S. seems destined to become a nation of
mongrels.
This blending process has almost certainly happened
in other parts of the world in the past, producing many of the ÒhomogeneousÓ
modern human populations seen today. In the United States, future white
supremacists and black power advocates must inevitably reconcile themselves to
being members of shrinking minorities. I predict that as the mongrels become
dominant, new rallying cries will be heard; perhaps, ÒBeige is beautiful.Ó
Before that happens, though, they must learn to identify themselves as members
of the new breed rather than as members of traditional races. The U.S.
government currently denies them that right.
Conclusions. I have argued that all
historical and present racial and ethnic classification systems for individuals
are nonsensical and so are the laws, court decisions, computer applications,
and bureaucratic superstructures that have been built on top of them.
The attempt to use computers to assist in racial
classification tasks has helped sharpen the issues because computers can't deal
with fuzzy concepts. If you try to define an ethnic code that is logically
complete, consistent, and determinable for every person using current
technology, you find that you can't.
There seems to be a silent conspiracy to deny the
existence of mixed racial groups in the United States. Most such people have
acquiesced to this conspiracy and don't even think of themselves in those
terms. Instead, they go along with the idea that they are members of one of the
races recognized by the government. In fact, they often identify with a
traditional race that represents only a small fraction of their genetic
heritage!
It appears that rigorous racial classification will
soon be possible through advances in genetics and the development of
computerized racial stereotypes, though the usefulness of such schemes is
suspect. Thus one answer to the title question, ÒCan
computers cope with human races?Ó is: ÒNot yet, but soon -- but who cares?Ó
Even if we fail to deal effectively with the racial
issues, it appears that the Òurge to mergeÓ will eventually settle this problem
for our descendants. It is encouraging to know that nature will handle it if we
screw up.
Many people have quietly resisted the persistent
nonsense of racial classification by either refusing to answer such questions
or by listing themselves as Òhuman.Ó Over the last 25 years, I have
consistently answered Òmongrel.Ó In order to turn back
the classifiers, I believe that it will be necessary to form an identifiable
movement with a distinctive title. As long as we're choosing a name, why not
identify with the long term winners? I propose USA
Mongrels.
I invite others to join in self-declassification,
with the hope and expectation that the bureaucrats and politicians will
eventually be forced to quit playing with this issue and will recognize that
the United States of America is a nation of egalitarian mongrels. I believe
that we will all be better off. So will the computers.
Finally, computer scientists who encounter problems
in adapting human concepts to computer use should not assume that the source of
such problems necessarily lies in the limitations of computers. There is a real
possibility that the concept itself may be flawed.
Acknowledgement. Thanks to B. Edwin Blaisdel of the Linus Pauling Institute for guidance on the
topic of genetic distance measures and to Peter Denning, CACM Editor in Chief,
for suggesting that I turn some of my electronic ÒflamesÓ into an article and
for a number of helpful suggestions on the manuscript.
References
[1] Ashley
Montagu, Man's Most Dangerous Myth, New York, Oxford University Press,
1974.
[2] Lancelot Hogben, ÒThe Concept of RaceÓ in his Genetic Principles
in Medicine and Social Science, New York, Knopf, 1932.
[3] Joshua
Lederberg, ÒThe Genetics of Human Nature,Ó Social Research, Vol. 40, pp.
375-406, 1973.
[4] Joseph Felsenstein, ÒNumerical methods for inferring evolutionary
trees,Ó Quart. Rev. Biology, Vol. 57, pp. 379-404, 1982.
[5] M.S. Waterman, ÒGeneral methods of inferring sequence comparison,Ó Bull. Math. Biology, Vol. 46, pp. 473-500, 1984.