LinkNYC’s “Super Fast Wi-Fi” Speeds Revisited

I never got the job of “Franchise Inspector” at DoITT. Evidently I was never even considered. But I do the job anyway, for the love of it, and to the extent that I have resource to do so.

I last connected to LinkNYC’s free WiFi in March, 2018, when I used a Samsung Galaxy Note 5 to test speeds on as many as a couple dozen kiosks in and around Astoria/Long Island City. Before that I had only connected a few times to the first machines that went live on Third Avenue in early 2016.

This time around I made more of a project of testing WiFi speeds on more kiosks and in more locations. Using a Samsung Galaxy S9+ I tested about 60 machines in three separate areas: Park Slope in Brooklyn, Astoria/Long Island City in Queens, and Jackson Heights in Queens. This table shows locations, download speeds, and upload speeds (in megabytes), according to Speedtest. I forgot to note the exact intersections of the last two, on 37th Avenue in Jackson Heights, but they would not have been far from the previous ones on that avenue.

Location	DL	UL
33rd St. @ 37th Ave.	45.5	20.3
37th Ave. @ 36th St,	76.4	64.1
37th St. @ 37th Ave.	7.84	3.21
38th Ave. @ 35th St.	249	199
31st St. @ 38th Ave.	78.1	87.8
31st St. @ Northern Blvd.	81.0	112
31st St. @ 40th Ave.	20.9	11.2
40th Rd. @ Northern Blvd.	69.6	42.9
Flatbush Ave. @ Sterling Pl.	245	248
7th Ave. @ Park Pl.	71.8	77.7
Flatbush Ave. @ Prospect Pl.	31.7	20.7
Flatbush Ave. @ St. Mark’s Pl.	241	196
Barclay’s Center	57.5	37.9
5th Ave. @ Bergen St.	247	228
Bergen St. @ Flatbush Ave.	57.7	27.9
Flatbush Ave. @ St. Mark’s Pl.	177	162
7th Ave. @ Park Pl.	240	275
7th Ave. @ Park Pl.	45.5	20.3
Northern Blvd. @ 42nd Pl.	66.1	59.8
Northern Blvd. @ 36th St.	n/a	n/a
Northern Blvd. @ 34th St.	68.6	67.9
31st St. @ Northern Blvd.	17.0	6.75
Northern Blvd. @ 40th Ave.	9.10	1.08
31st St. @ 38th Ave.	49.2	25.7
Broadway @ 21st St.	29.6	7.63
Broadway @ 23rd St.	89.7	91.1
Broadway @ Crescent St.	257	259
Broadway btw Crescent & 29th St.	48.0	25.0
Broadway @ 29th St.	n/a	n/a
Broadway @ 30th St.	40.9	14.4
Broadway @ 31st St.	22.5	16.6
Broadway @ 32nd St.	1.13	0.62
Broadway btw 32nd & 33rd St.	245	243
Broadway & 33rd St.	244	149
Broadway & 34th St.	44.5	17.7
Broadway & 35th St.	2.26	0.27
Broadway & 36th St.	n/a	n/a
Broadway & 37th St.	70.4	70.3
Broadway & 38th St.	4.42	0.81
Broadway & 38th St.	45.6	22.5
Broadway & 38th St.	4.99	2.87
Broadway & Steinway St.	n/a	n/a
35th Ave. @ 35th St.	240	243
35th Ave. @ 37th St.	243	233
35th Ave. @ Steinway St.	39.6	38.0
Northern Blvd. @ 44th St.	77.4	59.04
Northern Blvd. @ 45th St.	12.1	1.95
Northern Blvd. @ 48th St.	63.8	29.0
50th St. @ Newtown Road	69.3	37.8
Northern Blvd. @ 50th St.	238	148
Northern Blvd. @ 53rd Pl.	13.2	35.4
Northern Blvd. @ 62nd St.	72.8	36.8
Northern Blvd. @ 73rd St.	34.7	22.1
Roosevelt Ave. @ 95th St.	255	205
Roosevelt Ave. @ 95th St.	55.6	31.0
Roosevelt Ave. @ 95th St.	20.8	5.86
Roosevelt Ave. @ 94th St.	n/a	n/a
Roosevelt Ave. @ 93rd St.	255	241
37th Ave. @ 88th St.	91.1	91.2
37th Ave. @ 81st St.	223	228
37th Ave. @ ???	255	220
37th Ave. @ ???	42.7	18.5

Consistent with my last survey, results were all over the place, with some machines delivering wicked fast speeds capping out at 300/300, others offering barely a trickle, and some not working at all.

The last category (non-working) exposes what I’ve always assumed: That the LinkNYC Status dataset released by CityBridge paints an impossibly rosy picture of the network’s WiFi uptime with, as of this writing and for the last 2+ years, WiFi reportedly down on only one kiosk. I encountered 5 kiosks to which I was unable to connect.

I once again used the Speedtest app, which I assume to be reliable. When a connection failed I always tried a second time. Failed second attempts are listed as “n/a”.

Except for instances where homeless people had the kiosk surrounded I stood pretty much right in front of the machines. Even with LinkNYC encampments present I stood well within what should have been acceptable range to get the FULL FORCE of LinkNYC’s WiFi.

I’m aware that all this might be considered anecdotal. A more thorough survey would involve testing individual kiosks multiple times over a span of days, connecting with multiple devices, and deploying network diagnostics not available to me.

For instance: Is a kiosk’s WiFi stupid slow because 100 people are using it all at once? That’s possible but in many of these instances with glacial speeds it’s just not realistic that a mass of users is hiding somewhere within usable range, gobbling up all that “super fast” bandwidth.

The one at 37th Street and 37th Avenue, showing speeds of 7.84/3.21, is outside a car dealership but surely they have their own internet access. Even if they don’t their physical location is probably out of range anyway.

The same likely holds true for the one at Northern Boulevard and 40th Avenue, with sizzling speeds of 9.10/1.08. It’s outside a storage facility but that place must have its own internet access.

The one at Northern Boulevard and 53rd Place, at 13.2/35.4, is outside a Covid-19 testing facility but I would assume, or rather hope, they have their own private and secure internet access.

At one location showing single-digit speeds I asked someone who worked at a bar right there if she ever used or attempted to use the nearby LinkNYC’s WiFi. She said no, the bar had its own internet, but she had witnessed several people try and fail to connect to that kiosk’s WiFi.

Are WiFi Stats Bloated?

One other matter I had in mind involved my suspicion that usage stats published by CityBridge are somewhat bloated in terms of number of WiFi sessions initiated. The last stats published, from December of last year, report that on average about 7500 sessions were initiated per kiosk per week, averaging out to over a thousand sessions per kiosk every day.

I’d like to see kiosk-by-kiosk stats made public. It has to be that specific kiosks get blockbuster usage while others get virtually none.

For that matter I’d like to see the usage stats dataset, which ceased being published at the end of last year, come back to life. It would be interesting to see how or if the pandemic might have influenced usage of the kiosks.

To test my suspicion about bloated “sessions initiated” stats I checked if my phone reconnected automatically while its screen was turned off, and also while turned on. As expected, it reconnected automatically when turned on but not, as far as I could tell, while turned off.

This would take more testing with more devices to confirm, but reconnecting to the WiFi while the screen is turned on, and when I make no use of the signal, seems to inevitably increase the number of WiFi sessions initiated even when they are not intended or used.

But remembering that the initial vision of LinkNYC was of a free continuous blanket of uninterrupted WiFi everywhere I guess bloated numbers of sessions initiated reflects the nature of the beast.

At least the numbers do not appear to get further bloated by devices reconnecting while the screen is off.

Notes

I had said I encountered 5 kiosks with no WiFi signal, but when Speedtest shows speeds like 1.13/0.62 I have to ask if the kiosk in front of me is actually sending any signal at all, or am I picking up the faint signal of another one across the street, 100+ feet away. If the latter scenario is true then the number of kiosks with non-working WiFi in the list above would increase.

Standing in front of these kiosks revealed something else about them: They have not aged well. The paint is worn, some look like their advertising panels are ready to fall off, scratchiti and graffiti artists have done their thing…

Many of the tablets do not work, and some that had been smashed by vandals are covered with a black piece of material. As I type these words over 10% of LinkNYC’s tablets are reported down. Unlike the WiFi and phone calling status published by CityBridge the hourly report on tablets functional status, while probably not 100% accurate, appears to at least give some reflection of reality.

It mostly disappears into the soundscape of the city but these kiosks make a lot of noise, like an old air conditioner, though you’d probably never know that unless you try to catch a nap while leaning against one of them. It would be interesting to know how much energy these things consume.

LinkNYC advertising screens also freak out sometimes, blasting potentially seizure-inducing visual noise. This is not a new problem with LinkNYC but I had not witnessed it in a long time.