[gwright]

I'm not sure small ISPs are a viable business model like it was in 10-20 years ago, but if you want to build a small regional ISP, I don't think peering relationships are going to be the problem, you can simply purchase transit bandwidth from the big guys. To do that cost effectively you do need to be renting rack space at a carrier hotel though.

The really hard part is the layer-2 last-mile infrastructure to end users. For the most part this means dealing with the local incumbent telco (leasing copper/fiber/DSL/ATM backhaul) or building your own wireless infrastructure. Except in rare circumstance I think the regulatory environment is going make it infeasible to build out your own physical infrastructure to compete with the telco or with the local cable franchise.

Google Fiber was an attempt to roll out a new last-mile infrastructure and it hasn't faired well. It isn't going to be easy to disrupt the last-mile infrastructure market.

[karlshea]

I'm don't think that's true. US Internet[0] here in Minneapolis is privately-funded and they've been slowly building out an in-ground FTTP network. It's cheap (50Mb/$35, 100Mb/$45, 1Gb/$70, no contract), reliable, and fast.

They buy transit from a couple of different providers[1] and are co-located in an exchange downtown (MICE/Cologix) that's got the usual suspects like Google and Netflix.

I think the key is that they are growing responsibly and not trying to be everything to everyone but are just offering a pipe. Granted, this is a dense area so it's a lot different than trying to get fiber to a rural area, but so far that's what all the other fiber providers have been concentrating on.

They're certainly doing better than CenturyLink's pathetic rollout, and CenturyLink can just use poles instead of burying fiber.

[0] http://fiber.usinternet.com/coverage-areas

[1] https://www.reddit.com/r/Minneapolis/comments/7ju5ew/now_tha..., https://www.peeringdb.com/net/3996

[rayiner]

The US Internet model of "growing responsibly" in response to demand works. That model is also illegal in nearly every U.S. city. Municipalities require you to cover every neighborhood within some usually comically short time frame like 5 years. Indeed, US Internet has come under fire for this: http://www.citypages.com/news/us-internet-appears-to-be-redl....

[karlshea]

> Municipalities require you to cover every neighborhood within some usually comically short time frame like 5 years.

That is pretty unfortunate, because the USI model definitely seems to be working.

I don't buy the Powderhorn "redlining", their COO has been upfront about that neighborhood being on their map and the biggest issue with rolling out there has been finding a property for their network gear (which they finally seem to have found this year)[0]. Although I do get that it would be frustrating living there and watching the coverage map fill in around you.

[0] https://www.reddit.com/r/Minneapolis/comments/6oh5la/looks_l...

[gwright]

Dense urban areas are a special case and in these locations it may be possible to roll out your own fiber but the details are going to be very geographically specific (regulatory framework, conduit availability and ownership, etc.).

[karlshea]

Sure, that's all true, but dense urban areas let you cover the most people the quickest. And it doesn't explain why CenturyLink's rollout has been a disaster (although with USI undercutting them on price maybe they just gave up).

[iamnotlarry]

I visited a location this week where there were 2-4 houses per mile along country roads. The residents had a choice of ISPs. The house I visited had dialtone from a VOIP service.

Yes, it costs money to build out last-mile infrastructure. But even in very low density areas, it can pay off. Recurring revenues of $50-100/month per customer can in fact pay for the initial investment. Especially if the investment was a wifi tower. It's the kind of thing one motivated local can pull off.

Building a worldwide network of peers who all respect NN is much much more difficult. Even if you could set up some organization like EFF to define some "Bill of Rights" and then certify conformant ISPs with a cool logo of approval, customers will not choose ISPs based on that logo. Average American consumers will not demand NN.

So your local ISP--even a well-intentioned one--will do as you suggest: "Simply purchase transit bandwidth from the big guys." There ends NN.

[amorphid]

A relative of mine lives in a rural area of Washington State, USA. He has fiber internet through ToledoTel...

http://www.toledotel.com/our-services/internet/

...it's expensive, but he has it. Every time I hear someone say we can't have blazing fast Internet in San Francisco, it's never phrased as a question like "would you be willing to pay $MONTHLY_PRICE for 1gps fiber?" So far, I've only heard "Building fiber in San Francisco isn't possible because of $UNAMBITIOUS_EXCUSE."

[Spooky23]

That stuff was all part of the Obama era stimulus bills. Many folks with the ability to lay fiber did so with substantial capitalization from the Feds. Many municipal governments and counties built fiber rings as well.

The result was mostly institutional connectivity. Rural prisons, hospitals, government, factories, and schools got connected. But local franchise agreements make the last mile pretty much impossible.

[gwright]

Huh. ToledoTel is the telco it its service area. It isn't competing with the local telco. I don't think this is an example of municipal networking.

[Spooky23]

Sorry for the confusion in my response. I didn’t mean to imply that.

Rural telcos had access to this money for sure.

Municipal government really benefited as well — for its own purposes, not municipal broadband.

[gwright]

It looks like ToledoTel is a "Local Exchange Carrier" (LEC). These entities have existed for a long time and are effectively the local incumbent telco and are quite common in more rural areas.

Purchasing IP connectivity from ToledoTel is the same thing as purchasing IP connectivity from Verizon -- they are both the incumbent telco in their respective geographic areas.

[gwright]

Perhaps I should have said "analogous to" and not "the same thing as". I'm not trying to put forth an opinion here, just trying to point out that ToledoTel is the phone company in its service area.

[gwright]

I'd be interested in knowing who was really providing the layer-2 connectivity. In most places your options are the local telco and the local cable company. What were the other options here?

It is possible in some places to get IP transit from someone other than the telco or cable company but almost always that is happening using the layer-2 infrastructure from the telco. I've never heard of that happening via cable but I suppose it is possible.

I've been out of this for a while, but in Connecticut it used to be possible to lease layer-2 connectivity from the telco and then provide the IP services on top of that. This was done via DSL to the end-user with the layer-2 traffic aggregated and backhauled via ATM virtual circuits to the ISPs router in some data center. It was still difficult to compete with the local telco -- who also was selling IP transit of course via the same methodology. Mainly the problem was the economies of scale. The pricing favored high-volume which is hard for a small company competing against the market visibility of the local telco.