Updated: LinkNYC Real Time Kiosk Status Data Released

I first posted this on July 9, 2018, and updated it July 14 with follow-up observations.

The city recently, and very quietly released the LinkNYC Kiosk Status dataset, a spreadsheet which its creators claim presents “the most current listing of LinkNYC Kiosks, their location, and the status of the Link’s wifi, tablet, and phone.” This dataset supplements a variety of other datasets regarding LinkNYC, including approved kiosk locations, application permits, maps, etc. This release is intended to be a more or less real time listing of what kiosks work and which ones do not. Or is it? Is it really intended to be that? I have reviewed enough of the LinkNYC Kiosk Status dataset to conclude that it contains numerous inaccuracies, and that it amounts to nothing more than an attempt to paint a picture of the LinkNYC network that is far rosier than reality. In some ways the data do not even make any sense.

COLUMN P
With the exception of kiosks which have not yet been activated (these are listed as “Not Available”) it would appear from this dataset that WiFi is up and running on all but three of the 1707 active devices. This claim is simply impossible to believe but, even if true, is also not especially useful. As I experienced a few months ago, sometimes LinkNYC’s WiFi works, sometimes it does not, and even when it connects the speeds can be uselessly slow. Not reported in this dataset is something like an average connection speed or average number of successful connections per day, figures which might actually be useful to someone who is sitting at their desk looking for an outdoor space with usable free WiFi.

COLUMN Q, “TABLET_STATUS”
Without even leaving my desk I can report that as many as a dozen kiosks on which the tablets have basically never worked are reported as being up when this is not true. These are tablet screens I pass almost every day, so it’s not as if they went out of service after this dataset was released. A particular tablet on Broadway in Astoria has been down for months but this dataset reports it as up. Another tablet that, since the day it was activated, reports no network connectivity when I attempt to use any of the apps is listed in this dataset as up and running. There are also a few instances of tablets reported as down when, by my observation, they appear to work just fine.

COLUMN R
The  “PHONE_STATUS” field is the most interesting to me for how little it reveals. But first of all, the coupling of columns Q and R creates data points that to me do not seem to make any sense. Of the 126 kiosks where the tablet is reported as down the status of the phone is said to be up. How is anyone supposed to make a phone call from one of these devices when the tablet is not even turned on? I guess it’s possible that the remote pinging techniques used to monitor these kiosks see certain functions as independent of each other, so if the phone call feature is itself lurking within the kiosk it might be seen as active even when you cannot get to it. It seems that a default for this dataset should be that when the tablet is reported as down then the phone should be, too.

UPDATE July 14: I overlooked one scenario where these seemingly conflicting data points might make sense: if phone calls can be made when the tablet is down by using the tactile keypad. Yesterday and today I tried a number of tablets where the tablet screen has basically never worked. Interestingly, I found that in some instances you could actually use the tactile keypad to make phone calls when the tablet screen was dark. That seemed at first like a reasonably smart fallback option if it was done by design. But I don’t think it was intentional. I think what was really happening with those devices was the tablet software functionality was awake but the screen was just not plugged in or able to be turned on. I only found two kiosks where this appeared to be the case. I found several more kiosks where the tablet screen was dark, no phone calls could be made, and it appeared that the 911 button was not lit. As I expected, the kiosk status dataset claims the tablet and phone on those devices were all up and running.

But the status of a phone being up does not open much of a window onto the vagaries of actually using these devices to make phone calls. I can’t tell you how many times I’ve dialed a number where it appeared to connect, the timer started ticking up, but the call never actually went through. This behavior, which has existed since the first days of these kiosks, is exhibited consistently on some, while others only show it once in a while. These phantom calls can stay connected for hours. Also unaccounted for is the quantity of kiosks that emit a high-pitched screeching noise when a phone call is placed, those that connect a call but disconnect it after just a few seconds (there are lots of those), the functional status of the touchscreen interface and/or the tactile keypad, how loudly you have to crank up the volume to hear the person you called, and other kiosk idiosyncrasies known only to those of us who actually use these things.

Not included is the functional statii of the advertising billboards, which is of course the one thing all New Yorkers care most about. OK, maybe we don’t, but I would think potential advertisers would be interested in knowing how reliably their messages might be communicated by knowing how many of the ad panels are dark at any given time. I would also think that LinkNYC/CityBridge itself would have an interest in knowing how many of their panels are potentially causing seizures across the city without having to rely on social media reports. I spotted 6 of these in one day last week, and four or five a few days later. These screens could give anyone a migraine.

But what is or is not included in the dataset is not as important as the accuracy of what is provided. Much of the data can be assumed to be accurate, but there are enough errors for me to conclude that in its entirety this is a useless piece of Smart City propaganda.

To be fair to DoITT, the city agency that has made this data available, they are simply taking CityBridge’s word for it that this dataset is a meaningful representation of reality. All datasets like this will contain mistakes but the quantity of errors and omissions in this particular set makes its integrity, and the intentions behind publishing it, highly suspect.

I can appreciate the exclamation points in column B, which communicate a hearty stroke of enthusiasm from the Smart City New Frontier. Maybe they could add another column that is empty by default but is filled with the word “HURRAH!” when a Link kiosk is activated.

But without wasting any more of my precious time on this I conclude that the LinkNYC Kiosk Status data presented is of no value. I had considered dumping the latitude/longitude coordinates and other relevant data points into a map but that takes time and by appearances that would make it look like I somehow endorsed the false sense of reality this dataset presents. Maybe this is why the folks at NYC.gov’s data portal chose not to map it themselves, and why this data has not been integrated into other maps showing LinkNYC kiosk locations.