Monthly Archives: November 2017

Dak Prescott Should Start Ignoring Dez Bryant

When Cowboys’ running back Ezekiel Elliott was finally forced to serve his six-game suspension for domestic violence, the logical assumption was that quarterback Dak Prescott would lean more heavily on the third member of the Dallas troika, star receiver Dez Bryant. But the Cowboys offense has floundered in the ensuing three weeks, and one of the reasons is becoming obvious: Bryant has fallen into an abyss so deep that he’s not merely no longer great, he has actually become one of the least efficient wide receivers in football.

The most disconcerting thing about Bryant’s precipitous drop in production is that it’s not because he’s not getting opportunities. The Cowboys and NFL sophomore Prescott are throwing Bryant the ball — peppering him with 103 targets, the sixth-highest number in the league according to ESPN TruMedia.

This high volume of looks would make good sense if Bryant were the offensive force we last saw for extended stretches way back in 2014, when Tony Romo was his quarterback. But the former All-Pro wide receiver has not been the same since he broke his foot in 2015. His horrible inefficiency that year was attributed to his coming back too soon and then playing without Romo, who was injured, and instead getting targets from one of the most inept casts of backup quarterbacks ever assembled. His 2016 season was supposed to be an awakening. But even with hyper-efficient Prescott at the controls, Bryant was nowhere near peak form. He hauled in 52.1 percent of his targets compared with 62.6 percent through 2014. And he converted just 57 percent of all the air yards on passes thrown to him into actual receiving yards, versus 74.6 percent prior to his injury.

Still, Bryant flashed enough brilliance late in the year and in the postseason to lead Cowboys fans to think they were heading into this season with a legitimate weapon. Instead, Bryant is among the lowest-rated receivers in key efficiency metrics like catch rate (73rd out of 78 qualifying wide receivers with at least two catches per team game), receiving yards per target (75th), and receiving yards as a percentage of air yards (72nd).According to data from ESPN TruMedia.

“>1 Here’s every wide receiver with 22 or more receptions this season, broken up by targets and receiving yards.

Most damningly, Prescott’s passer rating collapses when he targets Bryant — the opposite of what is supposed to happen with a No. 1 wide receiver. When throwing to Bryant, Prescott is 53-for-103 for 578 yards, with four TDs and three picks. That’s a 69.2 rating, which would rank 35th in the NFL, between C.J. Beathard and Tom Savage.

In Elliott’s absence, things have gotten even worse. Prescott has a 54.4 rating when throwing to Bryant, and Dallas has scored just 22 combined points in three straight losses — the first team to score fewer than 10 points in each of three straight games since the 2009 Browns.

Bryant, an elite deep-ball weapon in his prime, is now completely useless downfield, with just one reception on a pass thrown over 20 yards from scrimmage all year. Between 2012 and 2014, he averaged nearly 10 such grabs per season (29 total). It’s not like he’s wide open on short passes either. That was never more apparent than it was on Thanksgiving, when Bryant’s 0.8 yards of separation on targets was lowest among all wideouts who got at least five targets, according to the NFL’s Next Gen Stats.

While Bryant just turned 29 in November, he has not only endured injury but also taken constant punishment as one of the game’s most physical receivers. He has been one of the league’s most vocal receivers about getting his share of volume, so if Prescott does start looking to his other targets more frequently, there’s no doubt that he will hear about. But it may be worth it: On all non-Bryant passes, Prescott’s rating of 95.3 is well above the NFL average.

Who Are The NBA’s Best Defenders Right Now?

For the past few seasons, the best defenders in the NBA have been some combination of Draymond Green, Rudy Gobert and Kawhi Leonard. But with Leonard and Gobert out for all or much of the season so far and Green’s production falling off just a bit, we’ve got some new blood in contention to be the NBA’s best defender.

To put a number to this, we’ll use data from Second Spectrum that shows us both the total number of shots defended by a player and how much he affected those shots. There are a few ways to get at this, but we’re going to use “quantified shot probability” (qSP) — which determines the expected value of shots defended using shot distance, shot location and defender position, among other variables — and “quantified shot making” (qSM) — which subtracts the expected effective field goal percentage on those shots from the actual eFG to determine how much of an effect the defender had. In other words, does the defender make shots worse? And if so, by how much?

We’ve used metrics like this in the past, but the rub with this set is it adjusts for who’s taking the shot, meaning weak defenders who guard bad shooters don’t get as much credit, and defenders tasked with guarding superstars aren’t punished for their assignment — the star’s excellence is baked into the stat.

The per possession numbers will look a little different from the overall effect — players like Tarik Black and Marreese Speights have defended a lot of shots very well in short minutes — but this chart should demonstrate that there is a wide spread on defender effect. What we’re missing here, though, is how many shots a defender actually affects.

Here we’ve got total shots as well as the qSM (the difference a defender makes on each shot) plotted against each other to show who’s affecting how many shots to what degree. This isn’t perfect. For one, a great defender doesn’t simply challenge shots — he denies them from happening, forcing a team to reset its offense. But by the same token, great defenders also work their way back into plays, affecting shots by playing smart help defense and covering acres of ground. Another thing these numbers don’t reflect the is overall quality of the shot — there are a few defenders who don’t depress value of each individual shot by as much as others but who force opponents into low-quality shots in general (through hard work and smart positioning) so that the overall effect is the same.

But despite the limitations, these metrics are a pretty good way to look at who’s having the best defensive season. A few players stand out on the list:

 

Anthony Davis, New Orleans Pelicans

From the time he was a rookie, Anthony Davis has had all the tools. He’d block shots and steal alley-oops and tug on his shorts and check the opposing point guard 30 feet from the basket. He came into the league as a destructive defensive force, but for reasons ranging from injury to scheme to sheer offensive burden, Davis’s aggregate effect on the defensive end hasn’t always lived up to the promise of those moments. This season, it has. Davis ranks seventh in the league in shots defended overall, and the defense gets 12.7 points per 100 possessions worse when he sits down. That’s nearly double his defensive on/off split from last season, which was already by far the best of his career.

Partly this is because of his partnership with DeMarcus Cousins. Defensive intensity hasn’t always been Cousins’s strong point, and that’s still true today, but he has defended the most shots in the league, and done so while holding opponents to a middle-of-the-road 51.4 eFG on shots he defends. Not great, certainly, but it’s a respectable backstop for the defense.

Because of Cousins’s role as a reliable constant, Davis is free to cover more ground without making risky moves to get back into plays. And because of that, his personal foul rate is at an all-time low (2.6 fouls per 100 possessions). This allows Davis to do what he does best — stick to his man, rotate to help in the paint without overextending, and blitz spot-up shooters on the perimeter faster than anyone else his size in the league.

 

Kevin Durant, Golden State Warriors

The Warriors are unfair. Draymond Green — for my money the best defensive player in the league — has for whatever reason been a little off during the first quarter of the season, possibly because the Warriors can sleepwalk to a bare-minimum 2-seed. He’s not been bad, mind you, just not quite as dominant as usual. That’s left quite a lot of slack in the Golden State scheme, and Durant has picked up more than his fair share of it. Golden State ranks eighth in defensive rating, down from second a season ago, but that almost feels like a threat more than a falloff given how flat the team has looked at times. Beware, it says, this is a top 10 defense even while almost completely half-assing things, just on the virtue of Kevin Durant showing up to work.

Durant has been improving as a defender for years, going back to his time with the Thunder. (If you ever go back and watch the 2016 series between OKC and San Antonio, watch Durant on defense — he was the best defensive big man in that series.) His style is also entirely his own: KD gives up relatively “great” shots — ones his opponents would be expected to turn into 52.3 eFG against an average defender, or about what Austin Rivers gives you as a defender. Not great. But because Durant is so long, so mobile and so smart positionally, the actual shots against him fall at a rate of just 45.2 eFG. This still happens within the overall Golden State system — Andre Iguodala and Steph Curry have similar opponent eFG numbers, albeit on fewer shots defended — but Durant’s role has him not only locking down his own man but also covering up for lapses by his teammates.

Green may remain the more important defender for the Warriors — he’s the anchor, and what he does from his position is impossible to replicate. But Durant’s play so far has been the strongest on a Golden State defense that should end up being a top-3-type unit, at minimum.

 

Marcus Smart, Jaylen Brown and Al Horford, Boston Celtics

The Celtic defense has continued to crush opponents. Boston is 18-4, with a league-leading defensive rating (100), the second-lowest effective field goal percentage allowed (48.6) and a roster full of live-bodied defenders who can switch practically any assignment. Even Kyrie Irving is chipping in on the effort. But while every Celtic is doing his part, some parts are more critical than others. It’s undeniably been the play from Smart, Brown and Horford that has carried the Celts so far.

Smart does not possess the most graceful set of skills, but what he provides is unmistakable to anyone watching. He’s all over the floor, hounding ball handlers, bodying up on bigger players in the post and bumping cutting players he isn’t even guarding, like a middle linebacker jamming a slot receiver.

Brown is a little different. While the Boston defense is built on interchangeable pieces switching and denying an offense space, it also needs players who can stick to their man across multiple screens, then square up and check their man as he plants and drives. Brown has stepped into that role and dominated so far. He can fight over screens (or just outright avoid them) to allow the defense to keep its shape, and he has the quicks to shut down first steps with the length to challenge pull-up attempts.

Horford, meanwhile, does not have nearly the same reputation as the other two, but this season, he looks livelier than he has in the past when making switches, and he has defended the most shots of any Celtic. The newfound agility in space is especially important — because the defense will semi-regularly ask him to survive on an island against a wing, but also because his role at the center of the defense requires him to scuttle shooters who’ve just run over two or three screens trying to escape Smart or Brown.

Honorable mention

Giannis Antetokounmpo and Joel Embiid have both been outstanding through the first quarter of the season, but they don’t have quite the same track record as some of the players listed above. Just the same, neither seems to be a fluke. We’ll be a little more certain how permanent their new defensive excellence is by the All-Star break. The same goes for Kristaps Porzingis, who last season was in the tier just behind Gobert and Green for overall defensive value, but this year fell off early before making up some ground in the past few weeks.

Some others seem more like early-season mirages — Josh Richardson, Gary Harris and Eric Gordon are all having excessively good years by the metrics but are playing way above the levels they had in previous seasons.

Politics Podcast: Congress’s Broken Culture

FiveThirtyEight

 

On Monday, lawmakers returned to Washington and immediately confronted questions about how they’ve handled sexual harassment in their own ranks. Amelia Thomson-DeVeaux joins the FiveThirtyEight Politics podcast to talk about why the workplace culture in Congress is conducive to misconduct and underreporting.

The team also discusses the latest polling and the unknowns in Alabama’s upcoming Senate election.


You can listen to the episode by clicking the “play” button above or by downloading it in iTunes, the ESPN App or your favorite podcast platform. If you are new to podcasts, learn how to listen.

The FiveThirtyEight Politics podcast publishes Monday evenings, with occasional special episodes throughout the week. Help new listeners discover the show by leaving us a rating and review on iTunes. Have a comment, question or suggestion for “good polling vs. bad polling”? Get in touch by email, on Twitter or in the comments.

How The Thunder Can Become A Dream Team

By Neil Paine, Chris Herring and Kyle Wagner, Neil Paine, Chris Herring and Kyle Wagner and Neil Paine, Chris Herring and Kyle Wagner

 

Welcome to The Lab, FiveThirtyEight’s basketball podcast. On this week’s show (Nov. 22, 2017), Neil, Chris and Kyle take a look at how the recent injury to Denver Nuggets forward Paul Millsap will affect the team. Millsap will reportedly have surgery on his wrist and could miss three months. Next, the crew discusses why the Thunder haven’t clicked yet. After bringing Paul George and Carmelo Anthony to Oklahoma City last summer, the Thunder were anticipated to be a contender in the West. But it hasn’t all come together so far. We investigate what’s going right, what’s going wrong and how they might turn the season around. Plus, a small-sample-size segment on Lonzo Ball.

Here are links to what we discussed this week:

  • Keep an eye on our 2017-18 NBA predictions, updated after every game.
  • Chris Herring wrote that the Thunder aren’t far from being good.
  • ESPN’s Royce Young took a look at the still-developing chemistry between Russell Westbrook and Paul George.

This Could Be The Biggest Alabama-Auburn Matchup Ever

Week 13 in college football — aka rivalry week — promises to be one of the most chaotic of the year, with dozens of teams playing hatred-filled games that will have major College Football Playoff implications. No. 2 Clemson travels to face in-state rival South Carolina; No. 7 Georgia hits the road for its annual meeting with Georgia Tech; and No. 9 Ohio State heads to the Big House to face “That Team Up North.” But of all those high-intensity games, it’s the one between Alabama and Auburn that has the most riding on it.

The Iron Bowl has given fans some of the most spectacular college football moments in recent memory, including “The Drive” in 2009, when Greg McElroy drove Alabama back to a 26-21 win at Jordan-Hare Stadium via a touchdown with less than 90 seconds left. Or “The Camback” in 2010, when Auburn, led by Cam Newton, stormed back from 24 points down in Tuscaloosa to stay undefeated (they’d eventually go on to win the National Championship). And most famously, there was the “Kick Six” in 2013, when Alabama took a 57-yard field goal attempt with one second remaining, but the ball fell short, and Auburn’s Chris Davis ran it back 109 yards to win the game.

But this Saturday’s Iron Bowl could top all of those others, given its importance and the sheer quality of this year’s Tide and Tigers squads. For starters, these two teams are the best they’ve been in decades. We can measure how strong a given matchup is using our Elo ratings, which assign each team a score based on their quality at any given moment. And currently, Alabama boasts the No. 1 Elo in the country (what else is new?) while Auburn ranks sixth. If we combine those Elo ratings using their harmonic mean — which allows us to look for matchups where both teams have high ratings — we find that this year’s Iron Bowl features the greatest combination of Alabama and Auburn squads since at least 1988Our Elo Ratings for college football only go back that far.

“>1:

On paper, this year’s Iron Bowl looks like the best in years

Highest combined Elo ratings (according to their harmonic mean) for Alabama-Auburn matchups, 1988-2017

SCORE ELO RATING
RK SEASON ALABAMA AUBURN ALABAMA AUBURN COMBINED
1 2017 ? ? 2374 2161 2262
2 2013 28 ✓34 2349 2096 2215
3 2016 ✓30 12 2468 1990 2204
4 2010 27 ✓28 2156 2210 2182
5 2014 ✓55 44 2300 1975 2125
6 1994 ✓21 14 2089 2108 2098
7 2015 ✓29 13 2390 1764 2030
8 2011 ✓42 14 2187 1864 2013
9 1989 20 ✓30 2051 1945 1996
10 1993 14 ✓22 1981 2004 1992

And the stakes of this particular Iron Bowl could scarcely be higher. Not only will the winner go to the SEC title game as West division champ,Either Alabama will have an 8-0 conference record or the teams will be tied at 7-1 in conference games and Auburn would have a head-to-head tiebreaker over the Crimson Tide.

‘>2 but the outcome will also have a big influence on who ultimately makes the College Football Playoff.

According to the FiveThirtyEight CFP prediction model, Alabama and Auburn are two of the eight remaining teams in the country with at least a 20 percent probability of making the CFP. If Bama wins, that would bump the Tide’s playoff odds from 67 percent to 84 percent; if they lose, that number falls to 42 percent (meaning ’Bama might well miss the CFP for the first time). The stakes are even higher for Auburn: If the Tigers prevail, their odds will rise from 21 percent to 52 percent, but a loss would effectively eliminate them from the CFP race. And the ripple effects extend beyond the SEC, as seven other teams figure to see their playoff probabilities shift by at least 1.5 percentage points depending on what happens on the Plains.

How the Iron Bowl could change the playoff picture

Swings in each team’s chances of making the College Football Playoff depending on who wins the Alabama-Auburn game

CHANGE IN CFP CHANCES IF …
TEAM CURRENT CFP CHANCE BAMA WINS AUB WINS CFP SWING
Auburn 21.0% -20.6 +31.0 +/-24.7
Alabama 66.9 +16.6 -25.1 20.0
Ohio State 27.8 +3.2 -4.8 3.8
Georgia 32.2 -1.9 +2.8 2.2
TCU 11.0 +1.8 -2.7 2.1
Clemson 70.0 -1.7 +2.6 2.1
USC 9.5 +1.6 -2.4 1.9
Miami 51.1 -1.6 +2.4 1.9
Oklahoma 58.1 +1.2 -1.9 1.5
Notre Dame 1.1 +0.4 -0.7 0.5

As of Nov. 20, 2017. Includes teams where the average swing is at least +/- 0.5 percentage points

So with 62.4 percentage points of total CFP probability on the line — in terms of how much the game projects to swing our model’s odds across every team in the country — Alabama-Auburn is easily the most important game remaining in the regular season:

The most important games of Week 13

According to total swing in all teams’ likelihood of making the College Football Playoff based on the game’s outcome

HOME TEAM AWAY TEAM
DAY SCHOOL ELO RATING WIN PROB. SCHOOL ELO RATING TOTAL SWING
Sat Alabama 2374 60.1% Auburn 2161 +/-62.4
Sat Michigan 1872 25.7 Ohio State 1992 25.5
Sat Georgia 2122 76.3 Georgia Tech 1654 19.4
Sat Oklahoma 2177 87.3 West Virginia 1734 16.5
Fri Miami 2216 84.4 Pittsburgh 1500 14.8
Sat Clemson 2192 81.4 South Carolina 1891 14.5
Sat Minnesota 1565 13.2 Wisconsin 2165 12.0
Sat Notre Dame 1997 55.7 Stanford 2004 10.4
Sat Washington 1961 76.5 Washington St. 1912 8.5
Sat Florida 1661 42.5 Florida State 1674 4.0

Elo ratings are scaled so that an average team has a rating of 1500.

In terms of ridiculous, heart-stopping finishes, it would be hard for Saturday’s game to top the 2013 edition, which ranked second in terms of combined Elo ratings, so don’t hold your breath for another miraculous kick six. But even though we don’t know how it will end, this Iron Bowl is set up to be special. The teams are at their best, the winner will likely go to the College Football Playoff while the loser will likely stay home, and the game’s outcome could help throw the rest of the country into chaos. The Alabama-Auburn rivalry is legendary and the game would still be tense even if both teams were 2-9 — but with the road to the national championship passing squarely through Jordan-Hare Stadium on Saturday, this season’s edition should be one for the ages.

Politics Podcast: Alabama Is Tied

FiveThirtyEight

 

The Alabama special Senate election is a dead heat after more women accused Republican candidate Roy Moore of sexual misconduct. The FiveThirtyEight Politics podcast team assesses attitudes among different voting groups in Alabama and discusses the candidates’ strategies thus far. The crew also looks at the pros and cons for Republicans of passing a tax reform bill.


You can listen to the episode by clicking the “play” button above or by downloading it in iTunes, the ESPN App or your favorite podcast platform. If you are new to podcasts, learn how to listen.

The FiveThirtyEight Politics podcast publishes Monday evenings, with occasional special episodes throughout the week. Help new listeners discover the show by leaving us a rating and review on iTunes. Have a comment, question or suggestion for “good polling vs. bad polling”? Get in touch by email, on Twitter or in the comments.

 

Democrats Missed A Chance To Draw A Line In The Sand On Sexual Misconduct

At about 11:15 this morning, an hour or so after Leeann Tweeden published an allegation that Democratic Sen. Al Franken of Minnesota had groped and kissed her without her consent in 2006, I assumed that Franken was headed toward resignation. I didn’t necessarily expect Franken to resign immediately or without putting up a fight. But barring some highly exculpatory evidence, I expected Senate Minority Leader Chuck Schumer and other prominent Democrats to be pushing Franken out the door.

Here’s why I thought that. First, the timing. The accusations against Franken came in the midst of a major scandal involving Roy Moore, the Republican nominee for Senate in Alabama, who has been accused of sexual misconduct toward multiple girls and young women. And it comes on the heels of scandals involving sexual assault or sexual harassment by some of the biggest names in Hollywood and the media business: Harvey Weinstein, Roger Ailes, Kevin Spacey and Louis C.K., to name some of many examples. It also comes about a year after Donald Trump was elected president even though he was accused of sexual misconduct by many women and was caught on tape bragging about grabbing women by their genitals. The conduct Franken is accused of is just the sort of behavior that he has condemned, potentially making he and other Democrats look hypocritical.

Second, there was the photograph that Tweeden published with her article. It appeared to show Franken groping Tweeden’s breasts while she was sleeping — not providing a lot of room for “if true” statements about Franken’s conduct.

And third, there was political expediency. If Franken were to resign, it probably wouldn’t cost Democrats a Senate seat. Instead, an interim replacement would be named by Minnesota Gov. Mark Dayton — a Democrat who would almost certainly appoint another Democrat. Then, a special election would be held next year to elect someone to serve the final two years of Franken’s term, which expires after the 2020 election. Next year’s midterms are likely to be blue-leaning (perhaps even a Democratic wave election), and Democrats are likely to hold Senate seats in states as blue as Minnesota under those circumstances. And Democrats have a deep and relatively diverse bench in Minnesota, with plausible candidates including State Auditor Rebecca Otto, Attorney General Lori Swanson, Lt. Gov. Tina Smith, U.S. Reps. Keith Ellison, Tim Walz and Collin Peterson, former Minneapolis Mayor R.T. Rybak, St. Paul Mayor Chris Coleman and others.Several of these candidates are running for governor in 2018, but some would probably be happy to switch to the Senate race because of the crowded gubernatorial field.

“>1

In other words, I thought the Democrats had an opportunity to maintain the moral high ground without having to pay a political price for it. They could keep the pressure up on Moore, who has put Republicans in a no-win situation in Alabama. And they could help to establish a precedent wherein severe instances of sexual harassment warrant resignation. In the long run, that might create more of a problem for Republicans than for Democrats, because the overwhelming majority of sexual harassment is conducted by men, and there are 265 Republican men in Congress compared with 164 Democratic ones.That includes the two independent senators who caucus with the Democrats.

“>2

Instead, Democrats basically punted on the question. Here’s what Schumer said, which echoes the statements made by many other Democrats:

Almost all of these comments said that sexual harassment must be taken very, very seriously. But the remedy they propose for Franken — referring the allegations to the Senate ethics committee, a step that Republican leader Mitch McConnell, White House press secretary Sarah Huckabee Sanders and Franken himself have also called for — isn’t particularly serious. Unless, that is, the committee process led to Franken’s expulsion. But there have been many ethics investigations and very few expulsions — none since 1862 — and none of the statements made by Schumer or the other leaders raised the possibility of expulsion.

Moreover, it’s not quite clear what behavior the ethics committee would actually be investigating: Franken hasn’t really denied Tweeden’s claim that he kissed her without her consent, and there’s already photographic evidence that appears to show he groped her. It’s possible the investigation could turn up evidence of similar incidents involving Franken and other women. But if Franken is a repeat offender — as so many sexual harassers are — that’s all the more reason for Democrats to want him out of office now instead of dragging the party through the mud.

Of course, what might be politically expedient for Democrats isn’t necessarily expedient for Schumer — or for McConnell, or for the White House, all of whom may be acting out of a sense of institutional self-preservation. If there’s a precedent that sexual harassment is grounds for removal or resignation from office, then a lot of members of Congress — including some of Schumer’s colleagues and friends — could have to resign once more allegations come to light, as they almost certainly will. President Trump’s conduct could also come under renewed scrutiny, as could the conduct of former presidents Bill Clinton and George H.W. Bush. Politics is a male-dominated institution, and a conservativeConservative meaning change-resistant, not politically right-of-center.

“>3 institution, and conservative, male-dominated institutions have pretty much no interest in flipping over the sexual harassment rock and seeing what comes crawling out from underneath it.

When we were thinking through the Franken story in FiveThrityEight’s internal Slack channel today, most of the men in our office thought that Franken was in deep trouble (“I think he’s toast,” I wrote at 11:07 this morning). Most of the women thought he’d hang in and survive. We’re less than a day into the story, but no surprise — it looks like the women will be right.

The FBI’s Explanation For Why It Released Less Crime Data Doesn’t Add Up

Late last month, FiveThirtyEight published an article that noted that the FBI’s most recent accounting of crime data in the United States was missing almost 70 percent of the data tables that had been included in past editions. The FBI has since disputed that the removal of those tables was out of the ordinary. But closer scrutiny doesn’t seem to bear this claim out.

In our earlier story, FiveThirtyEight reported that a large number of year-to-year data tables that typically appear in the FBI’s Crime in the United States Report had been removed in the most recent edition, which covers data from 2016 and is the first report of its kind published during the Trump administration. The yearly report is considered the gold standard of crime-trend tracking and is used by law enforcement, researchers, journalists and the general public. Changes to the structure of the report typically go through a body called the Advisory Policy Board (APB), which is responsible for managing and reviewing operational issues for a number of FBI programs. But this change was not reviewed by the APB. One former FBI employee told FiveThirtyEight the decision not to consult with the APB was “shocking.”

Following our story’s publication, the FBI informed FiveThirtyEight that it took issue with what Department of Justice spokesman Wyn Hornbuckle called “a false narrative.” The agency also publicly posted a statement on the data tables in the report, titled “Understanding Changes in ‘Crime in the United States, 2016.’” The statement, which was published about a month after the data was initially released, noted that the FBI “has been working on what has become the UCR-Technical Refresh since 2010” and that “throughout the planning of this project, it has been the intention of the UCR program to streamline the publications, including Crime in the United States (CIUS), and to reduce the number of data tables in the reports.”

But while the FBI says the plan to remove data tables had been in place since 2010, state-level UCR managers do not seem to have been informed of it until late 2016. “State UCR Program managers were advised of the FBI’s plan in the fall of 2016 through individual teleconferences,” Hornbuckle said in an email.

While it’s true that the FBI is on record saying it hoped to improve the report, the FBI had not publicly included the removal of data tables as part of those improvements until the statement it released following the FiveThirtyEight story. Instead, the FBI’s past statements said the agency aimed only to make data available more quickly and to improve digital features to allow users to access more data more easily.

The UCR-Technical Refresh page that the FBI links to in its statement asserting that the intention of the project has always included reducing the number of data tables, appears to have been created only relatively recently. A search through the Internet Archive fails to retrieve a cache of the page, suggesting a relatively new URL.

What does appear to have been referenced as early as 2010 is what was called “the UCR Redevelopment Project.” In 2010, the FBI described the goals of the UCR Redevelopment Project as: “decrease the time it takes to analyze data,” “reduce … the exchange of printed materials,” “provide an enhanced external data query tool,” and “decrease the time needed to release and publish crime data.”

In essence, the UCR Redevelopment Project appeared to largely be concerned with improving the process by which agencies submit data to the FBI, along with creating a tool for analyzing the data. This intended direction can be seen in project updates from March 2013 and December 2013.

In September 2016 the project became the “New UCR Project,” which had the stated goal “to manage the acquisition, development, and integration of a new and improved data collection system.” Sometime between September and October 2017, the New UCR Project became the UCR-Technical Refresh, though the overarching goals seem identical to its previous iterations.

While it is possible that reducing the number of available data tables was always a goal of the UCR Redevelopment Project, we were unable to identify any evidence of this objective in any publicly available publications or presentations on the subject. On the contrary, this project appears designed to improve the speed and efficiency with which law enforcement data can be submitted, processed and provided to the public.

A digital presentation dated April of 2012 describes the technical changes to the project with no mention of removing data tables. The earliest webpage reference to the project appears to be from March 2013, and it largely details technical upgrades intended to improve reporting from local law enforcement agencies to the FBI. The Internet Archive’s last capture of the site came as recently as May of 2016. We reviewed the documents and found no mentions of removing data tables in any of these iterations.

The April 2016 update lays out the plan in more detail. Of note is the last bullet point, which says, “We will continue to keep the APB and the broader UCR Program stakeholders apprised of our status as we work toward successful implementation of the new UCR System.” The APB was ultimately not involved in the decision to remove data tables from the 2016 report, but rather the tables were removed after consultation with the FBI’s Office of Public Affairs.

In a capture of the page that includes an update from February of 2017, the project is noted to have moved from development to implementation. One of its stated goals is to “provide a streamlined publication process that will give users quicker access to the data,” though it again makes no reference to data table removal.

When the New UCR Project became what is now called the UCR-Technical Refresh this fall, The FBI published a page devoted to the Technical Refresh that has very few details of the plan. But there is a change to wording that is striking.

One bullet-pointed goal of the project that once read, “Provide a streamlined publication process that will give users quicker access to the data,” now reads as two bullet points: “Provide a streamlined publication process” and “Provide users swifter access to the data.”

That may seem insignificant, but that grammatical sleight of hand could mean that the FBI is looking to justify a new interpretation of what it means to “streamline” the publication process. Rather than streamline by providing users swifter access to the data, perhaps the FBI is now looking to streamline by decreasing the amount of data they provide. Given that the FBI has not yet responded to further requests for comment, we’re left to read between the lines.