Uploaded July 2022 | Updated September 2026, 2 weeks ago
The criminal justice system is overburdened and expensive. What if we could harness advances in social science and math to predict which criminals are most likely to re-offend? What if we had a better way to sentence criminals efficiently and appropriately, for both criminals and society as a whole?
That’s the idea behind risk assessment algorithms like COMPAS. And while the theory is excellent, we’ve hit a few stumbling blocks with accuracy and fairness. The data collection includes questions about an offender’s education, work history, family, friends, and attitudes toward society. We know that these elements correlate with anti-social behavior, so why can’t a complex algorithm using 137 different data points give us an accurate picture of who’s most dangerous?
The problem might be that it’s actually too complex -- which is why random groups of internet volunteers yield almost identical predictive results when given only a few simple pieces of information. Researchers have also concluded that a handful of basic questions are as predictive as the black box algorithm that made the Supreme Court shrug.
Is there a way to fine-tune these algorithms to be better than collective human judgment? Can math help to safeguard fairness in the sentencing process and improve outcomes in criminal justice? And if we did develop an accurate math-based model to predict recidivism, how ethical is it to blame current criminals for potential future crimes?
Can human behavior become an equation?
*** ADDITIONAL READING ***
Sample COMPAS Risk Assessment: documentcloud.org/documents/2702103-Sample-Risk-Assessment-COMPAS-CORE
COMPAS-R Updated Risk Assessment: equivant.com/compas-r-core-transparent-rna
“The accuracy, fairness, and limits of predicting recidivism.” Julia Dressel. science.org/doi/10.1126/sciadv.aao5580
“Understanding risk assessment instruments in criminal justice,” Brookings Institution: https://www.brookings.edu/research/understanding-risk-assessment-instruments-in-criminal-justice/
“Machine Bias,” Julia Angwin, Jeff Larson, Surya Mattu and Lauren Kirchner, ProPublica: propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing
“The limits of human predictions of recidivism,” Lin, Jung, Goel and Skeem: science.org/doi/full/10.1126/sciadv.aaz0652
“Even Imperfect Algorithms Can Improve the Criminal Justice System,” New York Times: nytimes.com/2017/12/20/upshot/algorithms-bail-criminal-justice-system.html
Equivant’s response to criticism: equivant.com/official-response-to-science-advances
“A Popular Algorithm Is No Better at Predicting Crimes Than Random People,” Ed Yong: theatlantic.com/technology/archive/2018/01/equivant-compas-algorithm/550646
“The Age of Secrecy and Unfairness in Recidivism Prediction,” Rudin, Wang, and Coker: https://hdsr.mitpress.mit.edu/pub/7z10o269/release/6
“Practitioner’s Guide to COMPAS Core,” s3.documentcloud.org/documents/2840784/Practitioner-s-Guide-to-COMPAS-Core.pdf
State v. Loomis summary: harvardlawreview.org/wp-content/uploads/2017/03/1530-1537_online.pdf
*** LINKS ***
Vsauce2:
TikTok: tiktok.com/@vsaucetwo
Twitter: twitter.com/VsauceTwo
Facebook: facebook.com/VsauceTwo
Talk Vsauce2 in The Create Unknown Discord: discord.gg/tcu
Vsauce2 on Reddit: reddit.com/r/vsauce2
Hosted and Produced by Kevin Lieber
Instagram: instagram.com/kevlieber
Twitter: twitter.com/kevinlieber
Podcast: youtube.com/thecreateunknown
Research and Writing by Matthew Tabor
twitter.com/TaborTCU
Editing by John Swan
youtube.com/channel/UCJuSltoYKrAUKnbYO5EMZ2A
Police Sketches by Art Melt
Twitter: twitter.com/EelJammin
IG: instagram.com/jamstamp0
Huge Thanks To Paula Lieber
etsy.com/shop/Craftality
Vsauce's Curiosity Box: curiositybox.com
#education #vsauce #crime
The criminal justice system is overburdened and expensive. What if we could harness advances in social science and math to predict which criminals are most likely to re-offend? What if we had a better way to sentence criminals efficiently and appropriately, for both criminals and society as a whole?
That’s the idea behind risk assessment algorithms like COMPAS. And while the theory is excellent, we’ve hit a few stumbling blocks with accuracy and fairness. The data collection includes questions about an offender’s education, work history, family, friends, and attitudes toward society. We know that these elements correlate with anti-social behavior, so why can’t a complex algorithm using 137 different data points give us an accurate picture of who’s most dangerous?
The problem might be that it’s actually too complex -- which is why random groups of internet volunteers yield almost identical predictive results when given only a few simple pieces of information. Researchers have also concluded that a handful of basic questions are as predictive as the black box algorithm that made the Supreme Court shrug.
Is there a way to fine-tune these algorithms to be better than collective human judgment? Can math help to safeguard fairness in the sentencing process and improve outcomes in criminal justice? And if we did develop an accurate math-based model to predict recidivism, how ethical is it to blame current criminals for potential future crimes?
Can human behavior become an equation?
*** ADDITIONAL READING ***
Sample COMPAS Risk Assessment: documentcloud.org/documents/2702103-Sample-Risk-Assessment-COMPAS-CORE
COMPAS-R Updated Risk Assessment: equivant.com/compas-r-core-transparent-rna
“The accuracy, fairness, and limits of predicting recidivism.” Julia Dressel. science.org/doi/10.1126/sciadv.aao5580
“Understanding risk assessment instruments in criminal justice,” Brookings Institution: https://www.brookings.edu/research/understanding-risk-assessment-instruments-in-criminal-justice/
“Machine Bias,” Julia Angwin, Jeff Larson, Surya Mattu and Lauren Kirchner, ProPublica: propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing
“The limits of human predictions of recidivism,” Lin, Jung, Goel and Skeem: science.org/doi/full/10.1126/sciadv.aaz0652
“Even Imperfect Algorithms Can Improve the Criminal Justice System,” New York Times: nytimes.com/2017/12/20/upshot/algorithms-bail-criminal-justice-system.html
Equivant’s response to criticism: equivant.com/official-response-to-science-advances
“A Popular Algorithm Is No Better at Predicting Crimes Than Random People,” Ed Yong: theatlantic.com/technology/archive/2018/01/equivant-compas-algorithm/550646
“The Age of Secrecy and Unfairness in Recidivism Prediction,” Rudin, Wang, and Coker: https://hdsr.mitpress.mit.edu/pub/7z10o269/release/6
“Practitioner’s Guide to COMPAS Core,” s3.documentcloud.org/documents/2840784/Practitioner-s-Guide-to-COMPAS-Core.pdf
State v. Loomis summary: harvardlawreview.org/wp-content/uploads/2017/03/1530-1537_online.pdf
*** LINKS ***
Vsauce2:
TikTok: tiktok.com/@vsaucetwo
Twitter: twitter.com/VsauceTwo
Facebook: facebook.com/VsauceTwo
Talk Vsauce2 in The Create Unknown Discord: discord.gg/tcu
Vsauce2 on Reddit: reddit.com/r/vsauce2
Hosted and Produced by Kevin Lieber
Instagram: instagram.com/kevlieber
Twitter: twitter.com/kevinlieber
Podcast: youtube.com/thecreateunknown
Research and Writing by Matthew Tabor
twitter.com/TaborTCU
Editing by John Swan
youtube.com/channel/UCJuSltoYKrAUKnbYO5EMZ2A
Police Sketches by Art Melt
Twitter: twitter.com/EelJammin
IG: instagram.com/jamstamp0
Huge Thanks To Paula Lieber
etsy.com/shop/Craftality
Vsauce's Curiosity Box: curiositybox.com
#education #vsauce #crime
