In the run-up to last weekend’s blizzard, Uber announced it would cap its dreaded surge pricing — the folly of so many on New Year’s Eve, who awoke to find their five-block ride home had set them back a whole rent check.

But the ride-hailing giant didn’t specify what its cap would be. It offered only a convoluted explanation: “The state of emergency price will be set after excluding the 3 highest-priced, non-emergency days of the preceding 2 months for that particular market.”

Amid boredom over the weekend, Annie Wang, a data scientist from Northwest Washington, took matters into her own hands. In short, she used the company’s own data — a program interface for software developers — to dig up the information Uber wouldn’t volunteer. The number she arrived at: 2.9 times the normal fare.

From the height of the blizzard Saturday to Monday morning’s cleanup, Uber’s rates were at a near-constant of three times the normal fare. Much lower than the nightmare New Year’s fares of seven and eight times the norm, but still high enough for riders to rack up hefty tabs — especially on the snow-caked streets.

That’s if you could even get a ride.

“What was pretty striking was the relative unavailability of Uber,” Wang said Tuesday.

Another graph of Wang’s (below) shows ride wait times at various points in DuPont Circle, Columbia Heights, Georgetown and Arlington. See that big void in the left-third, a dot desert next to a universe of blips on the right? That’s the blizzard!

This period — the 24 hours from noon Saturday to noon Sunday, is when many riders reported being unable to hail a ride anywhere in the city. In the middle of the day. For what appear to be at least 12 hours between Saturday and Sunday, the capital’s ride-hailing network basically went dark.

Note that Wang’s data is a representative sample, not a record of every trip taken in the region. And there’s a significant chunk of data missing — that’s because Wang’s boredom (the impetus of this study) didn’t set in until the storm had set in on Friday. She recorded data from the following locations: the DuPont Circle Metro station, the Columbia Heights Metro station, the Rosslyn Metro station and for Georgetown, Georgetown Cupcake.

Wang says Uber’s cap was evident from the data. In the first graph, you can see clear plateaus at points just below the three times surge.

“You can really see a sort of clear discontinuity in a way that you almost never see with real data,” she said, “real” meaning absent of preconditions.

The other big conclusion: Uber was slow to recover. Monday morning, when the cap was lifted and roads became somewhat navigable, rates surged — a likely effect of there being too few drivers on the road.

Here’s how Wang explained it: “Friday night there are no cars because the weather is awful, but on Monday morning, the roads are clearly drive-able because there are estimates for the rest of the day. But that period is congested because there probably aren’t enough drivers on the road —- maybe because even 5x surge is not enough for drivers?”

In an email, Uber spokeswoman Kaitlin Durkosh declined to comment on the company’s surge cap.

Durkosh declined to specify how many Uber trips were taken in the region during the storm, saying only the number was in the thousands. She said demand was at its highest in the region shortly before the storm arrived, but it surged again on Friday night and Saturday afternoon.

(Washington Post owner Jeff Bezos is an investor in Uber).