MAITLAND, Fla. -- Andre Weingarten sticks his head through the space between the whiteboard and the edge of Dustin Smith's end-of-row cubicle. They watch one of three computer screens on Smith's desk. A YouTube video of Kyler Murray, the former Oklahoma quarterback who would become the No. 1 pick in the NFL draft six weeks later, is playing on one of them.

Together they watch. For hours. Daily. Prospect by prospect they go, the bespectacled 34-year-old Smith and the bearded 23-year-old Weingarten -- making the decisions for how digital players will rate in this year's Madden video game.

Ten years ago, Smith was a game-tester for Electronic Arts, searching for bugs in the game and ways to make the player likenesses more accurate. Weingarten was a teenage fan who eventually went to school to learn, among other things, scouting. Now, they make some of the game's most-discussed decisions at their two desks on the sixth floor of a building in a nondescript office park 30 miles northeast of Disney World.

Here, virtual NFL players come to life for millions of gamers around the globe. On this Monday, they are taking an initial look at Murray.

"What I'm looking for here is velocity, release and arm strength," Smith said, watching Murray make throws in a game. "Look, he's looking at the same read. One. Two. Three. Looking at the same read."

"That's really the biggest issue I have with Murray," Weingarten said. "He's not exactly known for running a complex offense."

During the season, they receive help from a set of employees known as Madden ratings adjustors -- many of whom have playing backgrounds in college or the NFL -- to deal with the breakneck speed of week-to-week ratings changes. For initial game ratings of rookies and veterans, Smith and Weingarten work through a year's worth of tweaking, poking, studying, analyzing, debating, searching for any morsel of information and staring at computer and television screens as they determine the individual and overall ratings of 2,900 players by the first week of June.

In this moment, they are trying to gauge Murray's throwing power -- one of 53 public ratings they'll create for him -- eventually landing at an 89 rating that will tie for 19th with Seattle Seahawks veteran Russell Wilson and fellow rookie Dwayne Haskins. Smith and Weingarten want the quarterbacks you play with in the game to feel more like how the quarterback plays in real life. Fifty-three quarterbacks had a throw power of 90 or better last season. This year, there are 18.

It's part of a bigger ratings spacing -- the brainchild of Smith and Weingarten -- to create more realism. Last year, 1,590 players were rated 70 or above overall at release. This year, 1,177 will have that initial rating.

"It's harder to explain to people on top of all that, because everybody is always tied into what the rating always meant, and this year we're doing a big stretch," Smith said. "Maybe not as big as I want it to be, but it's a stretch noticeably, where, like before your worst of your worst was maybe a 59.

"There may be now a 51. Overall."

Murray is in no danger of being that low. His talent, what he did versus college competition, scouting reports breaking down his strengths and weaknesses, and where he would rank against other quarterbacks mean he'll be one of the better-rated rookie quarterbacks when Madden '20 is released Aug. 2. Smith and Weingarten had a feeling in mid-March, when they opened up their process to ESPN for two days, that Murray might end up as the Arizona Cardinals' No. 1 overall pick.

Smith's desk looks like it belongs in Texas -- eight Dallas Cowboys helmets and four Cowboys jerseys surrounding him. It's an odd fit for a man who grew up in the Kansas City suburbs, a one-time defensive back at Raytown (Missouri) High School who used to hear flyovers at Arrowhead Stadium from his house. After graduation, he worked as a valet at the Argosy Casino in Riverdale, Missouri, playing Madden in his free time and logging on to the Operation Sports message board in 2008 and 2009. He would point out mistakes and make suggestions on what the game was missing, particularly with player likeness and gear.

It is knowledge Smith acquired as a kid thanks to a photographic memory of his football trading cards. The detail caught the attention of then-Madden designer Ian Cummings, who reached out and invited Smith to Florida to participate in a community forum in 2009.

While there, Smith asked how he could work at EA. Representatives told him he could start as a game tester -- known as a "QA" -- in Baton Rouge, Louisiana. Wanting to get into game development, he took the gig, searching for bugs in games for three years before taking a yearlong hiatus, staying in Baton Rouge and parking cars at Ruth's Chris Steakhouse.

He returned to game testing in 2014 for one more year before interviewing for a job at the home office in Florida. Rex Dickson, then Madden's creative director, hired Smith to work on gear, uniforms, likeness and authenticity in January 2015 as a contractor -- essentially creating the same things that he was criticizing on a message board seven years earlier.

Once hired, he worked with then-ratings czar Donny Moore. Smith learned the massive Madden database -- a spreadsheet with thousands of ratings -- and how to create ratings. Moore left for FanDuel later that year, and Dickson offered Smith the job.

Moore walked Smith through everything before he left, including how to create undrafted rookie free agents. Smith was overwhelmed. He worked in anonymity outside EA's offices in 2016 while gaining comfort with the system and the ratings.

"Did I feel I could do it? Yes," Smith said. "I knew there would be a ton of pressure. I had seen what people had said to Donny for years, and it wasn't going to be a simple task to try and jump into that role."

By 2017, Smith had a better feel; he understood the pressure in his 50-hour workweeks and realized that outside criticism comes from everywhere. Smith said that Odell Beckham Jr. blocked him on Twitter because he was upset with a rating. He was a one-man operation until last year, when EA hired Weingarten on a one-year contract, in part, to help Smith.

A native of New York's Long Island who transferred from Hofstra to United States Sports Academy in Daphne, Alabama, Weingarten added a mathematical component from a scouting background, differing from Smith's experience as a work-your-way-up Madden employee. They clicked fast. Now a full-time employee, Weingarten spends roughly 70% of his time working on rosters with Smith.

Having Weingarten allowed more film study to work through the eight to 10 rookies they evaluate and rate per day. They can have a whole position group of veteran players -- who only require small adjustments -- done in a day. With every player, they rate first and rank after.

They devised formulas to take some of the guesswork out of ratings. Five categories are pulled almost straight from combine numbers: Strength, jumping, speed, acceleration and agility. The other ratings -- including two private ratings, one for celebration and another they declined to name -- are determined from the research Smith and Weingarten do throughout the year and scouting reports they receive. They declined to name their scouting reports or give much information on their specific formulas for proprietary reasons. Draft guides are key. So too are specific Twitter accounts -- including those of Kent Lee Platte () for athletic ratings and Ian Wharton () for catchable ball and accuracy breakdowns.

The formulas are not the end-all for ratings. Game speed and position also are taken into account. It's one of the ways -- along with ratings spacing -- they are trying to add more reality to a game that attempts to thrive on it, similar to FIFA and NBA2K. Moore began working on formulas before he left. The ones Smith and Weingarten use now, including the strength formula Weingarten created, were refined in the past year.

It also is where a player like Murray becomes more challenging -- because they have no numbers to work off of since he didn't work out at the combine.

Smith drops Murray into the massive spreadsheet. Listed as a free agent, he'll be judged against every other quarterback in the league.

In order to build a rookie, they use a duplicate of another player, usually a lower-tiered one, who has the same height and weight. It's a time saver that will later be customized, although they use Drew Brees as Murray's duplicate because there are so few quarterbacks with his build. They search Google images to find photos of a player with arm sleeves, wrist tape, a different hairstyle for as much authenticity as possible.

Using a duplicate as a base gives them more time to focus on ratings. This year, they had 401 rookies built with basic information by Feb. 2. Last year, they only had 170 bodies by April 2.

Rookie quarterbacks take the longest to rate. There are eight position-specific ratings and physical metrics -- and rushing ratings for dual threats such as Murray. Most of those aren't math-based markers, either.

"Nobody is going to sit there and tell you, 'Oh, he's good at breaking sacks,' unless it's something really common he does," Smith said. "You have to go find information like that."

Wide receivers, running backs and tight ends also take time, but quarterbacks end up being the most painstaking -- and often the most critiqued. With Murray, speed is the start.

With nonexistent combine numbers, Smith and Weingarten scoured the Internet searching for a 40-yard, straight-line run in a game to time him for an approximate 40-yard dash and 10-yard split. It's harder than it will be a week later, when they would receive access to every college game film from the 2018 season.

This gave them a starting point -- a number eventually cut down by a few hundredths of a second because they found an angle that showed a little curl at the end of the run, landing him at a 91 speed rating. His 10-yard split, which is part of the acceleration formula, helped Murray earn a 92 in that category.

Agility, for which Murray rated a 90, is the hardest for them to compute, even with a formula.

"Higher emphasis on the three-cone, shuttle, and then we use the eye test for the majority of it," Weingarten said. "It's good to have that number to attach to it, just because we're flying blind if you don't have any number. Same thing with a 40 time.

"Like you can feel comfortable, but if you don't know what they ever tested at, it feels off."

Most of the ratings don't have combine numbers attached. This is where the reports matter, as well as their own study, ratings and the reality that where a player is drafted doesn't necessarily correlate with his rating. Smith looks at players in comparison to their peers at a position, not in a draft class.

Weingarten's head is through the space between the desks again. He leans forward in his chair, watching more film of Murray trying to make decisions on his accuracy on short, intermediate and deep throws. It's where the debate begins and the reality of creating a player with game-altering skills sets in.

"Based on what we're seeing here, I'm going to immediately make a decision on short and probably put it at 87, 88," Smith said.

"That's pretty high," Weingarten said.

"I know. And let's look at who is there with him. Sam Darnold is an 87," Smith said.

"But that's Darnold's strength. I would not describe his short accuracy as [Murray's] strength," Weingarten said.

"I know. It should be his deep, but we can't put his deep as the strength. That would literally break the entire mold of how we do quarterbacks," Smith said.

"Isn't that exactly what Kyler Murray is doing?" Weingarten said.

To settle it, they look at Wharton's charts breaking down Murray's catchable passes in each zone. Smith sees 32 of 41 short passes were catchable, numbers he is "not too horrified" at. Weingarten responds that it is 16th best of 65 qualifying rookie quarterbacks they've charted since 2012. For comparison, they looked at Darnold's numbers from last season, which put him at No. 8.

They also use Lamar Jackson, who had a 74% short completion rate and rated an 83 last year. They decide, initially, to make Murray an 84 in short accuracy -- a number that eventually will drop to 82 because of tweaking after watching college film and seeing where his rating fit with other quarterbacks.

Smith and Weingarten agree intermediate (known as "throw accuracy mid" in the game) will be Murray's worst passing rating. It's often the hardest for a rookie quarterback to adjust to, based on reports and film study. His height also plays a role.

"His passes sail in that area because he has to try and throw over guys and past underneath guys," Weingarten said. "It's not a good spot for someone of his height. That's why Russell Wilson has had some struggles in that area. [Baker] Mayfield had a little bit. Brees was a little hit or miss early in his career in that area."

Murray will land at a 78 in intermediate. Using a consensus that the deep ball is his strongest throwing attribute, Smith and Weingarten give him an 82 -- tying him at No. 23 with Ryan Fitzpatrick and Cam Newton.

These non-mathematical ratings and traits for rookies and vets are based off years of experience in the game, NFL and college game study throughout the year, watching film, scouting reports and Weingarten's notes from the Senior Bowl. Some ratings also depend on whether or not they actually performed something in college; for example, there was very little, if any, tape of Murray performing play-action under center, so he gets a 71.

Awareness is among the most subjective ratings, because it is only for use in Madden. Smith describes it as "how well a player knows his surroundings and what to do on the field." For quarterbacks, it means adjusting to defenses, reading playbooks and how they cycle through progressions. Rookies often start low, no matter the position, which is why Murray is a 65 -- tied for No. 70 among quarterbacks.

"I like to leave room for improvement, regardless of what they proved," Smith said. "Because they always could get a lot worse moving forward."

The ratings are tied into archetypes. They tried to make Murray an "improviser," which replaces "West Coast" in this year's game, but ended up with him as a "scrambler" -- one point ahead of his improviser rating. Smith said that depending on how he performs during his rookie campaign, Murray easily could switch archetypes in weekly updates.

"We know, like certain guys, they have to be a certain player type. So, if Josh Allen doesn't come out as a strong arm, even though he's a scrambling guy, it doesn't make sense that a 99 throw power isn't a strong arm. It's strange," Smith said. "It just doesn't read well. You know he's a strong arm. He'd be a scrambler and maybe people don't question it, sure, but I just don't think it looks good having a guy say scrambler and then you have 99 throw power. It would come off really strange."

Weingarten, who grades players like a scout, had an early second-round draft grade on Murray as far as true talent. Quarterbacks -- in Madden and in real life -- are looked at differently. Which is why they take the longest to build and often are the most focused-on players in the game.

Eventually, they give Murray a 73 overall rating -- one number down from their initial rating of 74 (he is a 72 as an improviser) -- tied for No. 32 among quarterbacks. Smith wondered if that rating was "generous."

"He's in the top 32," Smith said. "He's in a starting quarterback rank for a guy who has never played before, which is saying something."

Every day, in those same two cubicles, these debates happen between two men charged with making Madden ratings for the world -- a massive job that only grows by the year as the game increases in popularity and technology becomes more advanced.

They understand the attention paid to what they do. Over the next few months, they'll hear from everyone. They'll take the criticism. They'll keep taking notes. Then, as soon as the next football season ends, they'll start the process all over again. Because inside these offices, the ratings never stop. They just continue to evolve.