Hacker News new | past | comments | ask | show | jobs | submit login

I've worked for two ISPs during my career. One in California and another in Rwanda. Those experiences have convinced me that providing internet service just isn't a viable business: the margins are too thin, or non-existent.

Generally, I haven't seen private industry do a good job at providing infrastructure over long periods of time.


That said, when I see articles like this one I have to remind myself that the landscape is going to completely change once Starlink and/or OneWeb become active.

Most reports are that major ISPs are highly profitable. Undercapitalized local ISPs always seem to struggle, however -- as do companies like Google when they aren't serious.

I am not that excited about the satellite model.

The LEO satellites are not going to be very big or high powered. The ground terminal is probably going to be a phased-array job that costs $6000 or so. The satellite is going to be a bent pipe to Verizon, AT&T or whatever rent seeker is operating in your neighborhood. If people from NYC who can afford to smoke dollar bills to spite the cable company are using it, they'll interfere with people who don't have a choice 700+ miles away.

The LEO constellations might hark back to the pirate days of satellite when the dish cost $6000 but there was no subscription. Unfortunately the people who are behind them don't realize that normal folks aren't going to have money for an expensive subscription after buying an expensive antenna.

> Most reports are that major ISPs are highly profitable. Undercapitalized local ISPs always seem to struggle, however -- as do companies like Google when they aren't serious.

What reports? Verizon’s wireline division reports a 5% operating profit margin in a good year. It’s been 15 years since Verizon started deploying FiOS, and it’s not clear they’ve even recovered the initial capital cost yet. Charter, the biggest pure cable company, reported massive annual losses in the years leading up to the merger with TWC.

The idea that wireline broadband isn’t “highly profitable” doesn’t pass the smell test. Everyone is trying to get out of wireline.

If they were trying so hard to get out of wireline, then why are trying to keep governments out of it?

For the most part they’re really not. See my comment above—the article really blows the restrictions in many states out of proportion. That aside, nobody wants the government coming in and using tax dollars to drive them out of business,[1] even if the business is a low-margin one they’re looking to get out of.

[1] And that’s what will happen. Municipal broadband will kill any private competitor, because you can’t compete with a tax-subsidized alternative. If that happens, folks better pray that their municipal system doesn’t end up like say the DC Metro or New York Subway (initially good, but neglected in the long term and allowed to decay).

What alternative does a community like Cleveland [2] have to municipal broadband?

[2] https://arstechnica.com/information-technology/2017/03/att-a...

I was shocked by one of the comments saying that the only reason those neighborhoods have utilities like water and sewer now is that they didn’t used to be poor.

Let me turn that question around: how can a community like Cleveland afford to run a municipal broadband network, if the reason AT&T won’t expand service there is that people don’t have the money to pay?

Our approach to broadband in poor urban neighborhoods is completely screwed up. We blame the telecom provider for not building in areas it’s not profitable to service, or force them to do it to be allowed to service the city at all (and as a result, nobody gets service).

Broadband access in poor neighborhoods is a matter for the general welfare system and should be paid for with federal and state tax subsidies, just as we do for say schools in poor urban neighborhoods. Then, we can make a reasoned cost-benefit analysis about how to best use that tax money. (For example, maybe wireless is a more cost-effective way of providing a safety-net level of service than fiber.)

You might be interested in the story of Dennis Kucinich and Muny Light. Years later, it actually did turn out to be the right choice (and I’m no Kucinich fan). Maybe something similar could be done with wireless and the utility poles.

> It’s been 15 years since Verizon started deploying FiOS, and it’s not clear they’ve even recovered the initial capital cost yet.

They got a pretty good chunk of cash from us to cover much of those costs.

They did not. If you dig into reports about “subsidies” to Bell Atlantic (Verizon’s predecessor), you’ll see that the “cash” is actually “excess rates” resulting from telecom deregulation. There is no economic substance to the argument, it’s based on the idea that the rates and profits of Bell Atlantic should have been the same during the period when broadband demand was exploding as it was during the period when a phone line was just a phone line. (Then you call any difference between the two a “subsidy”).

How are they not viable businesses? Even if you remove the legal barriers there is a high cost to starting an ISP with the amount of physical capital that needs to be set up. Its not like its some business where a competitor can spin up on a whim and you're forced to deal with incredible amounts of competition.

Sure the ISPs profit would likely go down, but it wouldn't go to zero. Just like every other capital intensive business on the planet

Neither of the ISPs that I worked for had legal barriers that I noticed (especially in Rwanda), the barriers that I did notice were physical capital, as you mention.

You do raise a good point about "viable business", what I was trying to communicate was "a business where both the business and the customer benefit", and as you pointed out, that's not what "viable business" means.

I can only speak from my experience here, and my experience was that the financial constraints on both of the ISPs I worked for were such that many, many compromises had to be made. For example, the ISP in California used consumer-grade hardware, so the ARP tables would fill up after about 48 hours, so we wrote scripts to reboot the routers every 24 hours - this killed all active TCP/IP sessions and stopped service for about a minute. The ISP in Rwanda only had a ~3mbps satellite link, so we set up a transparent proxy to ignore HTTP headers and aggressively cache data to increase the apparent throughput.

These kind of compromises stand in contrast to similar work I did for a community college, where we didn't have to make compromises like that.

Personally I think that as soon as one of the promised satellite constellations becomes real enough to compete with Comcast/Time Warner/etc, we'll see new legislative barriers (bought and paid for) come into play.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact