One of the better ideas to handle "internet as a utility" is actually not to scrap private ISPs, but to give government a monopoly over the physical infrastructure in the "Last Mile" (i.e. fiber from central hub to business/home).
ISPs lease the Last Mile at near cost (to pay for physical maintenance/improvements) and then are responsible for peerage agreements, subscriber management, network equipment, data budgets, plans, customer service, and other services (e.g. VoIP, TV over IP, etc).
Nobody benefits from ten+ different fiber runs from data center to home or business, it is inefficient. So you have to consolidate it and then lease it, that way granting an actual competitive landscape where consumers would get many choices for their ISP (and their ISP is responsible for mediating with the government about physical issues/faults).
Other countries have done this successfully. It also keeps government from being responsible for actually providing internet services, which could have civil rights problems (or politicians trying to decide winners/losers, etc).
While I agree wholeheartedly with the sentiment of having a single last mile connection to each home, IMHO I think giving the government a role is a really bad idea. As a FTTH engineer that has deployed millions of feet of last mile optical fibers, I can tell you government is clueless as to how best do this (as I might add are the MSO's). The majority of small FTTH ISP's have received funding through the USDA RUS, which has specific engineering requirements. Those requirements are prehistoric and dramatically inflate the cost of deployment. For example, one of the better topologies is to implement fast spanning tree to each home over 2 route-diversified connections back to a municipal ring. This dramatically reduces cost vs a star topology build-out (typically endorsed by RUS), and increases fault tolerance. The RUS estimates cost per home well in excess of $1,800 while it is possible to reduce that to well under $1,000 using appropriate engineering. One of the big problems is that most larger organizations (e.g. MSO's) have separate departments for physical networking (buried/aerial cable) and for Ethernet networking (spanning tree etc). As these departments do not communicate well at the design/planning stage, they do not enjoy engineering economies. The MSO's actually believe quite sincerely that their cost of FTTH is at least an order of magnitude higher than it is - they are just plain wrong.
Dude, I live in the heart of Silicon Valley and have dealt with AT&T and two different offices and my home. And I can tell you that AT&T are both utterly clueless and shameless liars when it comes to FTTH and FTTN installations. You can not imagine the depth of my anger with those clowns.
I have no idea who you work for. But my personal experience with trying to get fiber from AT&T in what is supposed to be one of the most tech-forward regions of the world is laughable. The government may not be better, but at least I know where to aim my pitchforks and torches.
I'm sitting here on a Comcast connection that is expensive and AT&T has FTTH in my neighborhood. I just can't seem to pull the trigger on the order because of the horror stories I've heard, and from personal experience. When I had ADSL from AT&T, they absolutely could not give me a connect that didn't drop randomly 2-10 times a day. I had 10+ visits from them over several years and at one point they told me "don't call us anymore" because I was too expensive for them. This, after being told by someone that put a device on my line and found the problem was in the drop cable (from the pole to the house). The new drop cable was ordered and a guy shows up 2 hours later and tells me it's not the drop cable (he didn't measure anything). When pressed, he told me it was too expensive to replace the drop cable. A subsequent tech told me to have a "tree trimming accident" where the drop cable was severed. I was worried I would be charged many thousands of $'s so I didn't do it and just switched to Comcast. (I have had my own issues with Comcast, and I hate them with a passion.)
The situation with home internet is just so shitty.
FWIW, I have had the exact opposite situation in multiple locations across the country -- every DSL (and, once, FTTH) connection I've had has been rock solid -- every Comcast internet connection has been flaky -- in one place, the connection went down every afternoon. Multiple Comcast technicians couldn't figure out the problem.
I so agree with your last statement.
If you're in an area they serve, getting Sonic is much better, BTW. They can operate over AT&T FTTH and DSL. It's a tad more expensive, but the customer service is unbeatable.
Sonic uses AT&T in my area, so I will definitely check them out. Thanks. It's a little annoying I have to get a landline, which basically adds $15-20 to the price. <sigh> I have an asterisk/freepbx system that I will continue to use. I wonder if their TOS require you to hook up a phone. Not looking forward to reading that...
Without debating the technical/cancer type of issues... Isn't 5G supposed to finally make wires obsolete? You'll put a device in your house (from say, Verizon) which speaks 5G on one side, and maybe has an ethernet jack and/or speaks Wi-Fi on the other side.
In this case it'd likely be talking about corrupted, truncated, or otherwise mangled packets. So having to retransmit things. You'd still technically have the same bandwidth and latency on the connection itself even if it's causing all kinds of overhead on upper levels of the stack.
After 6 years of mobile internet with only 1GB per month my ISP basically trained me to not use it for anything unless absolutely necessary. The irony of course is that prices have dropped but I'm not going with a bigger plan. I switched to a "budget plan" and stayed on 1GB per month for which I pay only 5€ per month.
In addressing this at a political/regulatory level it became clear to me that while it is convenient to beat up on Comcast (however deserving they may be of such treatment), the real problem on the east coast corridor are the trade unions. They have raised the cost of doing business by at least an order of magnitude. It is a heartbreak for me personally - my grandfather was highly involved in the labor movement back in his day and today he would roll over in his grave at the rampant corruption.
If you're not looking to spend that much, Comcast gigabit (non-pro) is cheap and gets you 1gbps down, but only 35mbps up. They're annoying to interact with, but once it's set up it's reliable.
Pro-tip: Ignore the Comcast website prices/packages. The only way to work with them is to call and get a friendly sales rep, then ask them to look through their options to find the best gigabit only plan (they have a lot more promotions available than they list online). This might take a few tries if you get someone on the phone that isn't very good. Comcast Twitter support is also pretty helpful.
Wow, this is awful. I live in Portland, and have symmetric gigabit fiber from CenturyLink for $65/mo. Honestly baffling to me that people tolerate the politics of Silicon Valley.
I come from Western New York, where (at the time) 20mbps was the best available option and it was over $100. I hear things are a little better now (though 100mbps may still be the best available), but the options here are way better.
Sure it's a great speed, but $300/mo plus a $2k install just to get a decent upload? Seems like market segmentation where they're basically just ripping anyone off who wants to run a little server or backup their files to the cloud.
Yeah, it'd be nicer to have symmetric gigabit for less, but the gigabit pro is actual fiber to the home (it doesn't run over coax lines).
Comcast includes a Juniper ACX2100 switch for layer 2 and bringing the fiber into the house. I thought that was pretty cool. (Well 'includes', you have to pay to rent it from them.)
I'm in East Bay and also on Comcast gigabit non-pro only (no TV). Signed up at $90 with no contract and so far it's been as stable as I had in Seattle with CenturyLink (fiber to building). I did buy my own DOCSIS modem so that could be part of the equation.
That is not gigabit. That is 35MB. Big difference. Internet MUST be symmetrical. There are too many essential peer-to-peer technologies for it to be anything but symmetrical.
Gigabit Pro is symmetrical, normal Gigabit is not. Gigabit Pro is FTTH (whereas all of Comcast's other plans go over copper/DOCSIS3), and is symmetrical 2Gbps. For $300 per month with a mandatory $2k install, I might add.
to me, a symmetric connection is more of a "nice to have" than a "need". I have something like 400/25 right now. if I had 400/400 or even 1000/1000, I might consider reconfiguring my plex server to stream outside my LAN, but aside from that I can't think of any obvious application. can you list some of these "essential peer-to-peer technologies" that a normal person would be likely to use?
I totally agree. The prevailing perception that Silicon Valley (the original MAE-West) and Virginia's MAE-East are the touchstones of high bandwidth Internet is not true either - while they are the peering points for (inter)national networks, the local access Internet offerings in those regions are indeed laughable. Meanwhile the small rural towns in fly-over country tend to enjoy networks that are gigabit to the home or better for under $100 MRC. I know of businesses in Philadelphia that had to relocate to Winona, Minnesota simply to lower their Internet costs from thousands of dollars per month (with Comcast) to under $200 MRC for a direct gigabit optical Internet connection.
> AT&T are both utterly clueless and shameless liars
Seconded. We had AT&T Fiber a few years ago, and were having techs out just about every week to fix it and were given a nonsense reason like "this cable was loose" each time. Once we got fed up and canceled service, the tech told us that corporate fucked up the install for the whole neighborhood so they'd basically send techs out, tell them to lie, and have them add whatever bandaid was needed to fix it for just a little longer.
The government also has no idea how to build roads or power plants. They contract it out, and then take responsibility for owning and maintaining the infrastructure long term.
Depending on the city and area, opinions would vary if they’re good at maintaining it... but considering the huge amount of space the USA takes up, and the large amount of interstate and highways that criss cross the country generally being quite functional, I think they do a pretty good job.
And this does not work so well either. The problem with many government contracts is not just that they are usually cost plus, but that they do not take the total cost of ownership into account. In your example of roads, if the government put out requests for proposals based on the lowest total cost of ownership of a road over say 100 years, we would have roads engineered to a much higher standard that would not need to be rebuilt at such a high cadence. Whenever government thinks they know how to specify the best engineering method to solve a user requirement, they are generally wrong.
That sounds like a bad solution to a bad problem. The better way to set things up is to devise a process in which the incentives are from the default set up to benefit the most people.
People are upset at what trump is doing in relation to the USPS. Why would we want his administration in charge of broadband as well?
> Whenever government thinks they know how to specify the best engineering method to solve a user requirement, they are generally wrong.
It's not that they're usually wrong, they simply aren't attempting to optimize towards that at all. Most The government tends to make decisions optimized against political merit, not engineering efficacy or solution fit. The reason they generally appear wrong is because generally engineering efficacy and solution fit have little if any positive correlation with political merit.
I wonder if governments should not just find experts to design interfaces / protocols and cheap testing organs, then let competition run wild in these constraints.
>IMHO I think giving the government a role is a really bad idea.
You've described why the government shouldn't be the entity making sole architectural decisions - you haven't explained why the government shouldn't own and fund the last-mile fiber.
My county "owns" all the roads around me - city employees aren't building the roads and more often than not aren't the ones architecting them either (slope, water mitigation, etc). They do maintain them after they're in place (plowing the roads/mowing the ditches).
Why would fiber need to be any different? I don't think anyone expects city employees to actually trench the fiber, that would all be outsourced. If they're a large enough entity they could very easily train and hire some fiber splicers to fix cut lines.
I've got no problem with government owning and funding the Last Mile. The problem arises when government demands they specify the network design because of that ownership and funding. They are not qualified to make that decision, and usually senior telecommunications engineers are not qualified either. There is simply too much controversy over what is optimal. If the government could simply specify that a network needs to deliver 10Gbps per customer, then that would work, but when they specify HOW those packets need to be delivered then that is when things become non-competitive. There is a huge difference in cost between trenching from a central office to every home in a star topology vs trenching from a municipal network (already paid for by schools, hospitals, large business etc) to homes in a fast spanning tree ring. At least an order of magnitude difference in cost.
In my (limited) experience, that fear is unfounded. I work with about a dozen local government entities and every last one of them reaches out to consultants when making architecture decisions. I've never once run across one that pulls the "it has to be this way because we say so" in the tech realm. They'll absolutely provide requirements, but they won't dictate how those requirements are met.
I guess I would be shocked to find out there is any significant number of local governments out there that demand to dictate architecture of something they have no expertise in. If ANYTHING, the times I see them get out over their ski's is when they trust a vendor consultant TOO much. See: every Oracle implementation ever.
Comcast in Seattle came out to fix a signal issue at my apartment and ended up possibly needing to redo a rats nest of cables chained together just to find the line running to my place. Maybe the connection to the building was good, but the infrastructure inside (all Comcast built) was 12-15 years old and flaky at best.
At least with federal ownership we can mandate maintenance and repair requirements or timetables along with budgeting to do so in a timely manner. Or at least stop passing free money to ISP CEOs by giving millions of taxpayer dollars to them for upgrades and expansions that never come. I'd rather give a billion dollars to the Army CoE or a similar group to get things installed and maintained instead of profit to an ISP and at best slow expansion and data caps.
Thanks for sharing the knowledge here. It's not clear to me how the MSOs misunderstanding of FTTH/Last Mile costs means government can't do the job. I read this as what we have right now isn't working, which to me suggests we should try something new.
LOL because the MSO's own the government. One of the most powerful lobbying forces in the US. They control the FCC. Under the current administration they have finally overthrown net neutrality - https://en.wikipedia.org/wiki/Net_neutrality#United_States
The key thing is ownership. I was part of a large government network buildout a long time ago that was like an ISP for state/local that also provided secure connectivity.
We owned lots of fiber, which was managed by third parties. The downstream customers benefited from better terms and buildout of the last mile infrastructure before private entities were doing it.
If we had the political environment that allowed for old style utility regulation, I would agree that commercial utilities would be better - the old dial up ISP market demonstrated how robust and effective competitive market prices were. But we do not have that environment today.
Interesting comment! I think you’re describing a surmountable barrier.
If local governments can own the roads and power utilities are forced to share power infrastructure there is no reason municipalities shouldn’t be able to own their last mile installation.
If local government put a contract out to tender for implementing last mile so that small companies could bid on it then the $1k company would win over the $1,800 company. This is how most things should work in local government (corruption aside) .
A competitive market with a relatively low barrier to entry for last mile cable would benefit everybody. Everybody except AT&T and Comcast that is (& their millionaire shareholders in congress).
Even a nationally owned and run last mile would be better than what we have now though.
I don't know about the US, but in the EU a tender has requirements. The topology of a network is a perfectly reasonable requirement so they couldn't compete on that (even if it's wrong)
A tender for a brick house is, rightly, not going to be won by a wooden house proposal, even if it's cheaper and the company argues it's just as good.
We have a similar system for electrical distribution; per area a single organisation has sole responsibility for the physical infrastructure and they bill your energy company of choice. That means that on your invoice you have a fixed amount of infrastructure cost as a component, and your service/usage fee from the energy company. Same for gas.
This is probably common (or at least more common than internet-infra-as-a-utility), but the concept of doing the same for data would benefit a lot of people. We had that for POTS/PSTN and Coaxical data (CATV/cable/whateveryoucallit) as well, which was privatised and now opened up again so you get your infrastructure component as a national fixed fee and you simply pay your provider of choice for the actual service.
I found this terminology confusing for many years, hope this clarification is helpful:
- “Coaxial” refers to the design of the physical cable (Wikipedia has a better explanation than I can offer)
- CATV refers to “Community Antenna Television”, the first cable systems in the USA, which fed the signal from a shared “community antenna” to homes in areas with poor broadcast reception.
- “Cable” is the generic term for pay-TV and ISP services over coaxial cable
Yeah, the problem is indeed that everyone points to the same general concept but isn't using the same terms which adds to the confusion.
Technically you are using a DOCSIS connection which happens to use a coaxial cable. But not the same cable one as you'd have with CATV because that one didn't have the same ratings. There is also a difference with one-way broadcast and two-way communication where you also need a CMTS for that purpose. And to make it worse, sometime EOC is used which in itself is Ethernet, but not over CAT5/6/7 cables.
However, all of those words are better than people calling it 'the internet wire'.
Is the US government at all capable of taking on this role though? They apparently already wasted $400 Billion in contracting out broadband upgrades to telecom companies who pocketed the money and never delivered, and then everyone just forgot about it and moved on.
I try to be an optimistic person, but years and decades of gridlock and getting nothing done in the US is definitely wearing on me. There are so many things in this same category, of "We have Big Problem X", here's what research has shown works and doesn't work, here's how other countries are doing it more successfully, here's a very clear solution - but no, we'll just talk about it on and off every few months or years and then let inertia win and never actually do anything.
Why would the US government be doing this? Surely local government is better suited to these decisions?
> I try to be an optimistic person, but years and decades of gridlock and getting nothing done in the US is definitely wearing on me.
That's just like, your opinion.
We do a lot of things well in this country, you just have to be willing to look. Public infrastructure like roads and power seem like they have plenty of examples of what works to get this built out.
I don't see why any other utility model needs to change significantly for last mile.
Some states have regulated so municipal ISPs can't start up or have no control. I haven't read an article saying Municipal ISPs are any worse than commercial and often cheaper and less hostile too. Funny how that happens when the profit motivator is taken away.
I'd argue that ISPs end up using infrastructure that is common (think tragedy of the commons) like a water pipe or power line and should be open or publicly owned/maintained. By exploiting exclusive ownership of a common resource, ISPs become rent seeking and suck cash from both customers and taxpayers. As long as they hold ownership of the lines, they can charge what they want or you have to pay a huge cost for a new fiber line or have no access.
So seize the lines, just like water and power, and allow well regulated companies to facilitate the utility like many states do for power. Allow cities to push for municipal control and management (ex taking over line maintenance in exchange for providing services. Force it into a system where it pays for its own upgrades vs millions to execs and share holders, just like health insurance in many countries.
Power is a nightmare! We just precariously string up more high voltage lines on poles because we can't figure out how to do street construction at a reasonable cost or on budget. These now cause massive wildfires every summer, as well as frequent outages from random things like balloons hitting the lines.
San Francisco had a plan and still has a tax in place to bury power lines, but they pre-spent all the money in 2007, made much less progress than expected, and haven't made any additional progress on it since.
More generally, infrastructure like trains and bridges are incredibly difficult, unpredictable, and expensive to build now. The California high-speed rail and Caltrain extension are effectively indefinitely postponed at this point. It's a perfect example of exactly what I was saying - it's a problem that has been solvable in the past, there are examples of other countries doing a much better job of it, but the US is completely unable to do it.
Where exactly do you look at infrastructure projects in the US in the past 2 or 3 decades and see this success story you're talking about?
You don't need to give the government a monopoly on the last mile, you just need to allow them to compete on it.
As a Swede living in Japan, that system has worked well in both countries. In Japan you have the former telco monopoly NTT with its national flets network and competition in the private au, Sony NURO and various local networks. In Sweden you have many local municipality-owned companies as well as national competition fighting over customers.
It seems like Americans are always see options as black and white. The same thinking is reflected in your politics
For Japan, Flets (NTT) service is top and widest area FTTH service and it runs on NTT's own fiber. au Hikari (KDDI) is 2nd major service and it uses KDDI's own fiber and partially NTT's dark fiber. Nuro (So-net) is relatively small provider (due to limited area) it uses mainly NTT's dark fiber and AFAIK no own fiber. Other players like eo (K-opticom) are mainly by power company and area is limited.
Separating last mile ownership from ISPs was how it worked around the turn of the century with DSL in Seattle.
US West owned the lines to your home. Your DSL modem communicated with the DSLAM at a US West switching center, and US West relayed raw bits between you and your chosen ISP over something lower level than internet (ATM?).
US West essentially provided an unstructured dedicate virtual circuit bit stream between your DSL modem and your ISP, and your ISP provided internet on top of that.
Your ISP charged you for providing internet access, and US West charged you a fee for using their line.
I learned just the other day that a group of Utah cities have banded together to create such a network. It's called Utopia Fiber - https://www.utopiafiber.com/about-us/
They are also avoiding the GPON approach and running fiber direct from "fiber huts" to homes so that they can simply use standard networking equipment to service customers.
At this point, are the ISPs adding a significant amount of value? How much work does it really take to set up peering agreements?
I am absolutely ready to be convinced otherwise, but my gut reaction to this strategy is that it sounds like a needless carve out for private industry. If 98% of a project is taxpayer funded, private companies shouldn't be able to charge whatever they like for the final 2%. Maybe competition would stop that from happening, but why take that risk? What is the benefit?
You can achieve the same or better result but simply mandating that the same company can't own the fiber AND provide service. The electricity grid in Europe is run the same way.
I don't understand why you want to preserve these monopolies. They're already non-competitive. How on earth does granting private monopolies do anything other than give away free money to unaccountable executives? It's already clear, based on this suggestion, that the private "market" has failed so the government needs to step in. Just take these guys over and run them publicly.
The problem is that not all bits are the same, even in the last mile.
There is still significant improvement in last mile delivery, where different technologies enable different layouts and possibly different cost structures.
There isn't even a safety reason to make this a monopoly.
The monopoly in Internet provision is in the right of way. If new providers were able to construct their own last mile networks, then we would see more providers.
Something similar is done where I live, except the last mile is not owned by the government but by a commercial corporation. The government basically states open-access and a level playing field for costs as a requirement for giving the necessary permissions to dig and install fiber.
The last mile is an outdated notion. Many people would have mobile internet these days and not really need a separate connection at home. Where I live (Berlin, Germany), 5G connections are actually faster than anything I can get to my house at the moment. And I can only get it from the same provider responsible for deploying fiber optic as well (just not to my house). Which of course would be the local state backed monopolist Deutsche Telekom; better known as t-mobile in the US.
Many developing nations never had any last mile to begin with and probably will never have any. There are more phones than people in this world. Multiple companies are bootstrapping satellite networks to serve internet all over the planet. Going forward, the only reason to have a wired connection into your house would be monopolist telcos hogging the locally available radio-spectrum backed by lots of government protection.
So, instead of governing the status quo of 15 years ago, governments should focus on ensuring there are no infrastructure monopolies on mobile internet. That means ensuring multiple providers are active and that no barriers exist for customers to switch provider. This drives them to compete on quality rather than luring in customers to years long contracts and lots of hurdles to get out of those; which sadly is the status quo in many countries.
The only reason sim cards still exist is because they are a key control mechanism. Soft sims were proposed early this century already and the only reason for sim cards to continue to exist is that it 'solidifies' your relation ship to your local telco monopolist. It's the same reason they make you obsess over owning a phone number, which is an ancient relic from the last century.
I strongly disagree with this push to abandon hard-wired internet. I have never found wireless internet to be fast or reliable. My parents had it for years, and then the trees became too tall, and it stopped working. I live in the middle of the city, and yet I don't get LTE service in the back of my house for some reason. Even when I do get it elsewhere, it's slow as shit 50% of the time. I'll keep my gigabit fiber, thanks.
That's a weird take on SIM cards. I used to work in the wireless industry and the primary benefit was that you could easily swap carriers by swapping out your sim, but you could keep your phone. Software based SIMs sound like a security risk to me.
ISPs lease the Last Mile at near cost (to pay for physical maintenance/improvements) and then are responsible for peerage agreements, subscriber management, network equipment, data budgets, plans, customer service, and other services (e.g. VoIP, TV over IP, etc).
Nobody benefits from ten+ different fiber runs from data center to home or business, it is inefficient. So you have to consolidate it and then lease it, that way granting an actual competitive landscape where consumers would get many choices for their ISP (and their ISP is responsible for mediating with the government about physical issues/faults).
Other countries have done this successfully. It also keeps government from being responsible for actually providing internet services, which could have civil rights problems (or politicians trying to decide winners/losers, etc).