Earlier this week, Airbnb released a massive dataset about its business in New York City. The company wanted to counter the portrait painted last year by the New York state attorney general, which accused Airbnb of enabling illegal hotels, allowing a small number of hosts to earn millions while cutting down on the amount of affordable housing available to the average citizen.

Airbnb has pushed back against that perception, pledging to be an "open and transparent" company that would work closely with cities to ensure it paid its fair share of taxes and avoids eating into the supply of affordable housing. The release of this data was part of an agreement with the AG. In an interview with The New York Times, Chris Lehane, Airbnb’s head of global policy and public affairs, said that after looking at the numbers "our hope is that people will understand 99 percent of people on Airbnb in New York City are using it as an economic lifeline."

But a review of the data by The Verge found that Airbnb's numbers, covering November of 2014 through November 1st of 2015, largely confirmed the attorney general's accusations. A small number of hosts renting out multiple listings took home a disproportionate amount of the total revenue. And while roughly 71 percent of hosts rented out their home for three months or less, there were still thousands of "whole units," meaning an entire house or apartment, which were rented for six months or more during the last year.

For example the number of hosts with three or more listings on the platform accounted for less than 2 percent of the total hosts on Airbnb in New York City during the time period covered by the data. Yet that group took home 24 percent of the total whole home revenue. And that only covers people who are openly renting multiple listings. The New York Times interviewed one Airbnb host who rented five units, but only one under his own name. He estimated his annual income from the platform at $500,000, far more than the $5,110 Airbnb highlighted as the median host income.

In Airbnb's defense, a snapshot of the service from November 17th of this year showed a very different environment, with far less revenue going to the small group that hosts multiple units. Projected out over the rest of 2016, the group of hosts with more than three listings would earn just 6 percent of the total revenue, instead of 24. Wrede Petersmyer, Airbnb's city manager for the New York metro area, said that the change reflected a mix of increased enforcement by Airbnb and greater awareness among the average citizens the company wants to woo as hosts.

Still, that data was a snapshot of a single day, chosen by Airbnb. It showed zero hosts with five listings, which would be a dramatic reduction from the preceding year, when there were 190 hosts in this category. But that date still had over 300 hosts with six or more listings.

It was difficult to assess how much the hosts with multiple dwellings earned, because Airbnb would only share the median income. The company would not discuss the minimum or maximum amounts earned, citing the need to protect hosts' privacy. In fact, while the company stated on its website that it had made 170,000 lines of data available, the data was not published to the web. It was made available to journalists, upon request, on Airbnb laptops in a private meeting room. But that data was redacted in several ways, for example, it lacked any information on the roughly 25 percent of listings that are made up of private and shared rooms, instead of whole apartments or homes.

The data also appeared to contain mathematical errors. The Verge reviewed the data alongside another journalist, who noted that the figures for listings in each of the five boroughs, when added up, did not match the total listings enumerated at the end of each column. So, for example, the number of listings in each of the five boroughs did not match the total listed for all of New York City. The discrepancy amounted to just a few percentage points, but highlighted that this was far from a complete or unadulterated view of the data.

Overall the data is a big step toward meeting the company's pledge of transparency. But viewed carefully, the numbers tell a different story than the one put forward by Airbnb. Over the last year, hosts renting out multiple units for long periods of time still represent a significant portion of Airbnb’s income in New York, potentially taking housing stock off the market. And until full access is given to more accurate numbers, it will be tough to rely on their fuzzy math as concrete evidence of change.

UPDATE: In an email to The Verge, Airbnb asserted that the mathematical errors, acknowledged during a presentation to journalists by its New York City manager, were actually based on a misunderstanding of how to read the assembled charts, and not an underlying issue with the data itself.