Black Box Tinkering: Beyond Disclosure in
Description: Black Box Tinkering: Beyond Disclosure in Algorithmic Enforcement Maayan Perel niva elkin-koren Netanya Academic College Haifa Center for law technology TILTing perspectives 2017 Regulating a Connected World Presentation Structure Part
Related Topics
Download Presentation
"Black Box Tinkering: Beyond Disclosure in" is the property of its rightful owner. Permission is granted to download and print the materials on this website for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Presentation Transcript
slide1. Black Box Tinkering:Beyond Disclosure in Algorithmic Enforcement Maayan Perel & niva elkin-koren
Netanya Academic College Haifa Center for law & technology TILTing perspectives 2017
Regulating a Connected World<br>
slide2. Presentation Structure Part 1:Algorithmic Enforcement & Accountability Challenges
Part 2: Our Methodology & Benefits: Black Box Tinkering
Part 3: Application & Analysis<br>
slide3. Part 1: Algorithmic Enforcement<br>
slide4. Algorithmic Enforcement by Online Intermediaries<br>
slide5. Algorithmic Enforcement by Online Intermediaries<br>
slide6. Algorithmic Enforcement by Online Intermediaries ALGORITHM MONITORS PLATFORM BASED ON WHAT CRITERIA?<br>
slide7. Algorithmic Enforcement & Oversight: Why? Ubiquitous: algorithmic enforcement is ubiquitous
Privatization of Enforcement
- No Judicial Oversight
- No sufficient checks and balances
3. Critical implications for public discourse and fundamental rights
- Free flow of information
- Free speech
- Fair use
- Competition<br>
slide8. Algorithmic Enforcement & Oversight: How? Algorithms are complex and opaque + machine learning
Algorithms developed by private actors are protected by IP (trade secrets)
Transparency is insufficient
- Unintelligible
Information overflow
Partial/misleading/biased information<br>
slide9. Part 2: Our Methodology - Black Box Tinkering Challenge algorithmic decision-making with real-world content and record response.
Benefits:
Independent data extraction (vr. voluntarily disclosed)
Creating a condensed data-set (vr. too much info)
Dynamic & adjustable research methodology (vr. static)
Proactive: an ongoing check (vr. observational)
Social activism: encourages public participation/interaction<br>
slide10. Part 3: Application Case Study: Algorithmic Copyright Enforcement
Statutory framework: DMCA N&TD
DMCA Plus: beyond N&TD
Blocking, filtering ex ante, monetizing (Content ID)<br>
slide11. Experiment Flowchart Uploading Content
Check: was it uploaded? Blocked? Removed? Yes End No Yes Notice
Check: was It removed Counter Notice
Check: re-loaded? End Yes End No No Re-loading infringing materials
Does the system preserve information?<br>
slide12. Fair Use Consideration? “Little Lenz” (Justin Bieber baby)<br>
slide13. Non infringing video<br>
slide14. Infringing music only Yanni
Song for Antartica<br>
slide15. Infringing Image<br>
slide16. Infringing (w Content ID) House
Season 4 Episode 14<br>
slide17. Infringing (w/out Content ID ) House
Season 4 Episode 13<br>
slide18. Local N&T practices are Chaotic A wide variety of policies
- No single standard
- Pre-screening
- Removal of non-infringing materials
- Removal of fair use materials
- Takedown and stay-down
Low level of “due process”
- No notification
- Low level of ownership verification<br>
slide19. Barriers to Black Box Tinkering Restrictions defined by law
- Uploading infringing content
- Oath Requirement
- Computer Fraud and Abuse Act (CFAA): accessing website in a manner that violates the websites terms of use is a crime (Sandvig v. Attorney General)
Restrictions in Terms-of-use<br>
slide20. Solutions? Safe harbor for research purposes
A right to tinker?
Thank you!
maayanfilmar@gmail.com<br>
Netanya Academic College Haifa Center for law & technology TILTing perspectives 2017
Regulating a Connected World<br>
slide2. Presentation Structure Part 1:Algorithmic Enforcement & Accountability Challenges
Part 2: Our Methodology & Benefits: Black Box Tinkering
Part 3: Application & Analysis<br>
slide3. Part 1: Algorithmic Enforcement<br>
slide4. Algorithmic Enforcement by Online Intermediaries<br>
slide5. Algorithmic Enforcement by Online Intermediaries<br>
slide6. Algorithmic Enforcement by Online Intermediaries ALGORITHM MONITORS PLATFORM BASED ON WHAT CRITERIA?<br>
slide7. Algorithmic Enforcement & Oversight: Why? Ubiquitous: algorithmic enforcement is ubiquitous
Privatization of Enforcement
- No Judicial Oversight
- No sufficient checks and balances
3. Critical implications for public discourse and fundamental rights
- Free flow of information
- Free speech
- Fair use
- Competition<br>
slide8. Algorithmic Enforcement & Oversight: How? Algorithms are complex and opaque + machine learning
Algorithms developed by private actors are protected by IP (trade secrets)
Transparency is insufficient
- Unintelligible
Information overflow
Partial/misleading/biased information<br>
slide9. Part 2: Our Methodology - Black Box Tinkering Challenge algorithmic decision-making with real-world content and record response.
Benefits:
Independent data extraction (vr. voluntarily disclosed)
Creating a condensed data-set (vr. too much info)
Dynamic & adjustable research methodology (vr. static)
Proactive: an ongoing check (vr. observational)
Social activism: encourages public participation/interaction<br>
slide10. Part 3: Application Case Study: Algorithmic Copyright Enforcement
Statutory framework: DMCA N&TD
DMCA Plus: beyond N&TD
Blocking, filtering ex ante, monetizing (Content ID)<br>
slide11. Experiment Flowchart Uploading Content
Check: was it uploaded? Blocked? Removed? Yes End No Yes Notice
Check: was It removed Counter Notice
Check: re-loaded? End Yes End No No Re-loading infringing materials
Does the system preserve information?<br>
slide12. Fair Use Consideration? “Little Lenz” (Justin Bieber baby)<br>
slide13. Non infringing video<br>
slide14. Infringing music only Yanni
Song for Antartica<br>
slide15. Infringing Image<br>
slide16. Infringing (w Content ID) House
Season 4 Episode 14<br>
slide17. Infringing (w/out Content ID ) House
Season 4 Episode 13<br>
slide18. Local N&T practices are Chaotic A wide variety of policies
- No single standard
- Pre-screening
- Removal of non-infringing materials
- Removal of fair use materials
- Takedown and stay-down
Low level of “due process”
- No notification
- Low level of ownership verification<br>
slide19. Barriers to Black Box Tinkering Restrictions defined by law
- Uploading infringing content
- Oath Requirement
- Computer Fraud and Abuse Act (CFAA): accessing website in a manner that violates the websites terms of use is a crime (Sandvig v. Attorney General)
Restrictions in Terms-of-use<br>
slide20. Solutions? Safe harbor for research purposes
A right to tinker?
Thank you!
maayanfilmar@gmail.com<br>