next week I’d love to watch you all
next week I’d love to watch you all [Music]
likely to engage you this is a simple
likely to engage you this is a simple example of this tactic all predicted
example of this tactic all predicted
example of this tactic all predicted completely rejected policing is defined
completely rejected policing is defined
completely rejected policing is defined as the application technique to identify
as the application technique to identify
as the application technique to identify the targets for police intervention or
the targets for police intervention or
the targets for police intervention or solved infer from past projects where
solved infer from past projects where
solved infer from past projects where geographically punished apart or who as
geographically punished apart or who as
geographically punished apart or who as in which specific people are likely to
in which specific people are likely to
in which specific people are likely to get drunk police departments vary in the
get drunk police departments vary in the
get drunk police departments vary in the way that they incorporate the different
way that they incorporate the different
way that they incorporate the different pieces into their work knitting as with
pieces into their work knitting as with
pieces into their work knitting as with established to the target location all
established to the target location all
established to the target location all told us to be departed to predicted
told us to be departed to predicted
told us to be departed to predicted pieces
pieces
pieces logistic policing has bipartisan appeal
logistic policing has bipartisan appeal
logistic policing has bipartisan appeal both conservatives and liberals support
both conservatives and liberals support
both conservatives and liberals support entities predictive policing software
entities predictive policing software
entities predictive policing software drivers
drivers
drivers or project operations like Red Bull and
or project operations like Red Bull and
or project operations like Red Bull and July and Palantir they claim that their
July and Palantir they claim that their
July and Palantir they claim that their products open in four digit on the data
products open in four digit on the data
products open in four digit on the data on the subject is cloudy and web
on the subject is cloudy and web
on the subject is cloudy and web complaints are just rejected policing
complaints are just rejected policing
complaints are just rejected policing model is operated by static algorithm
model is operated by static algorithm
model is operated by static algorithm basically software country paddocks
basically software country paddocks
basically software country paddocks whose on the dataset in this case the
whose on the dataset in this case the
whose on the dataset in this case the patterns there is a record of crime
patterns there is a record of crime
patterns there is a record of crime without a citizen recording it or log
without a citizen recording it or log
without a citizen recording it or log discovered if you can then rock and no
discovered if you can then rock and no
discovered if you can then rock and no one notices
one notices
one notices did it happen according to the official
did it happen according to the official
did it happen according to the official time data the answer is no this is
time data the answer is no this is
time data the answer is no this is opinion wipe don’t have some amount of
opinion wipe don’t have some amount of
opinion wipe don’t have some amount of interaction with police as blacks and
interaction with police as blacks and
interaction with police as blacks and also a little bit same kind of
also a little bit same kind of
also a little bit same kind of interaction with police is class this
interaction with police is class this
interaction with police is class this difference is encoded in the conduct for
difference is encoded in the conduct for
difference is encoded in the conduct for instance they should live time data
instance they should live time data
instance they should live time data suggests that blacks use illegal drugs
suggests that blacks use illegal drugs
suggests that blacks use illegal drugs with whites certainly they are arrested
with whites certainly they are arrested
with whites certainly they are arrested and drug-related crimes I have a rape
and drug-related crimes I have a rape
and drug-related crimes I have a rape advice but more studies haven’t
advice but more studies haven’t
advice but more studies haven’t validated like white and blacks use
validated like white and blacks use
validated like white and blacks use illegal drugs and really
but without when is contradiction one
but without when is contradiction one answer is greater these presents and
answer is greater these presents and
answer is greater these presents and black neighborhoods this is one example
black neighborhoods this is one example
black neighborhoods this is one example of bias of the data but there are many
of bias of the data but there are many
of bias of the data but there are many others some are so embedded we don’t
others some are so embedded we don’t
others some are so embedded we don’t recognize the problem is that there is a
recognize the problem is that there is a
recognize the problem is that there is a lot of social historic and economic
lot of social historic and economic
lot of social historic and economic package built into the country
package built into the country
package built into the country this is particularly true of original
this is particularly true of original
this is particularly true of original rice primary that is compromised by a
rice primary that is compromised by a
rice primary that is compromised by a legacy of institutional racism the
legacy of institutional racism the
legacy of institutional racism the software abusers play in don’t use
software abusers play in don’t use
software abusers play in don’t use variables that describe protected
variables that describe protected
variables that describe protected classes by the grapes but it shows up
classes by the grapes but it shows up
classes by the grapes but it shows up anyway and proxy variables that they do
anyway and proxy variables that they do
anyway and proxy variables that they do use zip code is an example how think of
use zip code is an example how think of
use zip code is an example how think of bread lining the practice of preventing
bread lining the practice of preventing
bread lining the practice of preventing black homeowners are buying property
black homeowners are buying property
black homeowners are buying property outside designated areas by the 1900’s
outside designated areas by the 1900’s
outside designated areas by the 1900’s you can see the legacy of redlining in
you can see the legacy of redlining in
you can see the legacy of redlining in the zip code beta today redlining isn’t
the zip code beta today redlining isn’t
the zip code beta today redlining isn’t equal but I act importance with the data
equal but I act importance with the data
equal but I act importance with the data a lot of produced to perpetuate the in
a lot of produced to perpetuate the in
a lot of produced to perpetuate the in apology the Fair Housing Act is in place
apology the Fair Housing Act is in place
apology the Fair Housing Act is in place to stop this is example but there are
to stop this is example but there are
to stop this is example but there are other variables income low educational
other variables income low educational
other variables income low educational attainment higher east encounters all of
attainment higher east encounters all of
attainment higher east encounters all of these are also closely correlated with
these are also closely correlated with
these are also closely correlated with race that isn’t perfect but there are
race that isn’t perfect but there are
race that isn’t perfect but there are bridges crime data is notoriously there
bridges crime data is notoriously there
bridges crime data is notoriously there are so many variables that contribute
are so many variables that contribute
are so many variables that contribute high and then I’m social health and
high and then I’m social health and
high and then I’m social health and political law enforcement and many other
political law enforcement and many other
political law enforcement and many other vectors all there are ethical concerns
vectors all there are ethical concerns
vectors all there are ethical concerns too about how the law was right there
too about how the law was right there
too about how the law was right there at this point they easily played root
at this point they easily played root
at this point they easily played root level characteristics more than the
level characteristics more than the
level characteristics more than the initial Pharisees bring with the
initial Pharisees bring with the
initial Pharisees bring with the unemployment right there and you’re
unemployment right there and you’re
unemployment right there and you’re feeling these criminals are all the way
feeling these criminals are all the way
feeling these criminals are all the way that morality that your own history
that morality that your own history
that morality that your own history should any person’s interaction with
should any person’s interaction with
should any person’s interaction with police be determined based on others
police be determined based on others
police be determined based on others behavior is that last data sets indicate
behavior is that last data sets indicate
behavior is that last data sets indicate an accuracy even given the tools to
an accuracy even given the tools to
an accuracy even given the tools to correct inaccuracies in big data sets
correct inaccuracies in big data sets
correct inaccuracies in big data sets the work to do it with the expensive
the work to do it with the expensive
the work to do it with the expensive labor intensive altogether this evidence
labor intensive altogether this evidence
labor intensive altogether this evidence suggests that more than predicting crime
suggests that more than predicting crime
suggests that more than predicting crime predictive policing algorithms predict
predictive policing algorithms predict
predictive policing algorithms predict policing that is they predict where
policing that is they predict where
policing that is they predict where these will be deployed and I’m using
these will be deployed and I’m using
these will be deployed and I’m using them these departments riffs reproducing
them these departments riffs reproducing
them these departments riffs reproducing any increasing box in the criminal
any increasing box in the criminal
any increasing box in the criminal justice system
now the corporation’s that push
now the corporation’s that push predictive policing are that visual
predictive policing are that visual
predictive policing are that visual artistic are all we have to go on
artistic are all we have to go on
artistic are all we have to go on they’re better than nothing right the
they’re better than nothing right the
they’re better than nothing right the research that they produce in-house
research that they produce in-house
research that they produce in-house shows that difficulties in always lead
shows that difficulties in always lead
shows that difficulties in always lead to a modest reduction in reported crime
to a modest reduction in reported crime
to a modest reduction in reported crime but their stones haven’t been
but their stones haven’t been
but their stones haven’t been independently verified so very bad there
independently verified so very bad there
independently verified so very bad there are some things that we are sure about
are some things that we are sure about
are some things that we are sure about my colleagues have a human rights data
my colleagues have a human rights data
my colleagues have a human rights data analysis groups leasing the project
analysis groups leasing the project
analysis groups leasing the project William Isaac in Christian love did
William Isaac in Christian love did
William Isaac in Christian love did instead the results of glitch were
instead the results of glitch were
instead the results of glitch were published in the journal called
published in the journal called
published in the journal called significance but they found was alarming
significance but they found was alarming
significance but they found was alarming they perform a simulation on a synthetic
they perform a simulation on a synthetic
they perform a simulation on a synthetic population that approximates
population that approximates
population that approximates population in Oakland California this
population in Oakland California this
population in Oakland California this innovation is a visual representation of
innovation is a visual representation of
innovation is a visual representation of the results the black dots represent an
the results the black dots represent an
the results the black dots represent an incident of drug product reported in the
incident of drug product reported in the
incident of drug product reported in the official condom
official condom
official condom the red squares are where the predictive
the red squares are where the predictive
the red squares are where the predictive policing model this one used by Red Bull
policing model this one used by Red Bull
policing model this one used by Red Bull Ernest police took up the bottle since
Ernest police took up the bottle since
Ernest police took up the bottle since police officers to areas where their
police officers to areas where their
police officers to areas where their crime data shows or drops these
crime data shows or drops these
crime data shows or drops these communities have large black Latino and
communities have large black Latino and
communities have large black Latino and low-income populations
low-income populations
low-income populations please deploy to these areas where they
please deploy to these areas where they
please deploy to these areas where they find more time but it is because they
find more time but it is because they
find more time but it is because they are there this new data feeds back into
are there this new data feeds back into
are there this new data feeds back into the model which identifies a higher
the model which identifies a higher
the model which identifies a higher incidence of drug crime in that area
incidence of drug crime in that area
incidence of drug crime in that area relative to the neighboring areas and
relative to the neighboring areas and
relative to the neighboring areas and again suggests that it’s wide enough at
again suggests that it’s wide enough at
again suggests that it’s wide enough at least to that area on and on and on it
least to that area on and on and on it
least to that area on and on and on it goes and infects until you have this an
goes and infects until you have this an
goes and infects until you have this an image of estimated drug use versus an
image of estimated drug use versus an
image of estimated drug use versus an image a reported drug drop again police
image a reported drug drop again police
image a reported drug drop again police discover each time in an area does not
discover each time in an area does not
discover each time in an area does not mean that there is more illegal drug use
mean that there is more illegal drug use
mean that there is more illegal drug use there it only means that illegal drug
there it only means that illegal drug
there it only means that illegal drug crime was back so what looked like an
crime was back so what looked like an
crime was back so what looked like an unbiased objective prediction now looks
unbiased objective prediction now looks
unbiased objective prediction now looks like a self-fulfilling prophecy
in policing in an area is
in policing in an area is disproportionate to the amount of real
disproportionate to the amount of real
disproportionate to the amount of real criminal activity that is discriminatory
criminal activity that is discriminatory
criminal activity that is discriminatory policing and these models help reduce
policing and these models help reduce
policing and these models help reduce that this is especially concerning given
that this is especially concerning given
that this is especially concerning given that incidents of police violence
that incidents of police violence
that incidents of police violence committed against black and brown people
committed against black and brown people
committed against black and brown people over the years across the country police
over the years across the country police
over the years across the country police disproportionately arrest flex those
disproportionately arrest flex those
disproportionately arrest flex those arrests lead to more arrests stream
arrests lead to more arrests stream
arrests lead to more arrests stream stops important sweets in the courts it
stops important sweets in the courts it
stops important sweets in the courts it leads to harsher sentencing and higher
leads to harsher sentencing and higher
leads to harsher sentencing and higher admission rates given the situation it
admission rates given the situation it
admission rates given the situation it is not surprising that many communities
is not surprising that many communities
is not surprising that many communities would elect to avoid any interaction
would elect to avoid any interaction
would elect to avoid any interaction with law enforcement . the community’s
with law enforcement . the community’s
with law enforcement . the community’s needs are competing with these
needs are competing with these
needs are competing with these predictions and this straight least
predictions and this straight least
predictions and this straight least sybilla’s relationships even for other
sybilla’s relationships even for other
sybilla’s relationships even for other studies have examined the impact of
studies have examined the impact of
studies have examined the impact of different predictive models on targeted
different predictive models on targeted
different predictive models on targeted communities and the results are also
communities and the results are also
communities and the results are also versus predictive models have been used
versus predictive models have been used
versus predictive models have been used importance to predict recidivism like
importance to predict recidivism like
importance to predict recidivism like repeat Olympic or probation or parole
repeat Olympic or probation or parole
repeat Olympic or probation or parole averages and pull of the fact that one
averages and pull of the fact that one
averages and pull of the fact that one of these models was almost twice as
of these models was almost twice as
of these models was almost twice as likely to incorrectly predict a black
likely to incorrectly predict a black
likely to incorrectly predict a black defendant as high-risk reselected these
defendant as high-risk reselected these
defendant as high-risk reselected these models are either used in states across
models are either used in states across
models are either used in states across the country this is very concerning and
the country this is very concerning and
the country this is very concerning and top- in the criminal justice field agree
as these models become normalized in the
as these models become normalized in the criminal justice system they have
criminal justice system they have
criminal justice system they have significant impact their whole
significant impact their whole
significant impact their whole communities and the
communities and the
communities and the well justice system as a whole the data
well justice system as a whole the data
well justice system as a whole the data violence thus the models are biased and
violence thus the models are biased and
violence thus the models are biased and to even give it a perfect and complete
to even give it a perfect and complete
to even give it a perfect and complete record of the past statistics aren’t a
record of the past statistics aren’t a
record of the past statistics aren’t a crystal ball the past isn’t a perfect
crystal ball the past isn’t a perfect
crystal ball the past isn’t a perfect guide to the future
guide to the future
guide to the future the producers of these problems and the
the producers of these problems and the
the producers of these problems and the police departments that you still would
police departments that you still would
police departments that you still would do well to remember that and some of
do well to remember that and some of
do well to remember that and some of these departments are on several
these departments are on several
these departments are on several California police departments have
California police departments have
California police departments have decided against using these moments Oh
decided against using these moments Oh
decided against using these moments Oh Richmond Milpitas imperfect it means
Richmond Milpitas imperfect it means
Richmond Milpitas imperfect it means that as though I really that gives you
that as though I really that gives you
that as though I really that gives you use of analytics and data driven
use of analytics and data driven
use of analytics and data driven approaches in law enforcement and crime
approaches in law enforcement and crime
approaches in law enforcement and crime prevention period but that’s not the
prevention period but that’s not the
prevention period but that’s not the case instead what I’m saying is a we
case instead what I’m saying is a we
case instead what I’m saying is a we need a marketing plan consider the
need a marketing plan consider the
need a marketing plan consider the education police must assess their role
education police must assess their role
education police must assess their role within the community the various tools
within the community the various tools
within the community the various tools of their disposal and their values when
of their disposal and their values when
of their disposal and their values when thinking about the utility of new
thinking about the utility of new
thinking about the utility of new technologies in their work however
technologies in their work however
technologies in their work however complex and detailed the analysis
complex and detailed the analysis
complex and detailed the analysis algorithmic data mining techniques are
algorithmic data mining techniques are
algorithmic data mining techniques are ill-equipped to direct police on who
ill-equipped to direct police on who
ill-equipped to direct police on who what when where why and how to peace
what when where why and how to peace
what when where why and how to peace rather than relying on these tools for
rather than relying on these tools for
rather than relying on these tools for boundaries we should turn elsewhere they
boundaries we should turn elsewhere they
boundaries we should turn elsewhere they should look to the policy leaders the
should look to the policy leaders the
should look to the policy leaders the stakeholders the researchers and
stakeholders the researchers and
stakeholders the researchers and communities above all those are the
communities above all those are the
communities above all those are the people they serve and also they should
people they serve and also they should
people they serve and also they should be market appearing about how they
be market appearing about how they
be market appearing about how they ideally before implementing them
ideally before implementing them
ideally before implementing them inviting a litigation and evaluation we
inviting a litigation and evaluation we
inviting a litigation and evaluation we don’t get to know if these models work
don’t get to know if these models work
don’t get to know if these models work the way we want them to and frankly we
the way we want them to and frankly we
the way we want them to and frankly we can’t business because the companies
can’t business because the companies
can’t business because the companies that produce these models use
that produce these models use
that produce these models use proprietary techniques we don’t have the
proprietary techniques we don’t have the
proprietary techniques we don’t have the data or the algorithms they use to
data or the algorithms they use to
data or the algorithms they use to reduce their analysis if the data is
reduce their analysis if the data is
reduce their analysis if the data is kept either eyespot dermaluxe
kept either eyespot dermaluxe
kept either eyespot dermaluxe researchers civil rights groups and even
researchers civil rights groups and even
researchers civil rights groups and even the police departments lives Ellis can’t
the police departments lives Ellis can’t
the police departments lives Ellis can’t understand how they function
understand how they function
understand how they function despite all of this police departments
despite all of this police departments
despite all of this police departments are still buying music ridiculous models
are still buying music ridiculous models
are still buying music ridiculous models but I have even more collaboration with
but I have even more collaboration with
but I have even more collaboration with researchers to develop a more accurate
researchers to develop a more accurate
researchers to develop a more accurate picture of crime greater attention to
picture of crime greater attention to
picture of crime greater attention to police community relations and more
police community relations and more
police community relations and more equitable policing practices more leads
equitable policing practices more leads
equitable policing practices more leads Department will come to understand how
Department will come to understand how
Department will come to understand how data can use tools to be the stakes are
data can use tools to be the stakes are
data can use tools to be the stakes are very high in criminal justice
very high in criminal justice
very high in criminal justice it is important to get answers that are
it is important to get answers that are
it is important to get answers that are accurate fair and just
next week I’d love to watch you all [Music]
likely to engage you this is a simple
likely to engage you this is a simple example of this tactic all predicted
example of this tactic all predicted
example of this tactic all predicted completely rejected policing is defined
completely rejected policing is defined
completely rejected policing is defined as the application technique to identify
as the application technique to identify
as the application technique to identify the targets for police intervention or
the targets for police intervention or
the targets for police intervention or solved infer from past projects where
solved infer from past projects where
solved infer from past projects where geographically punished apart or who as
geographically punished apart or who as
geographically punished apart or who as in which specific people are likely to
in which specific people are likely to
in which specific people are likely to get drunk police departments vary in the
get drunk police departments vary in the
get drunk police departments vary in the way that they incorporate the different
way that they incorporate the different
way that they incorporate the different pieces into their work knitting as with
pieces into their work knitting as with
pieces into their work knitting as with established to the target location all
established to the target location all
established to the target location all told us to be departed to predicted
told us to be departed to predicted
told us to be departed to predicted pieces
pieces
pieces logistic policing has bipartisan appeal
logistic policing has bipartisan appeal
logistic policing has bipartisan appeal both conservatives and liberals support
both conservatives and liberals support
both conservatives and liberals support entities predictive policing software
entities predictive policing software
entities predictive policing software drivers
drivers
drivers or project operations like Red Bull and
or project operations like Red Bull and
or project operations like Red Bull and July and Palantir they claim that their
July and Palantir they claim that their
July and Palantir they claim that their products open in four digit on the data
products open in four digit on the data
products open in four digit on the data on the subject is cloudy and web
on the subject is cloudy and web
on the subject is cloudy and web complaints are just rejected policing
complaints are just rejected policing
complaints are just rejected policing model is operated by static algorithm
model is operated by static algorithm
model is operated by static algorithm basically software country paddocks
basically software country paddocks
basically software country paddocks whose on the dataset in this case the
whose on the dataset in this case the
whose on the dataset in this case the patterns there is a record of crime
patterns there is a record of crime
patterns there is a record of crime without a citizen recording it or log
without a citizen recording it or log
without a citizen recording it or log discovered if you can then rock and no
discovered if you can then rock and no
discovered if you can then rock and no one notices
one notices
one notices did it happen according to the official
did it happen according to the official
did it happen according to the official time data the answer is no this is
time data the answer is no this is
time data the answer is no this is opinion wipe don’t have some amount of
opinion wipe don’t have some amount of
opinion wipe don’t have some amount of interaction with police as blacks and
interaction with police as blacks and
interaction with police as blacks and also a little bit same kind of
also a little bit same kind of
also a little bit same kind of interaction with police is class this
interaction with police is class this
interaction with police is class this difference is encoded in the conduct for
difference is encoded in the conduct for
difference is encoded in the conduct for instance they should live time data
instance they should live time data
instance they should live time data suggests that blacks use illegal drugs
suggests that blacks use illegal drugs
suggests that blacks use illegal drugs with whites certainly they are arrested
with whites certainly they are arrested
with whites certainly they are arrested and drug-related crimes I have a rape
and drug-related crimes I have a rape
and drug-related crimes I have a rape advice but more studies haven’t
advice but more studies haven’t
advice but more studies haven’t validated like white and blacks use
validated like white and blacks use
validated like white and blacks use illegal drugs and really
but without when is contradiction one
but without when is contradiction one answer is greater these presents and
answer is greater these presents and
answer is greater these presents and black neighborhoods this is one example
black neighborhoods this is one example
black neighborhoods this is one example of bias of the data but there are many
of bias of the data but there are many
of bias of the data but there are many others some are so embedded we don’t
others some are so embedded we don’t
others some are so embedded we don’t recognize the problem is that there is a
recognize the problem is that there is a
recognize the problem is that there is a lot of social historic and economic
lot of social historic and economic
lot of social historic and economic package built into the country
package built into the country
package built into the country this is particularly true of original
this is particularly true of original
this is particularly true of original rice primary that is compromised by a
rice primary that is compromised by a
rice primary that is compromised by a legacy of institutional racism the
legacy of institutional racism the
legacy of institutional racism the software abusers play in don’t use
software abusers play in don’t use
software abusers play in don’t use variables that describe protected
variables that describe protected
variables that describe protected classes by the grapes but it shows up
classes by the grapes but it shows up
classes by the grapes but it shows up anyway and proxy variables that they do
anyway and proxy variables that they do
anyway and proxy variables that they do use zip code is an example how think of
use zip code is an example how think of
use zip code is an example how think of bread lining the practice of preventing
bread lining the practice of preventing
bread lining the practice of preventing black homeowners are buying property
black homeowners are buying property
black homeowners are buying property outside designated areas by the 1900’s
outside designated areas by the 1900’s
outside designated areas by the 1900’s you can see the legacy of redlining in
you can see the legacy of redlining in
you can see the legacy of redlining in the zip code beta today redlining isn’t
the zip code beta today redlining isn’t
the zip code beta today redlining isn’t equal but I act importance with the data
equal but I act importance with the data
equal but I act importance with the data a lot of produced to perpetuate the in
a lot of produced to perpetuate the in
a lot of produced to perpetuate the in apology the Fair Housing Act is in place
apology the Fair Housing Act is in place
apology the Fair Housing Act is in place to stop this is example but there are
to stop this is example but there are
to stop this is example but there are other variables income low educational
other variables income low educational
other variables income low educational attainment higher east encounters all of
attainment higher east encounters all of
attainment higher east encounters all of these are also closely correlated with
these are also closely correlated with
these are also closely correlated with race that isn’t perfect but there are
race that isn’t perfect but there are
race that isn’t perfect but there are bridges crime data is notoriously there
bridges crime data is notoriously there
bridges crime data is notoriously there are so many variables that contribute
are so many variables that contribute
are so many variables that contribute high and then I’m social health and
high and then I’m social health and
high and then I’m social health and political law enforcement and many other
political law enforcement and many other
political law enforcement and many other vectors all there are ethical concerns
vectors all there are ethical concerns
vectors all there are ethical concerns too about how the law was right there
too about how the law was right there
too about how the law was right there at this point they easily played root
at this point they easily played root
at this point they easily played root level characteristics more than the
level characteristics more than the
level characteristics more than the initial Pharisees bring with the
initial Pharisees bring with the
initial Pharisees bring with the unemployment right there and you’re
unemployment right there and you’re
unemployment right there and you’re feeling these criminals are all the way
feeling these criminals are all the way
feeling these criminals are all the way that morality that your own history
that morality that your own history
that morality that your own history should any person’s interaction with
should any person’s interaction with
should any person’s interaction with police be determined based on others
police be determined based on others
police be determined based on others behavior is that last data sets indicate
behavior is that last data sets indicate
behavior is that last data sets indicate an accuracy even given the tools to
an accuracy even given the tools to
an accuracy even given the tools to correct inaccuracies in big data sets
correct inaccuracies in big data sets
correct inaccuracies in big data sets the work to do it with the expensive
the work to do it with the expensive
the work to do it with the expensive labor intensive altogether this evidence
labor intensive altogether this evidence
labor intensive altogether this evidence suggests that more than predicting crime
suggests that more than predicting crime
suggests that more than predicting crime predictive policing algorithms predict
predictive policing algorithms predict
predictive policing algorithms predict policing that is they predict where
policing that is they predict where
policing that is they predict where these will be deployed and I’m using
these will be deployed and I’m using
these will be deployed and I’m using them these departments riffs reproducing
them these departments riffs reproducing
them these departments riffs reproducing any increasing box in the criminal
any increasing box in the criminal
any increasing box in the criminal justice system
now the corporation’s that push
now the corporation’s that push predictive policing are that visual
predictive policing are that visual
predictive policing are that visual artistic are all we have to go on
artistic are all we have to go on
artistic are all we have to go on they’re better than nothing right the
they’re better than nothing right the
they’re better than nothing right the research that they produce in-house
research that they produce in-house
research that they produce in-house shows that difficulties in always lead
shows that difficulties in always lead
shows that difficulties in always lead to a modest reduction in reported crime
to a modest reduction in reported crime
to a modest reduction in reported crime but their stones haven’t been
but their stones haven’t been
but their stones haven’t been independently verified so very bad there
independently verified so very bad there
independently verified so very bad there are some things that we are sure about
are some things that we are sure about
are some things that we are sure about my colleagues have a human rights data
my colleagues have a human rights data
my colleagues have a human rights data analysis groups leasing the project
analysis groups leasing the project
analysis groups leasing the project William Isaac in Christian love did
William Isaac in Christian love did
William Isaac in Christian love did instead the results of glitch were
instead the results of glitch were
instead the results of glitch were published in the journal called
published in the journal called
published in the journal called significance but they found was alarming
significance but they found was alarming
significance but they found was alarming they perform a simulation on a synthetic
they perform a simulation on a synthetic
they perform a simulation on a synthetic population that approximates
population that approximates
population that approximates population in Oakland California this
population in Oakland California this
population in Oakland California this innovation is a visual representation of
innovation is a visual representation of
innovation is a visual representation of the results the black dots represent an
the results the black dots represent an
the results the black dots represent an incident of drug product reported in the
incident of drug product reported in the
incident of drug product reported in the official condom
official condom
official condom the red squares are where the predictive
the red squares are where the predictive
the red squares are where the predictive policing model this one used by Red Bull
policing model this one used by Red Bull
policing model this one used by Red Bull Ernest police took up the bottle since
Ernest police took up the bottle since
Ernest police took up the bottle since police officers to areas where their
police officers to areas where their
police officers to areas where their crime data shows or drops these
crime data shows or drops these
crime data shows or drops these communities have large black Latino and
communities have large black Latino and
communities have large black Latino and low-income populations
low-income populations
low-income populations please deploy to these areas where they
please deploy to these areas where they
please deploy to these areas where they find more time but it is because they
find more time but it is because they
find more time but it is because they are there this new data feeds back into
are there this new data feeds back into
are there this new data feeds back into the model which identifies a higher
the model which identifies a higher
the model which identifies a higher incidence of drug crime in that area
incidence of drug crime in that area
incidence of drug crime in that area relative to the neighboring areas and
relative to the neighboring areas and
relative to the neighboring areas and again suggests that it’s wide enough at
again suggests that it’s wide enough at
again suggests that it’s wide enough at least to that area on and on and on it
least to that area on and on and on it
least to that area on and on and on it goes and infects until you have this an
goes and infects until you have this an
goes and infects until you have this an image of estimated drug use versus an
image of estimated drug use versus an
image of estimated drug use versus an image a reported drug drop again police
image a reported drug drop again police
image a reported drug drop again police discover each time in an area does not
discover each time in an area does not
discover each time in an area does not mean that there is more illegal drug use
mean that there is more illegal drug use
mean that there is more illegal drug use there it only means that illegal drug
there it only means that illegal drug
there it only means that illegal drug crime was back so what looked like an
crime was back so what looked like an
crime was back so what looked like an unbiased objective prediction now looks
unbiased objective prediction now looks
unbiased objective prediction now looks like a self-fulfilling prophecy
in policing in an area is
in policing in an area is disproportionate to the amount of real
disproportionate to the amount of real
disproportionate to the amount of real criminal activity that is discriminatory
criminal activity that is discriminatory
criminal activity that is discriminatory policing and these models help reduce
policing and these models help reduce
policing and these models help reduce that this is especially concerning given
that this is especially concerning given
that this is especially concerning given that incidents of police violence
that incidents of police violence
that incidents of police violence committed against black and brown people
committed against black and brown people
committed against black and brown people over the years across the country police
over the years across the country police
over the years across the country police disproportionately arrest flex those
disproportionately arrest flex those
disproportionately arrest flex those arrests lead to more arrests stream
arrests lead to more arrests stream
arrests lead to more arrests stream stops important sweets in the courts it
stops important sweets in the courts it
stops important sweets in the courts it leads to harsher sentencing and higher
leads to harsher sentencing and higher
leads to harsher sentencing and higher admission rates given the situation it
admission rates given the situation it
admission rates given the situation it is not surprising that many communities
is not surprising that many communities
is not surprising that many communities would elect to avoid any interaction
would elect to avoid any interaction
would elect to avoid any interaction with law enforcement . the community’s
with law enforcement . the community’s
with law enforcement . the community’s needs are competing with these
needs are competing with these
needs are competing with these predictions and this straight least
predictions and this straight least
predictions and this straight least sybilla’s relationships even for other
sybilla’s relationships even for other
sybilla’s relationships even for other studies have examined the impact of
studies have examined the impact of
studies have examined the impact of different predictive models on targeted
different predictive models on targeted
different predictive models on targeted communities and the results are also
communities and the results are also
communities and the results are also versus predictive models have been used
versus predictive models have been used
versus predictive models have been used importance to predict recidivism like
importance to predict recidivism like
importance to predict recidivism like repeat Olympic or probation or parole
repeat Olympic or probation or parole
repeat Olympic or probation or parole averages and pull of the fact that one
averages and pull of the fact that one
averages and pull of the fact that one of these models was almost twice as
of these models was almost twice as
of these models was almost twice as likely to incorrectly predict a black
likely to incorrectly predict a black
likely to incorrectly predict a black defendant as high-risk reselected these
defendant as high-risk reselected these
defendant as high-risk reselected these models are either used in states across
models are either used in states across
models are either used in states across the country this is very concerning and
the country this is very concerning and
the country this is very concerning and top- in the criminal justice field agree
as these models become normalized in the
as these models become normalized in the criminal justice system they have
criminal justice system they have
criminal justice system they have significant impact their whole
significant impact their whole
significant impact their whole communities and the
communities and the
communities and the well justice system as a whole the data
well justice system as a whole the data
well justice system as a whole the data violence thus the models are biased and
violence thus the models are biased and
violence thus the models are biased and to even give it a perfect and complete
to even give it a perfect and complete
to even give it a perfect and complete record of the past statistics aren’t a
record of the past statistics aren’t a
record of the past statistics aren’t a crystal ball the past isn’t a perfect
crystal ball the past isn’t a perfect
crystal ball the past isn’t a perfect guide to the future
guide to the future
guide to the future the producers of these problems and the
the producers of these problems and the
the producers of these problems and the police departments that you still would
police departments that you still would
police departments that you still would do well to remember that and some of
do well to remember that and some of
do well to remember that and some of these departments are on several
these departments are on several
these departments are on several California police departments have
California police departments have
California police departments have decided against using these moments Oh
decided against using these moments Oh
decided against using these moments Oh Richmond Milpitas imperfect it means
Richmond Milpitas imperfect it means
Richmond Milpitas imperfect it means that as though I really that gives you
that as though I really that gives you
that as though I really that gives you use of analytics and data driven
use of analytics and data driven
use of analytics and data driven approaches in law enforcement and crime
approaches in law enforcement and crime
approaches in law enforcement and crime prevention period but that’s not the
prevention period but that’s not the
prevention period but that’s not the case instead what I’m saying is a we
case instead what I’m saying is a we
case instead what I’m saying is a we need a marketing plan consider the
need a marketing plan consider the
need a marketing plan consider the education police must assess their role
education police must assess their role
education police must assess their role within the community the various tools
within the community the various tools
within the community the various tools of their disposal and their values when
of their disposal and their values when
of their disposal and their values when thinking about the utility of new
thinking about the utility of new
thinking about the utility of new technologies in their work however
technologies in their work however
technologies in their work however complex and detailed the analysis
complex and detailed the analysis
complex and detailed the analysis algorithmic data mining techniques are
algorithmic data mining techniques are
algorithmic data mining techniques are ill-equipped to direct police on who
ill-equipped to direct police on who
ill-equipped to direct police on who what when where why and how to peace
what when where why and how to peace
what when where why and how to peace rather than relying on these tools for
rather than relying on these tools for
rather than relying on these tools for boundaries we should turn elsewhere they
boundaries we should turn elsewhere they
boundaries we should turn elsewhere they should look to the policy leaders the
should look to the policy leaders the
should look to the policy leaders the stakeholders the researchers and
stakeholders the researchers and
stakeholders the researchers and communities above all those are the
communities above all those are the
communities above all those are the people they serve and also they should
people they serve and also they should
people they serve and also they should be market appearing about how they
be market appearing about how they
be market appearing about how they ideally before implementing them
ideally before implementing them
ideally before implementing them inviting a litigation and evaluation we
inviting a litigation and evaluation we
inviting a litigation and evaluation we don’t get to know if these models work
don’t get to know if these models work
don’t get to know if these models work the way we want them to and frankly we
the way we want them to and frankly we
the way we want them to and frankly we can’t business because the companies
can’t business because the companies
can’t business because the companies that produce these models use
that produce these models use
that produce these models use proprietary techniques we don’t have the
proprietary techniques we don’t have the
proprietary techniques we don’t have the data or the algorithms they use to
data or the algorithms they use to
data or the algorithms they use to reduce their analysis if the data is
reduce their analysis if the data is
reduce their analysis if the data is kept either eyespot dermaluxe
kept either eyespot dermaluxe
kept either eyespot dermaluxe researchers civil rights groups and even
researchers civil rights groups and even
researchers civil rights groups and even the police departments lives Ellis can’t
the police departments lives Ellis can’t
the police departments lives Ellis can’t understand how they function
understand how they function
understand how they function despite all of this police departments
despite all of this police departments
despite all of this police departments are still buying music ridiculous models
are still buying music ridiculous models
are still buying music ridiculous models but I have even more collaboration with
but I have even more collaboration with
but I have even more collaboration with researchers to develop a more accurate
researchers to develop a more accurate
researchers to develop a more accurate picture of crime greater attention to
picture of crime greater attention to
picture of crime greater attention to police community relations and more
police community relations and more
police community relations and more equitable policing practices more leads
equitable policing practices more leads
equitable policing practices more leads Department will come to understand how
Department will come to understand how
Department will come to understand how data can use tools to be the stakes are
data can use tools to be the stakes are
data can use tools to be the stakes are very high in criminal justice
very high in criminal justice
very high in criminal justice it is important to get answers that are
it is important to get answers that are
it is important to get answers that are accurate fair and just
Please follow and like us: