I tried to make some calls from LinkNYC terminals yesterday. I did not have much luck. After 25 attempts at 25 individual kiosks only two calls connected, although a casual observer and even LinkNYC’s traffic monitoring might think most of the calls went through. That’s because a majority of the calls looked like they connected, and the timer started ticking to show the length of the call, but whoever or whatever I called could not be heard. How many of these silent calls account for LinkNYC’s reported 600,000 calls per month?

It’s been like this since the earliest days of LinkNYC, the so-called “payphone of the future” that has already replaced thousands of trusty old payphones. The “ability” to make phone calls from LinkNYC kiosks should more accurately be described as a “possibility”.

Lately it seems to have gotten worse. A couple of months ago I attempted 99 calls from midtown Manhattan kiosks. Only 44 of those calls connected. Last year, as I know from having obsessive-compulsively dialed up demented ice cream truck music from as many of these machines as possible, I would guesstimate I encountered non-working LinkNYC phones only about 25% of the time.

VIDEO CALLS NOW AVAILABLE ON THE VIDEO RELAY SERVICE

It looked as if the LinkNYC rollout had completely stagnated, with no meaningful features added to the tablets, and virtually no new installations this entire year. 2019 has even seen a reduction in the number of activated LinkNYC machines. 1,776 machines are reported active as of today, down from 1,788 in July.

But recently, to end the stagnation, LinkNYC announced the availability of a video phone calling service. This means one more camera, this one located over the tablet screen, has been activated, but not for video calls between the kiosk and your cell phone or Skype. That long-promised feature will probably never make it to LinkNYC.

Instead this camera is reserved for use on the Video Relay Service (VRS), an FCC-sponsored program allowing deaf or hard of hearing individuals to make phone calls using human sign language translators as intermediaries. A hearing-impaired individual dials the phone number of a person they wish to reach, and a live video feed of a sign-language translator appears on the LinkNYC tablet screen. The translator, known as a Communications Assistant (CA), connects to the desired phone number and translates the hearing-impaired person’s sign language to the person being called, while also translating the words of the person on the phone into sign language for the hearing-impaired VRS caller to read.

There is no cost to the caller or the callee for this government-funded service.

My hearing is fine but out of curiosity I took a look at this service. I thought some hidden trick might allow me to make direct video calls from the kiosk to my cell phone. Nothing doing, as far as I could tell. After I entered my cell phone number into the VRS screen there appeared a low-resolution live video of my neck. The camera really is alive, I thought. The lens is too low to capture my face so I would have had to bend over some for the CA to see me.

Live video appeared on the tablet screen, showing a man hurrying to sit down at his desk as my VRS call came in. I did not intend to disrupt this person’s day, so my apologies to that individual for hanging up on him. I can imagine high school pranksters making a laugh riot of this curbside live video curio, delivering their own brand of sign language to these CA intermediaries.

Having not used the VRS in its intended manner I cannot say if the interface suffers from the same eavesdropping design flaw as encountered when making regular phone calls. If the CA’s voice is heard through the loudspeakerphone then everything the hearing-impaired person signs to the CA could be heard by any non-deaf people nearby.

As of this writing no mention of this newly activated camera has made it to the LinkNYC FAQ, which still says the camera over the tablet “could be used for video calls in the future”.

DID LinkNYC SURVIVE THE WEST SIDE BLACKOUT IN JULY?

On July 13, 2019, a ConEdison blackout cut off electricity to a swath of Manhattan’s West Side. Did the LinkNYC kiosks in the affected areas continue to function without power, and for how long? As far as I can tell no one asked this question on social media or anywhere else.

It is relevant because LinkNYC’s purported ability to remain alive during power outages was a key selling point in its original RFP to replace the old payphones. Access to a functioning public telephone network has been critical for public safety in past blackouts, and it was assumed that a replacement for traditional payphones would be built smart enough to take such things into consideration.

Traditional payphones, at least according to conventional wisdom, could be expected to stay alive during power outages, since their copper-based lines did not rely on the power grid. But that changed in recent years, at least in New York, with a majority of the city’s outdoor payphones converted to solar power and connected to the phone network not via copper landline but by cellular routers. Backup battery power should keep today’s outdoor payphones alive during a blackout, but that’s only useful if the Verizon wireless network to which they connect stays alive – a dubious expectation at best.

To analyze what might have happened with the LinkNYC network on the night of the blackout I looked at hourly datapoints from the LinkNYC Kiosk Status dataset. I’ve collected these datapoints since April. The data, provided by CityBridge and published at NYC.gov and other government data portals, is said to provide live hourly snapshots of how many of the LInkNYC’s network’s devices are working, and how many are not.

I only collect network-wide stats, not kiosk-by-kiosk figures, though I do post hourly snapshots of which specific machines are reported as not working.

Stats from the night of July 13 are interesting. The number of kiosks reported with non-working Wi-Fi went from 47, or 2.63%; to 159, or 8.89%. I never see that kind of percentage fluctuation from one hour to the next. The typical hourly difference is well under 1%.

The reported number of non-working tablets on the night of the blackout went from 33, or 1.85%; to 134, or 7.9%.

The reported number of non-working phones, miraculously, went from 1, or .06%, to 0. That cannot possibly be accurate but that is what the dataset from CityBridge has reported not just on the night of the blackout but pretty consistently since at least June.

The blackout started at about 6:45pm. ConEd reported all power restored by around 11:00pm.

My datapoint snapshot from 7:16pm, a ½ hour into the blackout, shows that 33 of LinkNYC’s 1,788 tablets, or 1.85% of the network, were offline. 8:16pm still shows 33 tablets down, a number about average for that day.

At 9:16pm, about 2½ hours into the blackout, the number of non-working tablets jumped from 33 to 78. At 1:16am, over two hours after power had been restored, the number of non-working tablets jumped again to 134, staying at that level until 5:16am, when it dropped to 15.

I find it hard to know what to make of this data. I’ve found its accuracy very suspect in the months that I’ve kept an eye on it. It purports to give live status snapshots of the LinkNYC network but time and again I encounter non-working machines that never appear in this data, even when one or more of the kiosk’s features had not worked for weeks.

The consistently rosy figures claiming virtually 100% uptime for LinkNYC’s phone calling feature is just wrong. I routinely encounter devices that cannot make phone calls for one reason or other, either on account of my above-mentioned experience of making calls that appear to connect but do not, or else calls cannot be made because the tablet screen does not work.

On rare occasion when the tablet is not working I’ve found that calls can be made using the physical keypad. I would estimate that is possible less than 5% of the time, but I’ve not put that guess to a rigorous test.

To take data from the night of the blackout at face value it looks like the LinkNYC kiosks in the affected area stayed alive for as long as a couple of hours after the power went out, then started going dark between 8 and 9pm. A second round of kiosks going dark appeared to have occurred around midnight. If the blackout ended at 11:00pm then why would another round of kiosks go dark at midnight and remain offline until 5:16am?

Whatever the integrity of the data at its most granular levels it seems obvious something interrupted service on a relatively large number of kiosks the night of July 13. Most likely impacted were kiosks in areas affected by the power outage. Common sense would further blame LinkNYC’s conspicuously high number of outages that night on the kiosks’ inability to stay powered up for very long during a power outage.

THEY GOT RID OF DARKSKY.NET.

One of the most interesting and useful discoveries I’ve made through LinkNYC was DarkSky.net, a weather website I’d never heard of until it was added to the kiosks in January. Access to that site was formerly available through the tablet screen. One of the tiles on the tablet’s home screen contained a weather snapshot which, when tapped, led to a somewhat crippled version of the DarkSky.net website. You could get to a lot of data but some pages were blocked.

Sometime within the last few months that access to DarkSky.net went away, and tapping the weather snapshot tile on the tablet screen has reverted to doing nothing. AccuWeather is now the weather data provider.

To me the difference between DarkSky and AccuWeather is that DarkSky’s rain forecasts are just scary accurate compared not just to AccuWeather but to any other weather source I know of. With the forgivable exception of a few tiny drops of rain on a couple of occasions I have not seen DarkSky get a rain forecast wrong, while AccuWeather and most other providers frequently claim it is raining outside when it is not.

THE PRESIDENT IS OUT

The now-former president of LinkNYC, Jen Hensley, announced her departure from the company. I know nothing about the company’s internals, or any circumstances surrounding this departure. But with the LinkNYC rollout having stagnated for well over a year it seemed obvious to this outside observer that the program’s early momentum had been lost, and something had to give. No replacement has been announced to the public.

I DID A RANDOM SURVEY OF HOW MANY ADS SHOW ON THESE SCREENS.

I decided to choose some LinkNYC locations at random, to see how many ads they show, versus PSAs, platitudinous NYC Quotes, and other filler content.I did this last month. The table below shows the location, the number of screenchanges I recorded, and how many ads versus everything else appeared. Clearly, some kiosks spit out a lot of ads, some don’t. The kiosk that performed in a manner I should have expected was the first one listed here, at 3rd Avenue and 49th Street. But the other midtown kiosk, at Lexington and 48th, locked in on “Charge your phone” and similar screens promoting the free services of the kiosk. I had intended to get at least 50 screenchanges per machine but I gave up the ones when it became apparent that glitches would prevent ads from appearing. The Court Square location showed almost nothing but “Find your way” screens. Artwork dominated the other 2 Queens locations, which look like they would average about 1 ad per every 50 screenchanges. The “blank white screen”, which I had seen plague other kiosks far worse than the ones shown here, is seen in the photo below this table.