[jkilpatr]

I have gigabit fiber at home. I also run a WISP.

This is from one of my wisp customers taken literally seconds ago.

  root@OpenWrt:~# ping 1.1.1.1
  PING 1.1.1.1 (1.1.1.1): 56 data bytes
  64 bytes from 1.1.1.1: seq=0 ttl=61 time=7.791 ms
  64 bytes from 1.1.1.1: seq=1 ttl=61 time=7.084 ms
  64 bytes from 1.1.1.1: seq=2 ttl=61 time=6.691 ms

Same server from my gigabit fiber to the apartment unit

  [justin@DESKTOP-UALBV95 ~]$ ping 1.1.1.1
  PING 1.1.1.1 (1.1.1.1) 56(84) bytes of data.
  64 bytes from 1.1.1.1: icmp_seq=1 ttl=56 time=11.4 ms
  64 bytes from 1.1.1.1: icmp_seq=2 ttl=56 time=11.2 ms

The point to point customer antenna added ~6ms of latency the remaining ~2ms is the fiber line to the internet exchange.

tl;dr Wireless point to point is easily competitive to Cable on speed and anything on latency.

[Nullabillity]

> The point to point customer antenna added ~6ms of latency

For one hop. The OP literally has Mesh in the name, so I'd expect a large multiple of that. Especially given that NYC's highrises would likely make it harder to provide direct line of sight.

> Same server from my gigabit fiber to the apartment unit

That doesn't say much unless they're colocated geographically, and the rest of the equipment is equivalent. Especially since 1.1.1.1 is heavily anycasted (like most other public DNS servers).

For the record, my results look like this (also using consumer-grade gigabit fiber):

     teo  ~  ping 1.1.1.1
    PING 1.1.1.1 (1.1.1.1) 56(84) bytes of data.
    64 bytes from 1.1.1.1: icmp_seq=1 ttl=58 time=1.45 ms
    64 bytes from 1.1.1.1: icmp_seq=2 ttl=58 time=1.49 ms
    64 bytes from 1.1.1.1: icmp_seq=3 ttl=58 time=1.47 ms

> tl;dr Wireless point to point is easily competitive to Cable on speed

Very curious about your customer density here. It may be competitive for a few customers, but I suspect that interference from neighbors would cause a sharp drop in urban areas if everyone bought in.

[jkilpatr]

> For one hop. The OP literally has Mesh in the name, so I'd expect a large multiple of that. Especially given that NYC's highrises would likely make it harder to provide direct line of sight.

First off I will note that the goal posts here have been moved. Your original statement was that any wireless effort was doomed. This has now been revised to include many hops and distance to the internet exchange.

NYC mesh uses very traditional WISP equipment and OSPF routing (for the most part, also last I talked to them was a while ago)

If you look at their map they have several two hop nodes.

https://www.nycmesh.net/map

althea.net also runs a multihop wireless network so I can get some samples to discuss right now.

You can find a comprehensive overview of our design here.

https://www.youtube.com/watch?v=G4EKbgShyLw

But to get back on subject Althea has exactly the sort of network your interested in since we're focusing on building a distributed, incentivized, mesh WISP. More than half of our users are behind two or more hops.

Here's a rough sample of some two hop users.

   fd62:e3a0:6100:ae2:2e0a:d8a7:7f25:294f metric  246  RTT  26.976  price  42840
   fd64:30f4:7ea7:5700:f391:6c14:e9c7:6f11 metric  193  RTT  15.750  price  42840
   fd65:9f14:c542:d97e:88dd:13ff:99f8:20d7 metric  239  RTT  29.966  price  42840

Most of these links are degraded by trees and distance, causing higher than ideal latencies. The users are ecstatic though, their only other options are 1mbps DSL or sat internet. No other WISP will enter the area because line of sight is so challenging.

I will also note this is literally the cheapest gear off the shelf. It gets 100mbps to the user but has serious problems with retrying packets it should just drop.

Each hop provides many degrees of freedom, we haven't found a need for more than two hops at all yet.

If we where in an urban area getting a couple of ms would be much easier, interference free 60ghz and shorter distances.

[NotSammyHagar]

Nullability sounds like a frustrated Comcast exec. The tradeoffs sound reasonable to me on the free alternative.