Human Rights in Post-Truth World: The Statistician's Mission | Patrick Ball
Summary
TLDREl discurso destaca la importancia de los datos en la defensa de los derechos humanos y cómo pueden ser engañosos. Se menciona la dificultad de obtener estadísticas precisas, especialmente en situaciones de violencia política. Se presentan casos de cómo el análisis estadístico ha sido crucial para demostrar políticas de violencia y para condenar a líderes responsables, como en el caso del presidente de Chad Hissène Habré. Además, se discute el desafío de lidiar con grandes cantidades de datos, como los registros de muertes en Siria, y cómo la estadística ayuda a comprender la magnitud de lo no observado, esencial para la justicia y la verdad.
Takeaways
- 📊 La importancia de los datos en la defensa de los derechos humanos y cómo estos pueden llevar a decisiones erróneas si no están bien interpretados.
- 🔍 La dificultad de obtener estadísticas precisas en situaciones de violencia política, donde los perpetradores ocultan y distorsionan la información.
- 🌟 La labor de reunir y analizar datos de diferentes fuentes para construir un cuadro más completo de la realidad en casos de violaciones de derechos humanos.
- 📚 El uso de la estadística para demostrar políticas de violencia sistemática, como en el caso del presidente de Chad Hissène Habré.
- 📈 La utilización de la estadística para estimar la magnitud de eventos no observados, fundamental en situaciones donde la violencia es intencionalmente oculta.
- 🔗 La integración de bases de datos y el desafío de lidiar con la duplicación de datos, especialmente en contextos donde no existen identificadores únicos.
- 📖 La aplicación de la estadística para determinar si se cometió genocidio, como en el caso de los mayas ishial en Guatemala.
- 👥 La colaboración entre organizaciones de derechos humanos, la comunidad académica y organismos internacionales para recopilar y analizar datos.
- 🔍 La importancia de la calidad de los datos en la búsqueda de la verdad y la justicia, y cómo la mala calidad de los datos puede llevar a conclusiones erróneas.
- 🌐 El impacto global del trabajo en derechos humanos, abarcando desde organizaciones locales hasta instituciones internacionales como las Naciones Unidas.
Q & A
¿Qué importancia tiene el dato en el ámbito de los derechos humanos según el discurso?
-El dato es crucial en los derechos humanos porque permite conocer la verdad y el panorama general de la situación, lo que es esencial para tomar decisiones informadas y para la responsabilidad política.
¿Por qué puede ser peligroso confiar en un conjunto de datos incompleto?
-Un conjunto de datos incompleto puede llevar a decisiones incorrectas, lo que puede tener consecuencias desastrosas, especialmente en temas de derechos humanos, donde las decisiones basadas en datos incorrectos pueden resultar en abusos y violaciones de derechos.
¿Cuál es la función de las estadísticas en el trabajo de derechos humanos mencionado en el discurso?
-Las estadísticas son fundamentales para entender las políticas de violencia y para demostrar la responsabilidad política en casos de abusos y violaciones de derechos humanos. A pesar de que no son la única evidencia, son cruciales para contextualizar y dar seguimiento a las violaciones.
¿Qué desafío representa el manejo de bases de datos con información sobre muertes en Siria?
-El desafío principal es la deduplicación de registros, ya que carecen de identificadores únicos como los números de seguridad social. Esto complica la integración de datos y la obtención de una imagen precisa de la magnitud de la violencia.
¿Cómo se utilizó la estadística para demostrar la política de violencia del presidente Chad en su juicio?
-Se utilizó la tasa de mortalidad en las prisiones对比 la tasa de mortalidad normal de los adultos en Chad y la mortalidad de prisioneros de guerra estadounidenses en la Segunda Guerra Mundial, para demostrar que las muertes no eran incidentes aislados sino parte de una política de negligencia y abuso sistemático.
¿Qué es la genocidio según el contexto del discurso y cómo se midió en el caso de Guatemala?
-El genocidio se define como el asesinato focalizado de ciertos grupos específicos. En el caso de Guatemala, se midió comparando las tasas de asesinato entre la población indígena y no indígena, encontrando que los indígenas tenían un riesgo relativo de ser asesinados 8 veces mayor.
¿Cómo se abordó el problema de la falta de identificación única en los registros de muertes en Siria?
-Se abordó utilizando aprendizaje automático y técnicas de deduplicación para integrar y limpiar los datos de las bases de datos, a pesar de la dificultad de no contar con identificadores únicos como los números de seguridad social.
¿Qué papel jugó el análisis estadístico en la condena del ex presidente de Guatemala, José Efraín Ríos Montt?
-El análisis estadístico proporcionó evidencia consistente con la afirmación de que se había cometido genocidio, al demostrar que la tasa de asesinato de personas indígenas era significativamente mayor que la de no indígenas, lo que fue citado en la sentencia del juez.
¿Cómo se utilizó la estadística para determinar la responsabilidad en el caso de desaparición de García en Guatemala?
-Mediante un proceso de muestreo aleatorio, se caracterizaron estadísticamente el flujo de documentos dentro de la Policía Nacional, lo que permitió demostrar que los documentos utilizados en el caso tenían el mismo flujo y metadata que el resto de los documentos, lo que indicaba una operación organizada y no simplemente acciones aisladas.
¿Qué mensaje se quiere transmitir con la historia del caso de García en el discurso?
-El mensaje es que la estadística y la investigación en derechos humanos son fundamentales para dar voz a las víctimas y para lograr la justicia, permitiendo que las familias de las víctimas obtengan respuestas y cierren un capítulo doloroso de su historia.
Outlines
📊 Importancia y peligros de los datos en los derechos humanos
El orador expresa su entusiasmo por hablar con una audiencia que comprende la importancia de los datos y cómo estos pueden ser engañosos. Resalta la necesidad de basar decisiones en datos confiables, especialmente en temas de derechos humanos. Describe su experiencia trabajando con diferentes organizaciones, desde pequeños grupos no gubernamentales hasta la Oficina del Alto Comisionado de Derechos Humanos de las Naciones Unidas, enfocándose en revelar patrones de violencia sistemática. Presenta un caso de Chadianos donde se utilizó evidencia estadística para demostrar una política de violencia, y cómo esta fue crucial para condenar al presidente Hissène Habré. Además, menciona el desafío de lidiar con grandes conjuntos de datos, como los registros de muertes en Siria, y la importancia de la integración y limpieza de datos para obtener conclusiones precisas.
🔍 Estadísticas para comprender lo desconocido en derechos humanos
El orador profundiza en el uso de estadísticas para entender fenómenos más amplios que los datos individuales pueden revelar. Utiliza la metáfora de dos habitaciones oscuras y bolas de goma para ilustrar cómo, a través de la frecuencia de choques entre las bolas, se puede inferir la magnitud del espacio. Esta idea se aplica para estimar la magnitud de violencia no observada. Describe el caso de Guatemala, donde se reunieron historias de víctimas y se analizaron para determinar si el ejército había cometido genocidio contra el pueblo ishial. Las estadísticas mostraron una tasa de muerte significativamente mayor para los ishiales, lo que apoyó la acusación de genocidio. El orador enfatiza la importancia de obtener la historia correcta y cómo las estadísticas pueden ayudar a dar voz a las víctimas y a responsabilizar a los perpetradores.
📜 Archivos históricos y justicia en Guatemala
Se narra la historia de los archivos históricos de la Policía Nacional de Guatemala, descubiertos en 2006, que contenían ochenta millones de páginas. El orador describe cómo, junto con sus colegas, diseñaron un proceso de muestreo aleatorio para caracterizar estadísticamente el flujo de documentos a través de la policía. Este trabajo estadístico fue crucial para demostrar que los documentos utilizados en el caso de desaparición de un estudiante y líder sindical, Sr. García, eran representativos y que seguían el mismo flujo y estructura de metadatos que otros documentos del archivo. La evidencia estadística llevó a la convicción de los responsables de la desaparición de García y a la condena de un alto mando de la policía, demostrando la importancia de la estadística en la justicia y en la amplificación de las voces de las víctimas.
Mindmap
Keywords
💡Derechos humanos
💡Datos
💡Engaño estadístico
💡Violencia estructurada
💡Responsabilidad política
💡Deduplicación de datos
💡Estadísticas
💡Genocidio
💡Relación de riesgos
💡Amplificar las voces
Highlights
Emphasizes the importance of data in decision-making and its potential to mislead if not properly understood.
Speaks to an audience knowledgeable about the significance of data in human rights.
Introduces the concept of 'speaking truth to power' and the necessity of accuracy in statistics.
Shares experiences working with various human rights groups, from small NGOs to the UN.
Discusses the challenge of identifying patterns of violence through data analysis.
Presents a case study involving the secret police records in Chad and their significance in a trial.
Highlights the role of statistical evidence in convicting former Chadian President Habra.
Discusses the complexity of deduplicating large datasets, such as records of deaths in Syria.
Explains the importance of understanding both observed and unobserved data in human rights investigations.
Uses a metaphor of dark rooms and rubber balls to illustrate statistical estimation of unobserved data.
Transcripts
I can't tell you how excited I am to
talk to you guys and how appreciative I
am to speak to an audience of people who
understand how important data is and
even more to understand how data can
mislead us so horribly how we can have
an incredibly exciting data set that
doesn't have the piece we need and we
make the wrong decision that's a really
terrible outcome and human rights is
even more terrible we have a saying in
human rights that's already been said
once tonight that we speak truth to
power let me propose to you that if
we're gonna do that it better be true
and with statistics that's actually very
very hard as I mentioned I've been at
this about 25 years and we've worked my
colleagues and I have worked with groups
from as small as tiny little
nongovernmental organizations in Sierra
Leone and in El Salvador to the office
of the High Commissioner for Human
Rights of the United Nations and many
many groups in between universities big
international human rights groups UN
missions and so forth and what all these
groups share is an interest in knowing
the truth in knowing what is the big
picture now if you want the video Oren's
got your back ok and if you want to know
the ecosystem of how those groups are
all going to work together hey trading
healthcare but if you need to know the
numbers if you need to convict someone
for responsibility for a policy of
violence that's where we come in we're
not about helping you understand a
particular event but rather the
conjuncture of events that indicate a
policy of violence that is political
responsibility so I'm gonna walk you
through a couple of cases of that really
quickly so that you can see what I mean
and I hope too that you can share my
concern that bad data can get you a bad
answer and how we can fix that so this
is a pile of trash that a really good
friend of mine and a friend of some of
yours I think is standing in and that
pile of trash that Reed Brody from Human
Rights Watch is standing and turn out to
be the operational records of the secret
police in Chad those are pretty
important records notwithstanding that
they're mixed with chicken bones and rat
feces and so after many many mass
nations in almost 20 years of work
led by Reid and two really important
human rights groups in Chad we managed
to organize those all of those records
and identify the records from that that
were the internal records of the
detention centers of the police and
that's relevant because turned out to
stand trial for his crimes at the
extraordinary African chambers in
Senegal he was that man - hey Bri was
the president of Chad in the 1980s and
I'll focus on just one aspect of the
trial which had hundreds of witnesses
this is the evidence I presented and I
want to emphasize this is not the most
important evidence the convicted
President Habra but this is statistical
evidence is never the most important
evidence but it's crucial evidence
explaining the policy of his violence
that the deaths in his prisons were not
isolated incidents the result of an
occasional abuse of guard rather visited
per day or a policy of sustained neglect
neglect and abuse so severe that that
mortality is 540 times the rate of
normal adult mortality of men in Chad
furthermore it is three to five times
greater than the mortality of u.s.
prisoners of war in Japanese custody in
World War two the reason that's relevant
is because in the Tokyo Tribunal in 1947
the abuse of US prisoners by the
Japanese was ruled a war crime this
statistic was cited three times in the
verdict of the judges who convicted
President Habra and I think one reason
that they cited my evidence so
thoroughly is because it was so well
written in French because Katherine
helped me edit it and I appreciate her
deeply for that assistance but let me
talk to you about the bigger problem
something that we face all the times
that we get a bunch of databases people
send us databases our partners are
accumulating lists of deaths in Syria we
have five partners now we have over
500,000 records of people dead in Syria
that's not 500,000 Syrians that's
500,000
how many are duplicates now this is a
hard problem because a third of all
Syrian men share the name Mohammed okay
so we don't have social security numbers
or national health numbers we have no
unique identifiers so the machine
learning problem to deduplicate and to
create all these databases well I have
to say that's the really really fun part
of my job actually is writing that code
but the point is that if you can
integrate those three databases you're
still left with a question you're living
in the world of the white circles the
data you can see but are you living in
the world on the Left where you can see
a third of the data or are you living in
the world on the right where you see
almost all of it these are very
different worlds and the reason they're
so different is that what we don't know
is different from what we do know now
I'm sorry to lead you down an
epistemological path here but if what we
don't know is systematically different
from what we do know what is it that we
know well it turns out this is exactly
what statistics is for if we knew all
the data it wouldn't be statistics it
would be accounting and it's not
accounting its statistics because we
have to figure out what we don't know
and so this is the most exciting part of
the talk let's do algebra so so here's
the story I'm gonna go through I'm not
going to do this derivation although I
have done it for judges and war crimes
trials many times instead I'm going to
give you a metaphor imagine that you
have two dark rooms they are of
different sizes you'd like to know which
is larger the only tool you have to
determine the size of the rooms is a
handful of little rubber balls and these
balls have a curious property that when
they hit each other they make a noise so
you throw the balls into the first room
bounce bounce bounce bounce and you hear
you gathered the balls you go to the
second room you throw them with equal
force which rooms bigger the second room
is bigger right because the balls have
spread out they have encountered each
other less frequently they've spread out
that intuition is what is formalized in
this very simple derivation and is
entirely unco incidentally on the cover
of our report which I would love to
share with you but the point is that
this allows us to know what we don't
know it allows us to estimate the
magnitude of what is not observed
and that's a crucial part of human
rights work because violence is almost
always unobserved violence is hidden by
the perpetrators on purpose it is
distorted on purpose and so in order for
us to capture the patterns of reality in
order to hold powerful people
accountable we have to figure out what
it is we don't know and that's we didn't
Guatemala throughout the 1990s we
gathered over 19,000 stories from
newspapers from individual victims and
from other NGOs and we accumulated them
into a database I add gated that
information with information from the
truth commission from the Catholic
Church from the national program from
cover compensation from ecumene records
and from many other sources thousands
upon thousands of records to study the
events of one specific area three
counties the counties in which the
shield people live and the question was
did the army commit genocide against the
ischial people now genocide does not
mean mass killing if they killed
everyone that could be the fog of war
genocide means focused killing it means
specifically targeting some people and
not targeting others and so the
calculation that we made is to compare
the killing rates between the two
populations and we found that the ratio
of being killed for an indigenous person
relative to non-indigenous person is 8
to understand that differently if you
lived in one of these three counties in
March of 1982 at the beginning of the
regime of josé efraín ríos Montt and you
were an indigenous person your
probability of having been killed by the
army was 8 times greater than your
non-indigenous neighbor for context
that's called a relative risk in
epidemiology the relative risk of being
bosniak relative to being Serb was 3 to
see versus Hutu was 5/8 so this is
evidence consistent with the claim by
the prosecution that acts of genocide
were committed by the army and this was
cited four times in the verdict that the
judges wrote general Rios Montt was
convicted his conviction stood for 10
days and then reversed on a technicality
they ordered a new trial it has not
happened look it's really important that
we get the story right so I'm gonna tell
you one last story to close this man
it's a student and labor leader in
Guatemala in the early 1980s and one day
in February of 1984 he left his office
and he didn't come home his wife it's
not naive she knew something could have
happened she looked everywhere she went
to police stations she went to embassies
she went to the army bases saying do you
have my husband do you know where he is
what has happened never a few witnesses
said well we think some people in
civilian clothes put him into a car with
no license plates but that's all we've
got and that's where it stood now she
never gave up she founded what became
one of Guatemala's most important human
rights groups the mutual support group
and now she's a very important
politician in Guatemala but in 2006 the
historical archives of the National
Police were discovered eighty million
pages of paper covered in dead insects
bat guano and other filth and mold these
of these had been abandoned after the
transition in Guatemala in 1996 that had
closed the National Police in favor of a
new civilian force and when those
records came to came to came to light
one of the first questions is what the
hell are 80 million pages of paper we're
not gonna read them all no one can read
that much and so my colleagues and I
designed a random sampling process in
which we were able to statistically
characterize the flow of documents among
offices of the National Police now this
has two important features because one
of the things they found for their
meritorious conduct on the day that he
was disappeared on the way mr. Garcia
was disappeared and they were tried and
they confessed saying well we were just
following orders our statistical
evidence first showed that the documents
used in the case were consistent with in
every statistical sense all the rest of
the documents in the case all the rest
of the Q's me in the in the archive but
a second and much more important finding
is that the documents in the case were
had the same flow they had the same
metadata that we now know that were
from other purposes they had the same
metadata structure the same flow through
the organization in which operational
strategies are designed on top plans are
developed at the next layer orders are
delayed are defined at the next layer
passed to the operational layer the
operational layer does their job they
generate reports that go back up the
chain the bureaucracy was fully
functioning and seeing that the judge
said thank you very much you were just
following orders that is not a defense
guilty 40 years goodbye
by the way madam prosecutor go find
their boss at the time of mr. Garcia's
disappearance he was brought into court
and our statistical evidence was used a
second time and I testified against him
he was convicted and sentenced to 40
years in jail this is thank you this is
an important act for justice and this is
an important act I think that I would
like many police forces around the world
to keep in mind right now but most
importantly for me I'd like to take it
back to it being important for people
her grandmother mr. Garcia's father one
of the most fundamental things we do in
human rights as we speak to amplify the
voices of those who've suffered we speak
to amplify the voices of the victims and
among the proudest things I've ever done
in my life is to present statistics that
have allowed the Garcia family to know
when they can speak of their loved ones
in the past tense thank you very much
for your time
you
you
5.0 / 5 (0 votes)