PREDICTIVE CODING

48 downloads 138 Views 919KB Size Report
PREDICTIVE CODING: Debunking the Seven. Biggest Myths. David J. Kessler, Partner. Fulbright & Jaworski. Howard Sklar, Senior Counsel. Recommind, Inc.
PREDICTIVE CODING: Debunking the Seven Biggest Myths David J. Kessler, Partner Fulbright & Jaworski Howard Sklar, Senior Counsel

Recommind, Inc.

Meet Our Speakers

David Kessler Partner Fulbright & Jaworski [email protected]

Howard Sklar Senior Counsel Recommind, Inc. [email protected]

2

What is Predictive Coding?  Sampling

is used to determine baseline responsiveness

 Prioritized

review results in highly responsive review batches (30 to 95% responsive)

 Responsiveness  Sampling ●

drops dramatically after limited review

is used to prove that junk pile is really junk

If Responsive Documents are found – additional Predictive Coding iterations are completed until team is satisfied

 Typically

all documents produced are reviewed by

humans

3

Predictive Coding Workflow Step 1

Step 2

Step 3

Step 4

Predictive Analytics to Create Review Sets Human Review

System Training on Relevant Documents

Human Review of Computer Suggested

Statistical Quality Control Validation

Computer Suggested

Adaptive ID Cycles (Train, Suggest, Review)

4

MYTH 1: PREDICTIVE CODING IS AUTOMATED CODING 

Approach 1 – Prioritization and Resource Management ●



Approach 2 – Cull the Review Set Using Predictive Coding ●



Review All; human review to generate seed set and to prioritize

Review Seed Set and Computer Suggested; sample documents not suggested

Approach 3 – Automated Coding ●

Review Seed Set and Produce All Computer Suggested; sample documents no suggested 5

MYTH 2: Defensibility Depends Primarily on the Technology 

This should be put to rest by Judge Peck in De Silva Moore



Defensibility is about process and execution ●



Not about the tool, but whether your discovery process was reasonable and you implemented it in a reasonably way.

Technology matters ●

Technology enables process



Certain technology is more cost effective (note: not cheaper)

6

MYTH 3: Predictive Coding Is Inherently Risky 

Bad assumption (generally based on Myth 1)



Human review is considered reasonable ●

Not necessarily cost effective



Watch Kleen Products from N. D. Ill.



The order a party reviews documents is not a defensibility concern



Approach 1 poses little if any defensibility risk

7

MYTH 4: Culling Documents Using Predictive Coding is Risky 

Predictive coding is not spoliation



What is the risk?



1.

Dispute centered on predictive coding arises.

2.

Dispute cannot be resolved by parties.

3.

Requesting party raises the dispute to the court.

4.

Court finds for the requesting party on all issues.

Absent extraordinary circumstances, the risk from culling is having to review the un-reviewed documents

8

MYTH 5: Technology is Replacing Judgment

 Predictive coding elevates legal judgment  What is important to the case? What does the litigation team?  Old mass contract lawyer model leveraged litigation team using staff counsel  Contract lawyers still critical going forward

 Technology provides more efficient visibility into the data

9

MYTH 6: 99% is Reasonable (and 95% is NOT) 

Lies, Damn Lies and Statistics



Statistics are important, BUT quality checking is more important



What is your story? ●



Why did you make the choices you made and based on what information? ●



Lawyering is still about persuasion, even discovery law

If you were 95% confident that there were no more responsive documents remaining that would be unreasonable?

Discovery is inherently about reasonableness ●

This is not manufacturing or science 10

MYTH 7: Transparency is the Answer 

Not against transparency



Against compelled transparency ●



Transparency is a TOOL



Practical downside to transparency



11

Runs afoul of Fed. R. Civ. P. 26(b)(1)



Savings going to be lost in coding objections?



Privilege logs for irrelevant documents (and new disputes)?



Serial litigants have serial opposing counsel.

Are we shuffling toward perfection?

For More Information . . .

Read the article by David Kessler “Debunking the Seven Biggest Myths of Predictive Coding”

12

Thank You

David Kessler Partner Fulbright & Jaworski [email protected]

Howard Sklar Senior Counsel Recommind [email protected]

13