1:03

Introducing carbon and value tracker where you can

1:05

track your cars value over time and learn

1:07

what's driving it. It might make you a

1:09

sign of low. Didn't know my car was

1:11

valued this high in might make you nervous.

1:14

Oh, markets flooded my car's value. Just dip

1:16

two point three percent. It might make you

1:18

optimistic a low mileage as paying off our

1:20

values up and it might make you realistic.

1:23

Car prices have gone up in a couple weeks?

1:25

maybe? It's time to sell, but it will

1:27

definitely make you an expert on your cars.

1:29

Value Carvalho value Tracker! Visit carbona.com to start

1:32

tracking your cars Valued. Today. The

2:00

Brick Gladstone is off this week.

2:02

I'm Michael O'Linger. You

2:06

might not know it, but recently there

2:08

was some breaking news in the retail

2:10

world that has links to a story

2:12

that has been doing the rounds for

2:14

quite a while now. Tonight a rash

2:16

of smash and rap fans targeting retailers

2:19

around the country, just as the holiday

2:21

shopping season picks up. A mob of

2:23

20 to 30 male and female suspects

2:25

wearing masks and hoodies, seen smashing

2:28

merchandise, running out of Nordstrom and

2:30

escaping. The items that you would

2:32

normally just grab and throw in

2:34

your purse are now under

2:36

lock and key. This thing is

2:38

an epidemic. It's spreading faster than

2:40

COVID, Steve. Sensational reports

2:42

use a new term, organized

2:45

retail crime. Retailers

2:47

we talk to are losing billions of

2:50

dollars to organized retail crime, and

2:52

authorities are warning that this has

2:54

become an absolute threat to public

2:56

safety, with violent gangs, dangerous international

2:58

crime rings and even groups with

3:01

suspected ties to terrorism increasingly getting

3:03

involved. Whether it's new laws or

3:05

amending existing laws to be able to

3:07

stiffen the penalties for repeat

3:09

offenders who are doing this

3:11

for financial gain, it's

3:14

a start. That last voice is

3:16

David Johnston. He's the Vice President

3:19

of Asset Protection and Retail Operations

3:21

for the National Retail Federation. The

3:24

National Retail Federation, the N.R.F.,

3:26

the biggest retail lobbying group

3:28

representing more than 16,000 companies,

3:32

including Target and Walmart. If

3:35

you've seen or read coverage about

3:37

organized retail crime and the new

3:39

bills aimed at cracking down on

3:42

this type of theft, you've likely

3:44

encountered N.R.F. spokespeople and statistics. For

3:47

a broader look at the issue in the retail

3:49

industry, we want to bring in Matt Shea,

3:51

these National Retail Federation's President and CEO. The

3:54

N.R.F.'s latest report on the issue shows that

3:56

retail shrinkage is on the rise. organized

4:00

retail crime really is a growing and a

4:02

persistent threat as the study you referenced just

4:04

a minute ago determined

4:06

and really illustrated in very stark

4:09

ways." That study, which included some

4:11

dubious data, is the subject of

4:13

this piece. But before

4:15

we go further, I think we should

4:17

define our terms. Let's start with shrinkage.

4:21

You mean shrinkage. Yes! Significant

4:23

shrinkage. So you

4:25

feel you were short-changed. Yes! No,

4:29

not that shrinkage. So shrink,

4:32

shrinkage is inventory

4:34

loss. Daphne Howland is a

4:36

senior reporter at Retail Dive,

4:38

an industry website reporting on

4:40

news and trends in retail.

4:42

Unaccounted for inventory. What

4:45

happened to these items? Lost,

4:47

damaged, maybe an

4:49

accounting mistake, and

4:52

stolen. Just the overall

4:54

number. Overall

4:56

number of goods unaccounted

4:58

for in a year. And

5:01

then recently we've learned

5:03

this kind of new term, organized

5:05

retail crime. It means something very

5:07

specific. Can you describe it? The

5:10

N.R.F. has a little bit

5:12

of a convoluted definition, but it boils

5:14

down to three or more

5:18

individuals robbing a

5:20

store and taking the

5:22

stolen goods to resell as

5:25

opposed to personal use. That's

5:27

key because a lot of shoplifting, whether

5:30

it's a teenager swiping a

5:32

pack of gum or maybe an

5:34

impoverished mother who needs diapers or

5:36

something, that's personal use.

5:39

This is a whole different scale. It means someone's going to probably

5:42

sell it online or through FB Marketplace

5:44

or a place like that. In

5:47

its latest annual report originally

5:49

released in April, the National

5:51

Retail Federation estimated that in

5:53

2021, shrinkage, overall loss of

5:55

inventory cost these stores $95

5:58

billion. and

6:00

that organized retail crime made up

6:02

nearly half of that. This

6:05

claim referenced data from another retail association

6:07

that estimated that groups of thieves had

6:09

swiped a total of $45 billion worth

6:11

of stuff in 2021. $45

6:17

billion. A

6:19

couple weeks ago, Daphne fact-checked that number

6:21

and found it to be baseless. After

6:24

my story came out, the NRF removed

6:27

this reference to $45 billion

6:30

of losses to organized retail crime from

6:33

a crime report that they released

6:35

in April. It almost sounds small

6:37

potatoes when you're like, National

6:39

Retail Federation retracts a number

6:41

from 2021 until you

6:44

realize that that's the only number they

6:46

got. And now they don't have it. There

6:49

is no number related to organized

6:51

retail crime. That doesn't mean

6:53

there's not organized retail crime. It just means

6:56

nobody has a number. I

6:58

began trying to figure out how

7:00

that now retracted claim first entered

7:03

the NRF's crime report, the media,

7:05

and even the congressional record. It

7:08

seemed to get significant attention following

7:10

a September 2021 Wall Street

7:12

Journal profile of a man

7:14

named Ben Dugan. So

7:17

he heads up a group

7:19

that is basically a partnership

7:21

between the retail industry and

7:23

law enforcement and loss prevention.

7:25

Clear the coalition of law

7:27

enforcement and retail. He also

7:30

works for CVS as a key

7:32

loss prevention officer for them.

7:35

I'm Ben Dugan. I'm the director

7:38

of organized retail crime and corporate

7:40

investigations for CVS Health. So instead

7:43

of like Law and Order SVU,

7:45

it's like Law and Order CVS.

7:49

Exactly. Something like that. This is

7:51

Dugan speaking with Ryan Knudsen, the

7:53

host of The Journal, the flagship

7:56

Wall Street Journal podcast for an

7:58

episode all about shoplifting. One

8:00

trade group estimates that shoplifting costs US retailers

8:02

$45 billion a year, which

8:06

is up 50% from a decade

8:08

ago. That

8:10

trade group is Ben Dugan's clear. Note

8:13

that in this reference, the $45 billion

8:16

is attributed to shoplifting writ large,

8:18

a much bigger bucket than organized

8:20

retail crime, which is a kind

8:22

of shoplifting. There's no

8:24

interrogation of that statistic. No one

8:27

asks how Dugan got that number.

8:29

In this podcast or in the extensive

8:32

print profile of Dugan that preceded it,

8:34

the print story in the Wall Street

8:36

Journal creates some buzz in the retail

8:38

world. And then two months later, we

8:41

hear from Dugan again, this time on

8:43

a much bigger stage. Good morning, Chairman

8:46

Durbin, ranking member Grossley and members of

8:48

the committee. Here's Dugan speaking before the

8:51

Senate Judiciary Committee on November 2nd, 2021.

8:55

I want to share firsthand today what I've

8:57

experienced over 30 years of working on this

8:59

problem. Organized retail crime

9:01

represents a massive and growing threat

9:03

to the tune of $45 billion

9:06

a year. These criminal

9:08

organizations, employee teams or crews are

9:11

professional thieves that steal the products by

9:13

any means necessary and sell

9:15

them through online marketplace. Now

9:18

the $45 billion in losses

9:20

is attributed solely to organized

9:22

crime. It's a big claim. It's

9:24

a definite claim. Daphne Howland,

9:27

when she started digging into the

9:29

2023 National Retail Federation's report, she found

9:32

that the source for the claim that

9:34

roughly half of the shrinkage losses in

9:36

2021 came from

9:38

organized retail crime was Dugan's

9:40

Senate testimony. Dugan said

9:43

that clear estimates this $45

9:46

billion figure, but I

9:48

couldn't find any reports

9:51

or research statements or anything

9:53

on the clear website or

9:55

anywhere. And actually, as

9:57

the LA Times did that same year in

9:59

2020, in 2021, they questioned

10:01

this number and tried to figure

10:03

out where it came from.

10:06

She eventually just went to Ben Dugan himself and

10:08

asked where he got that number. In an

10:10

email that Daphne shared with us,

10:13

Dugan said the National Retail Federation.

10:16

That $45 billion was a

10:18

number offered up by the National

10:20

Retail Federation in 2016, which was

10:24

their estimate of total

10:26

shrink or total inventory loss

10:29

for 2015. So he

10:31

goes to the Senate, he

10:34

tells them that organized

10:36

retail crime accounts for $46 billion of

10:39

loss for retailers. And he's

10:42

just talking about a

10:44

number from 2016 that describes something completely

10:47

different. Exactly. The

10:50

National Retail Federation was saying that roughly

10:52

half of all shrinkage in 2021 was

10:55

the result of organized retail crime. Their

10:58

source was Ben Dugan,

11:00

who, it turns out, was

11:02

quoting old numbers from National

11:04

Retail Federation, quoting themselves in

11:06

effect. Quoting themselves and

11:08

quoting themselves wrongly, you

11:11

know, making their own math

11:13

mistake. And it seems

11:15

like a mistake that should have been

11:17

taught, or maybe put it this way,

11:19

it should have been a number

11:22

whose provenance should have been

11:24

sort of investigated. So

11:27

we can and should put that $45 billion

11:29

number to rest. But, you

11:32

know, maybe these lobbying groups

11:34

don't mind if journalists fail

11:36

to scrutinize their claims. We've

11:38

heard from the National Retail Federation saying

11:40

that organized crime is behind it. I'm

11:42

seeing estimates here that it's costing retailers

11:44

nearly $100 billion a year. It's

11:50

a crime that cost businesses $96.5 billion in 2021,

11:52

according to the National Retail Federation.

11:58

The National Retail Federation. full

14:00

statements head to onthemedia.org.

14:05

Coming up, bad shoplifting coverage

14:07

leads to bad shoplifting laws.

14:10

This is On the Media. On

14:19

the Media is supported by Masterclass. Masterclass

14:21

makes a meaningful gift this season for

14:24

you and anyone on your list because

14:26

any recipient can benefit from learning from

14:28

the best to become your best. From

14:30

leadership to effective communication to cooking. With

14:33

a Masterclass annual membership it's $10 a

14:36

month. Memberships start at $120

14:38

a year for unlimited access

14:40

to classes with over 180

14:42

Masterclass instructors. Here's something we

14:44

could all benefit from no matter who you

14:46

are. Learning how to negotiate a raise with

14:48

Chris Voss. Through the session you'll

14:50

learn among other important takeaways, 12 proven

14:53

tactics to use every day at work

14:55

for big meetings or small moments. Boost

14:58

your confidence and find practical takeaways you can

15:00

apply to your life and at work. And

15:02

if you own a business or are

15:04

a team leader use Masterclass to empower

15:06

and create future ready employees and leaders.

15:09

This holiday season give one

15:11

annual membership and get one

15:14

free at masterclass.com/otm15. That's two

15:17

memberships for the price

15:19

of one at masterclass.com/otm15.

15:21

Offer terms apply.

15:24

WNYC Studios is supported by

15:27

Carvana. Introducing carbon and value

15:29

tracker where you can track your cars value

15:31

over time and learn what's driving it. It

15:33

might make you a sign of low. Didn't

15:35

know my car was valued this high in

15:37

might make you nervous. Oh, markets flooded

15:39

my car's value. Just dip two point three

15:41

percent. It might make you optimistic a low

15:43

mileage as paying off our values up and

15:46

it might make you realistic. Car prices

15:48

have gone up in a couple weeks? maybe? It's

15:50

time to sell, but it will definitely make

15:52

you an expert on your cars. Value Carvalho

15:55

value Tracker! Visit carbona.com to start tracking your

15:57

cars Valued. Today. Hey y'all I'm

15:59

Rhiannon Gibson. host of Aria Code

16:01

and I'm here to spread the gospel

16:03

of opera. We are all gathered to

16:05

experience the magic of great human voices,

16:09

beautiful staging and big human

16:11

drama. We're bringing together singers,

16:13

experts and unexpected guests to

16:16

reveal the complexities of opera

16:18

and life one aria at

16:20

a time. Don't miss the

16:22

new season of Aria Code. Listen

16:24

wherever you get podcasts. This

16:30

is on the media. I'm Michael Lowinger. Retailers

16:32

have been lobbying the press and

16:34

legislators to support a slate of

16:36

new laws aimed at fighting the

16:39

retail crime epidemic. But

16:41

when Nicole Lewis, the engagement editor

16:44

at the Marshall Project, began digging

16:46

into shoplifting data from the National

16:48

Retail Federation, other lobbying

16:50

groups and law enforcement, she

16:52

became less convinced there was

16:54

an epidemic at all. In

16:57

February, she published a piece called What

16:59

the Panic Over

17:01

Shoplifting Reveals About American Crime Policy.

17:04

She says that the crime stats

17:06

provided by the stores themselves might

17:08

help clarify the situation if

17:10

only she had a chance to look at them. So

17:13

every year these retailers report their numbers. This

17:15

is a self survey. When I asked the

17:17

retail federation to say, can I dive into

17:19

this raw data? I'm really curious how it's

17:22

collected. They said, no, of course not. No,

17:24

we're not going to get that to you.

17:26

Right? That's proprietary information. And

17:28

when you say self survey, you mean

17:30

a trade organization asking CVS, Walgreens, Target.

17:32

What are you seeing? Exactly.

17:35

And then the other layer here is that

17:38

when I went back and I dug into

17:40

every single survey that they've conducted between 2016

17:42

and now out of curiosity, I was like,

17:44

well, has this problem actually been getting worse?

17:48

The share of stolen merchandise

17:50

that comes from actual retail

17:52

theft has stayed stable around

17:55

1.4%, 1.5% for each of

17:57

those years. years.

18:00

So basically what that is saying to

18:02

me is yes, the cost

18:04

of the overall amount has increased, but

18:06

the share that they're attributing to people

18:08

actually stealing has stayed the same from

18:10

2016 to 2021 when they did their

18:13

last survey. Wait,

18:15

sorry, I'm a little slow. So explain that to me. Yeah.

18:18

So basically the headline is it's

18:20

actually not getting worse. Things are

18:22

getting more expensive. So if someone were

18:24

to steal 10 gallons of milk in

18:27

2016, that's one cost. But

18:29

if you stole 10 gallons of milk today, it's more

18:32

expensive. So that's how we get from 90 to 94 billion.

18:36

So their own data does not bear this

18:38

out. Let's talk about what should

18:40

be more reliable data like law

18:43

enforcement data. Are they seeing

18:45

this big spike in so-called

18:47

organized retail crime that we've

18:49

been hearing about? Organized

18:52

retail theft is a

18:54

new category of this kind of

18:56

crime. The umbrella term would

18:58

be property theft, just generally stealing.

19:01

So in a lot of places,

19:03

the police couldn't even really charge

19:05

you with organized retail theft because

19:07

it doesn't exist as a crime

19:09

category yet. When I asked police

19:11

departments and ask states and ask task

19:13

forces, is this something you're telling? They

19:16

say no. For the most part,

19:18

shoplifting exists in this amorphous

19:20

lumpy category of property theft. Moving

19:24

the obvious sort of propaganda

19:26

elements aside, New

19:29

York has seen a spike

19:32

in shoplifting, right? That we

19:34

can prove. Correct. Just

19:36

recently, the Council on Criminal Justice

19:38

looked at the police data that

19:40

we do have available to say,

19:42

what's happening here? And in New

19:44

York, they found something really important,

19:46

that shoplifting in New York is

19:48

16% higher in the first

19:50

half of 2023 than

19:52

in the first half of 2019. But

19:55

I think what's important to know is when we take New

19:57

York City out, the trend changes. And

20:00

so shoplifting actually declined over those

20:02

periods. And out of

20:04

the 24 cities that the Council

20:07

on Criminal Justice looked at, 17

20:09

reported decreases in shoplifting. So again, it

20:11

really does contradict that national narrative

20:13

that is surging out of

20:15

control everywhere and tells us

20:17

that in some places, this is

20:20

a really concentrated major issue. So

20:23

the data is either incomplete and

20:25

unreliable or outright does not support

20:28

this narrative. And yet, you

20:31

can't tell me that those

20:33

videos of cars crashing into

20:35

stores or shoplifters flashing

20:37

guns or knives on Home Depot

20:39

employees or flash mobs rushing out

20:41

of a store with loads of

20:43

goods are not real. Yeah,

20:46

no, this is a really important point because

20:48

I don't want to make

20:50

it seem like I'm saying, oh, calm down,

20:52

it's fine. We've all seen those

20:54

videos. And I was actually even in a

20:57

CVS not that long ago in my neighborhood. And

21:00

a man came rushing in with a duffel

21:02

bag. He clears off a shelf. He

21:04

runs out. He does say, thank you. Polite.

21:08

Right, thank you for these stolen goods. All

21:11

of the CVS attendants are kind

21:13

of standing around. The police are

21:15

nowhere in sight. So absolutely, something

21:17

is happening. I think the question

21:19

here is, there are a couple. One

21:22

is, is it really as bad as

21:25

the executives say or

21:28

as they'd like us to believe? And

21:30

then the other big question is, what

21:32

do we do about it? Are the

21:35

methods that they're advocating for, is

21:37

the involvement of law enforcement stiffening

21:39

of penalties, is that actually going

21:41

to solve the problem? When I

21:43

think about how consequential crime policy

21:46

is to Americans' lives, that's really where

21:48

I'm sort of saying, I think we

21:50

need to pull back and really make

21:52

sure that we're understanding the scope of

21:54

this problem, what we can and

21:56

cannot say about it. Yeah, I

21:58

was reading a piece on the Wall Street. Journal about

22:01

the so-called brazen burglars who

22:03

are leading part of

22:05

this shoplifting epidemic. And

22:08

the article made it clear that

22:10

on the West Coast with quote-unquote

22:12

more lenient law enforcement policies, we're

22:15

seeing a bigger problem. So tell

22:17

me how this perception, this

22:19

fear of an increase in

22:21

shoplifting is translating into policy.

22:25

Yeah, I mean, it's so fascinating,

22:27

right? Because even hearing that talking

22:29

point in the Wall Street Journal,

22:31

I'm thinking, this sounds so shockingly

22:33

familiar to what the lobbyists for

22:35

retailers were telling me, right? But when

22:37

I go to talk to criminologists and

22:40

I say, is it true that

22:42

lenient policies around property theft are

22:45

driving up property crime? Do we

22:47

have any data or research

22:49

that backs that up? And

22:51

the resounding answer is just that is not

22:53

how it works. And so one good example

22:56

of this comes from Pew. There

22:58

has been this trend over the last several

23:00

years for states to

23:02

basically increase the threshold, increase the

23:05

amount of goods you'd have to

23:07

steal before triggering a felony charge.

23:10

And this is really important because let's say

23:12

you set that number in the 1990s. So

23:16

today, again, just like the milk

23:18

analogy, you don't have to steal

23:20

nearly as much milk before you would trigger a

23:22

felony. And so that means that people were getting

23:25

felony charges for a thing that wouldn't actually have been

23:27

a felony in the 90s. If you

23:29

set the price needed for a felony higher,

23:31

then in theory, you're locking up fewer people.

23:33

Right. So this is a really important reform.

23:37

And a number of states said, oh, my God, yeah,

23:39

we should probably do something about this. And

23:41

so Pew went back and they looked and they said,

23:44

OK, well, now we have this natural experiment where we

23:46

can say, does it actually matter if

23:48

people are going to be charged with a

23:50

felony? Does that increase or decrease the amount

23:52

that people are going to steal? And

23:55

the answer was clearly no. States that

23:57

didn't change their threshold. did

24:00

change their threshold had the same overall

24:02

decline in the overall property crimes in

24:04

the last few decades of states that

24:06

did not. Adam Chapnick Give me a

24:08

sense of how widespread these shoplifting laws

24:10

are. I mean, are they growing? Dr.

24:26

YOUNG-MAYER Virginia lawmakers passed a bill making

24:28

it a class three felony to steal more than $5,000 worth

24:39

of retail from one or more stores over the

24:41

course of 90 days. Organized

24:44

retail theft, that's been an issue here

24:46

in Minnesota. Adam Chapnick Yeah, bipartisan legislation

24:48

aimed at fighting organized retail theft passed

24:50

this spring. It adds much stiffer penalties

24:52

including prison time. The Alabama District Attorney's

24:54

Association says that the state will adopt

24:57

new laws for stiffer penalties. There's a

24:59

new law that's designed to crack down

25:01

on this type of crime in North

25:03

Carolina. Adam Chapnick In New Mexico, prosecutors

25:05

will now be able to combine the

25:07

value of merchandise stolen from various stores

25:10

over a 90-day period, making

25:12

it easier to charge serial shoplifters

25:15

with felonies rather than misdemeanors. Virginia

25:18

lawmakers So this is a large share of

25:20

states now that are being influenced by

25:22

a trade association and making these

25:24

penalties harsher or adding

25:27

new categories of crimes to

25:29

their statutes. Adam Chapnick

25:31

So let's just talk a little bit about incentives here.

25:34

Obviously, CVS, Walgreens, Target, they

25:36

don't want people to steal from

25:38

their stores, clearly. Why

25:41

else are they pushing this

25:44

unfounded narrative of a national shoplifting

25:46

spike? Virginia lawmakers This story that

25:48

I wrote even started because

25:51

we heard a Walgreens executive

25:53

reporting out of an earnings call basically

25:55

saying, we cried too much

25:57

last year about how many products

26:00

disappearing from our shelves. Basically

26:02

just admitting like why say

26:04

that? Yeah well they said

26:07

this because there's just huge financial

26:09

implications. They had paid to hire

26:11

private security to police their

26:13

stores to make sure that the merchandise stays

26:16

on the shelves and that's the cost

26:18

to them and then the data

26:20

says maybe we didn't need to spend that much

26:22

so they walked it back

26:24

and he was basically announcing the decision

26:27

to let those private security companies

26:29

go that was sort of the

26:31

context of the admission but just

26:33

a few years earlier Walgreens

26:36

had tried to use shoplifting and

26:38

rampant shoplifting as the reason behind

26:40

their decision to close five stores

26:42

in San Francisco but what we

26:45

know about the Walgreens scenario

26:47

is it turned out not to be

26:49

true. These decisions had been made months

26:51

if not years prior and then I'd

26:53

say there's one other thing that becomes

26:56

really important and it's all about the

26:58

cost to these retailers. So

27:00

if you can make it so

27:02

that there's a public panic, the

27:04

public sees these videos, they go

27:06

into CVS and items are behind

27:08

plexiglass, there's a public concern, there's

27:11

a need to say we really

27:13

got to get a handle on

27:15

this. Lawmakers tend

27:17

to be pretty responsive to that

27:19

kind of public energy about an

27:21

issue and so the

27:23

retailers get to externalize the cost

27:25

of dealing with this onto state

27:28

lawmakers and law enforcement to say

27:30

hey it's police and

27:33

it's prosecutors and it's the state legislature that

27:35

needs to change the laws in order for

27:37

us to get a handle on it. And

27:40

there's reason to believe that they

27:42

will get their way because they're

27:45

lobbying both at the national level

27:47

and at the sort of regional

27:49

level is just super strong. Yeah

27:51

I mean retailers are a very

27:54

very important part of our economy.

27:56

They're major employers and

27:58

the transactions that happen in stores

28:00

generate a ton of sales tax.

28:02

So states have a huge interest

28:04

in making sure that retail interests

28:07

are taken care of. I've

28:09

seen credulous profiles

28:12

of executives who are

28:15

combating the shoplifting epidemic

28:18

at their store nationwide.

28:21

We've seen this kind of

28:23

breathless TV reporting that includes

28:26

these sensational but otherwise not

28:28

representative clips on

28:30

TV news. We

28:32

know what bad coverage looks like.

28:35

What is good coverage of shoplifting

28:37

look like? There's one simple

28:40

thing, just one thing that

28:43

reporters can do when they're faced

28:45

with these stories, when they have law

28:47

enforcement or they have executives coming to them

28:49

saying, this problem is so out of hand

28:52

and we got to close all these stores.

28:54

You say, thank you for this information. How

28:56

do you know it is true and can

28:58

you show me? We need

29:00

to see the data and the sources

29:02

that they're looking at that inform their

29:04

talking points. Nicole, thank you very

29:06

much. Thank you so much for having me. Nicole

29:09

Lewis is the engagement editor for

29:11

the Marshall Project. When

29:15

it comes to fear around shoplifting,

29:17

it's more about vibes and dramatic

29:20

footage than data. The same could

29:22

be said about crime in general.

29:25

Jeff Asher is a data analyst

29:27

and co-founder of A.H. DataLitix. In

29:29

his recent sub-stack column, Americans are

29:32

bad at perceiving crime trends, he

29:34

posed the question, why

29:36

do people always think that crime is

29:38

rising? The Gallup poll shows more Americans

29:40

fear becoming victims of crime. A near

29:43

record, 40 percent say they're afraid to

29:45

walk alone at night within a mile

29:47

of their home. Every year, Gallup conducts

29:49

a survey about how crime is perceived

29:52

in the U.S. Its latest

29:54

results were released in October. The

29:56

last time we saw this level of concern about crime

29:58

was back in 1990. 93,

30:00

which Gallup points out was, quote, what

30:03

are the worst crime waves in U.S.

30:05

history? The problem with the

30:07

way Gallup conducts the poll, says

30:09

Jeff Asher, is that while the

30:11

term crime covers everything from jaywalking

30:14

to theft, when polled, people go

30:16

straight to murder. It is the

30:18

crime that people tend to think

30:20

about the most, but it only

30:22

makes up 0.2 percent

30:24

of all major crimes. So we've

30:27

created this situation where we're asking

30:29

people, hey, what do you think

30:31

about crime without actually defining it? It's

30:33

like saying, you know, what do you think about football

30:36

without defining? Do you mean football,

30:38

American football? Do you mean soccer? Do you

30:40

mean Premier League? Do you mean NFL? Do

30:43

you mean college football? Do you mean my

30:45

five year old's team? When it comes to

30:47

the way that Gallup is asking people about

30:49

their perceptions and crime, which might also reveal

30:52

the misconceptions people have in general is that

30:54

the people conducting the poll are kind of

30:56

baking the misconceptions into the questions with this

30:58

kind of ambiguity? Absolutely.

31:01

Because of the ambiguity, if you mean

31:03

major crime, major crime rose in 2022

31:06

because property crime after the sort

31:08

of artificial dip in 2020 and 2021 during coronavirus

31:12

where everybody was home, it

31:14

went down. And so it rose relative to that

31:16

in 2022. Other

31:18

types of crime, violent crime fell, murder

31:20

fell in 2022. But

31:24

without knowing which of those the person the

31:26

respondent is thinking about, it's hard to necessarily

31:28

respond. The second factor that you

31:30

cite for why Americans aren't great at

31:32

knowing how our country is doing on

31:34

crime is that specific questions about crime

31:36

are actually hard to answer. Things

31:39

like violent crime have fallen substantially

31:41

since the 90s. 40%

31:43

decline in violent crime, large decreases

31:46

in murder, large decreases in property

31:48

crime since the 90s. But

31:50

they're not asking if these things have changed since

31:52

the 90s. They're asking, have things changed since last

31:54

year? And the year to year

31:56

changes have been way more subtle. So

31:59

violent crime. fell very slightly

32:01

in 2022 according to the FBI data,

32:03

it rose very slightly in 2021 according

32:05

to the FBI data. If it's

32:08

a one or 2% change in prime for

32:10

five to 10 straight years, I

32:12

think it's a lot harder for somebody

32:14

that's not inherently a data expert to

32:16

understand year on year, whether

32:19

or not it's going up or it's going

32:21

down. Which brings us to your

32:23

third point that the data is hard to

32:25

come by. First off, can

32:28

you just kind of describe for

32:30

us how national crime data get

32:32

compiled? So each

32:34

agency reports each year, they

32:36

have until April of the following year to

32:39

send to typically their state

32:41

UCR, State Uniform Crime Report

32:43

Program, all of the

32:45

major crimes that occurred in a given

32:47

year. They send it to their state

32:49

UCR program, the state UCR program collects

32:51

it, sends it to the FBI, the

32:54

FBI collects it all and publishes it.

32:56

So there's a long lag between formally

32:58

when these agencies get the data and

33:01

when they have to report it up the

33:03

chain and when it's actually reported nationally. Usually

33:05

it's a nine or 10 month lag. There

33:08

are a handful of agencies, probably a couple

33:10

of dozen, maybe 100 agencies that

33:13

report their crime data directly on their website.

33:15

But there are 18,000 agencies nationwide. So

33:19

yes, it's great that 100 agencies maybe

33:21

do this, but that's a tiny sliver

33:23

of all of the agencies. The vast

33:25

majority of people that live in

33:27

the United States live in a place where it's sort of

33:29

a desert for crime stats that

33:31

are updated. It makes it really hard

33:33

for people to necessarily answer a question of, is

33:35

crime going up? You really have to go by

33:38

sort of the fields rather than the data. If

33:41

we just acknowledge that the public perceptions

33:43

of crime polling from Gallup were kind

33:45

of spotty in the 90s, I'm

33:48

just curious if we go back to that decade,

33:50

given the fact that in the 90s, we

33:52

had nearly 2 million violent crimes a year

33:55

compared to now when we're about 1.2 million,

33:58

were people saying similar things? looking

36:00

at? Between 2000 and 2020, when

36:02

Bush was in office, more Democrats and Republicans

36:07

thought crime was rising. When

36:09

Obama was in office, more Republicans

36:12

than Democrats thought crime was rising.

36:14

When Trump was in office, more Democrats

36:17

thought crime was rising. Under

36:20

the Bush years, when crime was reasonably down, a

36:22

majority of Democrats were saying that

36:25

crime was rising each year. Now over

36:27

the last few years, things have kind

36:29

of gotten broken. In 2023, the Gallup Survey

36:32

showed 91% of Republicans

36:34

saying that crime was rising in the

36:36

last year versus 58% of

36:39

Democrats, which is the highest percentage of

36:42

Democrats that have ever said that crime

36:44

was rising. Yeah, so there's kind of

36:46

two things going on. There's the fact

36:48

that right-wing concern for crime rates, is

36:51

that an all-time high? Yeah,

36:53

I mean not even close. Prior to 2020, it was under

36:56

60% every year. Democrats too

36:58

are also citing higher crime rates,

37:01

which flies in the face of

37:03

a trend where typically, you know,

37:05

their guy is in office and

37:08

therefore they would seem less concerned.

37:10

Why? What accounts

37:12

for these changes? Previously, partisanship

37:14

was a decent answer for this. Now

37:17

it's some sort of hyper partisanship,

37:19

broken media vacuum that's pumping wrong

37:22

information and misinformation into the system

37:24

and is leading to nine out

37:26

of ten Republicans saying that crime

37:28

is rising. I want

37:31

to ask you about how journalists and

37:33

news consumers can do a better job

37:35

of closing the gap between the anecdote-borne

37:38

perceptions and the data, you

37:40

know, our perceptions of crime

37:42

and the facts. So

37:44

readers, viewers, listeners, what should we keep

37:47

in mind or look out for when

37:49

we're consuming news about a crime? We

37:51

should keep in mind that crime

37:53

data is generally flawed. Do you

37:56

mean like the numbers are wrong

37:58

or that the numbers simply... lack

38:00

context like year to year or

38:02

decades long trends etc. Yes.

38:07

The numbers are wrong. They're always

38:09

wrong. They're always estimates. And they're

38:11

estimates because not every agency reports,

38:14

not every agency reports perfectly. And

38:17

there's no good way of adding up data from 18,000

38:19

agencies and saying there were 18,242

38:22

murders last year. We

38:25

just don't have that level of precision. So

38:28

we're always looking at estimates. That's very important

38:30

to understand. The second

38:32

thing is that if pickpocketing

38:34

in New Orleans surged in 2022, 50, 70, 80%

38:40

increase, that doesn't mean that

38:42

we have a rash of pickpockets, that we

38:44

have a serious pickpocket problem. It

38:46

just means that in 2021, we didn't have Mardi

38:48

Gras. And in 2022, we did have Mardi Gras.

38:51

And whenever there's Mardi Gras, you're going to have pickpockets. Looking

38:54

at it just year to date does not

38:56

tell you the story of what's going on.

38:58

You need a longer term view of what's

39:01

happening. And year to date sometimes

39:03

is the best that's available to us. But

39:06

frequently, if you were choosing what analysis

39:08

to do on data, you wouldn't

39:10

do year to date. You'd show rolling over

39:12

time. What's at stake

39:14

when such a

39:16

large portion of Americans have

39:20

unfounded concerns about

39:23

rise in crime? Only

39:26

from an electoral standpoint, it just

39:28

leads to fear and not

39:30

making choices based on reality. I

39:33

think from a policymaker standpoint, it means

39:35

that you're not necessarily making smart choices

39:38

throughout the criminal justice system, how you're

39:41

using your resources, the number of

39:43

officers you're hiring, how you're

39:46

approaching incarceration reform, how you're

39:48

approaching sentencing reform. And

39:50

just are people scared? Are they nervous? Are they

39:52

worried? Do they think things are getting better or

39:55

things are getting worse? I think in all facets

39:57

of American life, when things are getting worse, I think

39:59

that's better we should have

40:01

optimism. And I'm going to go to

40:03

the fields now. It feels like we've lost

40:05

some of that sense of optimism

40:07

at the things that are getting better while

40:09

still taking seriously those issues and taking seriously

40:11

the issues that are getting worse. Jeff,

40:14

thank you very much. My pleasure. Jeff

40:17

Asher is a writer and data

40:19

analyst. His latest article, Americans Are

40:21

Bad at Perceiving Crime Trends, is

40:24

available on his sub-stack, Jeffalytics.

40:28

Coming up, mixing journalism and

40:31

AI. What could go wrong? This

40:34

is On On The Media. This

40:42

is On The Media. I'm Michael Loehninger. This

40:47

week, two top executives were

40:49

fired from Sports Illustrated's publisher.

40:52

The news comes a little over

40:54

a week after the tech publication

40:56

Futurism noticed that something was off

40:59

with certain author profiles on the

41:01

Sports Illustrated site. Authors

41:03

like Drew Ortiz, who according to

41:06

a since-deleted bio, grew up in

41:08

a farmhouse. Or there's

41:10

Sora Tanaka, who loves to dry

41:12

different foods and drinks. The

41:15

problem is, both their photos were

41:17

repeatedly, reportedly found on

41:20

a website that sells AI-generated

41:22

headshots. In a recent

41:24

company-wide call, the majority stakeholder

41:26

reportedly told employees to quote,

41:28

stop doing dumb stuff. Sports

41:31

Illustrated has said the dismissals this

41:33

week were unrelated to the AI

41:35

scandal. The outlet is

41:37

one of several media companies that

41:39

have come under scrutiny for their

41:42

alleged or stated use of artificial

41:44

intelligence. In August, the

41:46

country's largest newspaper company, Gannett, rolled

41:48

out a new, non-generative AI service

41:50

that would provide automated high school

41:53

sports coverage in a number of

41:55

states. But readers quickly

41:58

discovered that bizarre phrase like

42:00

close encounters of the athletic kind,

42:02

had shown up in hundreds of

42:05

local news stories. Our client had

42:07

a PR problem on their hands.

42:09

Jay Allred is the CEO of

42:12

Source Media Projects, which includes Richland

42:14

Source, a local news organization in

42:16

Ohio. He's also the co-founder

42:18

of Lead AI, the company

42:21

that built the technology that Gannett

42:23

was using to automate some of

42:25

its coverage. Gannett put an indefinite

42:28

pause on the project of

42:30

reporting high school sports results using

42:33

AI with us. In

42:35

September, Jay agreed to speak to me

42:37

about what happened. His first extensive interview

42:39

since his deal with Gannett blew up.

42:42

He told me that his team began

42:44

building and using Lead AI in his

42:47

own newsroom at Richland Source a few

42:49

years ago, after they learned that they

42:51

could draw on high school sports results

42:53

from a service called Scorestream. Which

42:56

collects game results often recorded by

42:58

fans. So if we're looking at

43:00

a football game, we're trying to figure

43:02

out, was it a close game? Was

43:04

it in overtime? Was it a blowout?

43:06

Was it a come from behind win

43:08

in the fourth quarter? We've kind of

43:11

grouped those different outcomes into scenarios. And

43:13

then we're going to pull

43:15

from a library of pre-written

43:17

templates, plug those variables into

43:20

those pre-written templates for the

43:22

customer. The goal

43:24

was that you could basically

43:26

be offering, let's say, fairly

43:29

rudimentary coverage of high school sports all

43:32

across Ohio or wherever. That your writers

43:34

and editors wouldn't necessarily have to be

43:36

solely on the hook for producing. And

43:38

then they could go and do more

43:41

meaningful coverage. We're a small newsroom. There's

43:43

only 10 of us. And there's only

43:45

one full-time sports reporter. There's well over

43:47

20 high schools in our region. What

43:50

this lets us do is be able to

43:53

provide coverage to communities that we wouldn't have

43:55

been able to be at that game at

43:57

all. Our sports reporter coverage is a very

43:59

important part of our community. the A game

44:01

or the number one game will cover the

44:03

B and the C game with our two

44:05

other reporters. And then LeadAI will

44:07

be in to write the briefs for us for

44:09

those other three games. From that

44:12

standpoint, our editor that's on the desk

44:14

that night can call coaches, flesh out

44:16

that LeadAI story, combining the

44:19

technology that LeadAI provides with

44:21

the actual journalism our newsroom

44:24

provides. And how do you

44:26

communicate to readers that what they're reading was not

44:28

written by a human? Every

44:30

single story that publishes on Richland Source

44:32

has an author and that author is

44:34

called Auto News Desk. And

44:36

if you click on Auto News Desk, it identifies

44:39

itself as an AI tool right out of the

44:41

gate. At the bottom of

44:43

the article, we are disclosing that it's an

44:45

AI tool that we're using. We're actually linking

44:47

to LeadAI's website. We

44:49

have a feedback form that publishes with every

44:52

piece of content that we publish. How do

44:54

people react? In general, the

44:56

readers understand it's information, it's not

44:59

journalism. Of course, a lot

45:01

of times readers want the content to be

45:03

longer and then to include player names and

45:05

photos and video. They want it to be

45:07

a reported article. Exactly.

45:10

How exactly do

45:12

you attempt to make

45:14

LeadAI produce human

45:16

sounding articles? I mean,

45:18

I know that with some of these

45:21

large language models, they require lots of

45:23

data. And this has

45:25

led to a lot of controversy around

45:27

AI startups scraping enormous parts of the

45:29

web, including books, entire news outlets, entire

45:32

forums like Reddit. So explain to

45:34

me how you feed language and

45:37

templates to LeadAI. Every

45:39

single word, every comma, every semicolon

45:41

in our database has been written

45:43

by a person. And then it's

45:45

been checked by another person and

45:47

checked by a person after that.

45:50

It's what allows us to be confident

45:52

in all cases that if

45:55

we're using our standard data set

45:57

that the content that we're producing

46:00

is accurate as long as the data is

46:02

accurate and it's very accurate. Okay,

46:04

that's interesting because in late August,

46:06

people on social media began posting

46:09

some of the really awkward phrases

46:11

that LeadAI has put into some

46:13

local news sources. The one that

46:15

caught a lot of attention on

46:17

Twitter, for instance, was

46:19

a piece in the Columbus Dispatch

46:21

and some other Gannett-owned papers. Readers

46:24

finding examples of LeadAI using

46:27

phrases like, quote-unquote, close encounters

46:29

of the athletic kind. There

46:31

were a lot of articles referring to

46:33

high school sports action or

46:36

how one team, quote, took victory

46:38

away from another team. You

46:41

know, these are phrases that most human

46:43

journalists would consider ranging from awkward to

46:45

poor writing. So how did that happen?

46:47

I knew you were going to go there and I'm glad you did. In

46:50

mid August, our technology powered a really

46:53

big launch with Gannett across, I

46:56

think, six or seven major markets

46:58

in the US. We

47:00

had written some custom code for

47:03

that particular customer and the code

47:05

had bugs in it, Micah. Some

47:08

of those things that showed up in those

47:11

Gannett articles, especially the

47:13

errors, were the result of

47:15

a small company working

47:18

really, really hard to get ready for a

47:20

launch with a very big company. As

47:23

far as the awkwardness of the phrasing

47:25

and the now infamous close encounters of

47:27

the athletic kind, a human being wrote

47:29

that, Micah. A person

47:33

wrote that and we got called out

47:36

on a few phrases and

47:38

they are no longer in our database. It

47:41

was as simple as taking them out. I

47:44

was curious to know if this was

47:47

a feature or a bug. I actually just

47:49

searched some of these phrases on the Richland

47:51

source. I counted over

47:53

140 articles on the

47:55

Richland source from this year that featured

47:57

the phrase close encounters of the athletic.

47:59

athletic kind or similar phrases, including

48:02

50 articles from

48:04

this year that featured the phrase close

48:06

encounter of the winning kind in the headline.

48:08

So I don't really buy that it

48:10

was just a fluke that

48:12

happened with launching a new service with Gannett.

48:15

Like you have been publishing these sentences

48:17

for years. No, and I

48:19

appreciate you calling that out because that phrase

48:21

has been in our code for years. The

48:25

things that were unique to the Gannett

48:27

launch were some other great, he says

48:29

sarcastically, some other unfortunate stuff.

48:32

For example, there were a couple of

48:34

leads that published in some of the

48:36

papers where we had plugged

48:39

in a variable where there should

48:41

have been a mascot name. There

48:43

were instances where we

48:45

published two very similar

48:47

lead paragraphs. They

48:50

said exactly the same thing in terms

48:52

of factual information, but they said it's

48:54

slightly differently. And those were

48:56

bugs that were built into that custom code.

48:58

But those awkward phrases that the internet called

49:00

out, that's been there for years. I guess,

49:02

but see, I guess this is what

49:06

sort of sends a shiver down the

49:08

spine of media critics and

49:10

journalists and editors, because we're talking

49:12

about high school sports. Like this

49:14

is not the highest stakes beat

49:16

in all of journalism, but

49:19

it seems like it does speak

49:21

to the risk of automation, where

49:23

one small mistake when

49:25

automated becomes 150 small mistakes

49:28

all across the country. Yes, absolutely.

49:31

What if this had been crime reporting? What

49:34

if these had been arrest reports? Real

49:36

harm could have been done as leaders in

49:39

the industry. I think it should give us

49:41

all pause. It's why

49:43

I'm having this conversation with you. And I

49:45

appreciate that. I appreciate your vulnerability and your

49:48

openness to introspection. Are

49:51

you at all concerned that local newsrooms

49:53

would see the promise of

49:56

lead AI? Maybe think that it's capable of

49:58

doing more than it is. and

50:00

kill entry-level jobs. I think about

50:03

that every single day. In

50:06

three years of talking to news

50:08

leaders around the country, I've

50:10

never once heard one of them say, I'm

50:12

super excited for AI because I get to

50:14

reduce my head count. Well, no one says

50:16

those things. They say, we would like to

50:18

be more efficient. I agree with you. And

50:21

with all of those things said, I

50:24

still lose sleep over it at night. What

50:26

do you lose sleep over? Is the

50:28

intention to use your euphemism to

50:31

find efficiency and to do that

50:33

through less people? Or is

50:35

the intention to create more value for

50:37

consumers so that we can get the

50:40

nose of this airplane pointed up and

50:43

we can start to create a future

50:45

where local news entrepreneurs can think of

50:47

local news as a good small business?

50:50

And I think that there's lessons to be learned

50:52

here and we can grow as an industry and get

50:54

better. Because the reality is this

50:56

stuff is, it's not coming, it's here. That's

50:59

what I've heard in some of your answers. There's still this

51:02

implicit belief that the kind of rushing river

51:04

of technology is coming no matter what. And

51:06

I wonder if this is a moment in

51:09

time to say, there might be some uses

51:11

for AI, we don't

51:13

just have to see it to its logical

51:15

conclusion just because technology is great, bro. Yeah,

51:17

I agree with you. I think we

51:19

should use tech like

51:22

LeadAI to report unreported

51:24

stories that would never go reported

51:27

otherwise. We should interrogate

51:30

that technology vigorously and make sure that

51:32

it can be trusted and be accurate.

51:35

And I know that there are ways to do that. You've

51:38

been forthcoming about the mistakes your team

51:40

made and the limitations of the technology.

51:42

Do you feel that any of the

51:44

backlash to AI within the media has

51:46

been unfair? Like has any of it

51:48

you think kind of missed the mark?

51:51

I think that our

51:53

industry has a tendency

51:55

to respond to stuff

51:57

like AI from a very defensive. position.

52:00

It's super understandable. Our

52:03

industry has done nothing but cut

52:05

newsrooms to the bone for going

52:08

on two decades now. I

52:10

wish we could get into spaces where

52:13

we understood that we were more all

52:15

in this together and that

52:17

we are trying to figure it out. I

52:19

think we as an industry need to be able to hold

52:22

multiple things to be true at the same

52:24

time, which is the level

52:27

and deployment of AI inside of our

52:29

industry is going to hurt our industry.

52:32

Intentional, thoughtful deployment for the benefit

52:34

of readers and communities and reporters

52:37

can benefit our industry. Both

52:39

things might happen. I hope it's the second. That's

52:41

going to be the work I continue to do.

52:44

Jay, thank you very much. Thank you,

52:46

Micah. I was glad to be invited onto your

52:48

program. Jay

52:50

Allred is the CEO of Source

52:53

Media Properties. That's

53:06

it for this week's show. On

53:08

the Media is produced by Eloise

53:10

Blondio, Molly Rosen, Rebecca Clark Callender

53:12

and Candice Wong with help from

53:14

Sean Merchant. Our technical director is

53:17

Jennifer Munson. Our engineer this week

53:19

was Brendan Dalton. Katya Rogers is

53:21

our executive producer. On the Media

53:23

is a production of WNYC Studios.

53:26

Brooke will be back next week.

53:29

I'm Micah Loehringer. everywhere.

54:00

Because businesses that grow, grow

54:03

with Shopify. Sign up for a $1 per month

54:06

trial period at shopify.com/trial. Go

54:08

to shopify.com/trial now to grow

54:10

your business no matter what

54:12

stage you're in.