The MTA has two main metrics for measuring delays, and neither way captures the rider's experience.

One measure of service is called On Time Performance. Take the L Train, for example, with an On Time Performance score of 91 percent. That measurement looks at how often a train gets to the end of the line within five minutes of its scheduled arrival time.

The problem is riders don't ride from one end of the line to the other.

The L train is the only line in the system that has Computer Based Train Controls, or electronic signals, which allows the MTA to run more L trains and trains that run closer together. But riders still can't count on getting to work on time. 

"There's frequent trains, but they're just very overcrowded," Williamsburg resident Noel Shirian, 34, said on a recent morning. Even during the summer, when commuters aren't competing with students for space, Shirian let two crowded trains pass. "I'm waiting on a third — hopefully, lucky number three will get me there."

The MTA doesn't include how long riders wait on a platform in the vast amount of data it collects on its service. It also doesn't know when someone just gives up entirely.

“Sometimes we get here and people are all the way to the other side of the platform and we just walk right back up," said another regular L train rider, Samantha Gordon. "We have to get an Uber or figure it out, because it’s just not realistic.”

The MTA has said a better measurement of its service is a convoluted metric called Wait Assessment.

"We believe Wait Assessment is a better reflection of the customer experience," MTA spokesperson Beth DeFalco told WNYC.

It sounds like it's a good measurement but the formula suggests otherwise. 

Essentially, Wait Assessment measures whether trains are running at regular intervals, which is what riders need to know; how long until the next train arrives. The problem comes in the way the score is calculated.

Wait Assessment measures how often a train arrives at a station on schedule, plus 25 percent. So if a train is supposed to come every 5 minutes, and it arrives one minute and 25 seconds late, it's considered on time. But if that train is one second over that, it's considered failed. If a train is 20 minutes late, it has also failed, but Wait Assessment treats these failures equally. 

Zak Accuardi, a senior program analyst with the think tank Transit Center suggests the MTA measure performance the way they do in London, using a metric called Excess Journey Time.

"This is going to help them improve operationally, it’s going to help the board conduct oversight of the agency," Accuardi told WNYC. "This is going to better reflect the experience that riders have using the system.”

While the MTA continues to assert Wait Assessment is its best measurement of service, the new MTA Chairman Joe Lhota said one of the first things he wants to do is bring in new service metrics, an MTA spokesperson told WNYC.

And at a recent City Council hearing, MTA Managing Director Ronnie Hakim admitted the agency needed to do better.

"Everything needs to be reviewed right now. There’s this recognition that the way we’ve been reporting stats don’t really help our customers,” she said.

The MTA declined to say whether Excess Journey Time will be adopted.