They've had a 1TB cap + pay per GB overages in basically everywhere they don't have competition already. For years. I've always hated looking at a several hundred dollar internet-only bill.
Vote with your wallet, switch providers if you can. If you can’t, put the time into spinning up a coop or muni fiber ISP in your community (or a locally owned for profit ISP even). Failing all else, StarLink.
I ran the second I could. Proper fiber is great. Sick of Comcast's 5 Mbps "up to" upload on their gigabit plan that they just use the word gigabit everywhere on but it takes 4 hours to upload a docker container.
well if its cable provisioned gigabit (docsis3) has limits on upload of 35Mbit. but i hear you, i pay for gigabit and get an average of 600mbps down for many months now
I live in a city which ostensibly has ISP competition, but it’s basically a fiction. Most of the city is in an enclave where one of the two major cable providers has exclusive rights. The parts of the city that require both providers to compete are dense, mostly apartments, and the providers get exclusive rights to each building. In most cases, the only other option is extremely slow DSL.
I’ve researched my options at every place I’ve lived and I’ve consistently been stuck with Comcast. It’s really a shame.
This doesn't make any sense. A Starlink satellite's bandwidth is shared with everyone under its radius, so any throttling/bandwidth caps is going to be even worse. A quick search says that each starlink satellite provides 20 Gbps each, which works out to 6.57PB per month, or around 5000 1.2TB/month users. This is the best case scenario, assuming that their usage is uniformly distributed per month. Most bandwidth use is bursty (think netflix in the evening), so in reality it's actually going to be much worse. Assuming a 4K stream at 25 Mb/s, one satellite would only service 800 users. Given comcasts' statistics (5% of users use more than 1.2TB/month), that would mean you need dozens of satellites to service all the heavy users in even a small town.
This was actually not an option in the past. In one of those pre-existing markets you'd basically just purchase from Comcast Business instead for something like 100M down, same 2-5M up on coax for a few hundred a month w/o cap (and with mildly better support but still absolutely no help and will just hang up on you if you say you use Linux or your own router or modem).
That page is new, and it is particularly obnoxious and sickening that the only way to get an unnecessary, arbitrary cap removed - the cheapest option is to use their hardware that you are not permitted to control, that emits Wifi hotspots that you cannot control, and remotely-controlled passwords. It's pretty much a Meraki rental that isn't yours. It's bordering on AT&T home internet's equivalence of "if you don't want us to sell all of your detailed DPI'd browsing history, add $50/m to your plan cost".
Internet usage should be based on a standard connection charge and then a per GB usage charge... this is generally how power companies charge for their service. If you think about it, the average users are subsidizing the minority that have multi-TB monthly usage.
They should also charge reasonable wholesale rates in that case if you remove the monopolization of the last mile. In other countries this is actually how it works already, the underlying cabling is municipal, you just pick an ISP to transit over it.
Comcast charges $0.20/GB for coaxial. Verizon charges $10/GB for LTE. My actual cost for hosting side projects is below $0.50 per 1 TB. It is unreasonable if they are going to fuck with your traffic, tamper with your HTTP payloads and inject scripts into them, deprioritize certain data and still charge multiple orders of magnitude more per GB.
It seems like you're comparing the cost of serving data out of a purpose built location to the cost of transporting it to and from an individual person. Not saying Comcast or Verizon are offering good or bad per GB prices just that it seems you're comparing the cost of your apples (unless your side project is a LOT larger than I'm assuming) and then saying the oranges are mispriced as a result.
I am, but this was with the implication that the last mile fiber was already built out, city/municipality-owned and you are already paying an access fee to use the cabling, therefore you are effectively just transporting packets to/from your own core anyway.
If they really want to align incentives it makes sense to switch to time-of-use based pricing like electricity providers have done. Since most of the cost is building infrastructure to support peak traffic, suggesting that consumers shift their usage pattern to be off-peak would be a notable cost benefit.
The obvious problem besides the added complexity for users is that I suspect there is relatively little "background" traffic usage. Almost all of it is people watching streaming video or interactive downloads. I doubt things like software updates that can be scheduled add up to much.
It would be interesting though. I wonder if it would lead to services promoting their download options more (example auto-download likely videos every morning).
That would make sense if ISPs' costs veried per GB like a power suppliers costs vary per kWh. But they don't really. 99% of the cost of them is running the initial fibre/copper and then maintaining it. Until you reach the limit of the infrastructure, the cost per extra GB is basucally nothing.
Quality of service is also not correlated with usage. packet loss, queuing and dropped connections result in higher usage and are directly the fault of the ISP. they can oversell a region and under provision and then charge excessive bandwidth fees because of it.
For comparison this works out to 3.7 megabits/sec, which is less bandwidth than most DSL connections! That means if you're a heavy downloader, you should seriously consider switching to DSL.
What I find especially galling is that the first 1.2TB costs half as much ($10 per 100GB, assuming a $120 monthly bill) than the subsequent data, even ignoring the fixed cost of the connection. I think if they merely charged a fair decreasing rate for usage, rather than trying to price gouge heavy users, they wouldn't be seen as such villains.
Alas, muni fiber for the win. Gigabit symmetric and forget about it.
Northern California has had the cap for a long time. I had to negotiate a plan for 45 minutes with them to remove the cap, which including going to 1gb/s which I don't need. I suspect this will be more common anywhere that does not have competition or where ISP's wink-wink agree not to step on each others territory.
It’s not even a wink-wink. In many states, for example, Connecticut[1], the territory of each. cable provider is regulated by the state. There’s only a few tiny spots that have an overbuild of another provider. The state claims that DSL is competition, but let’s be real, 12mbps DSL is no longer competitive.
If only 5% uses 1.2TB each month, why would Comcast want to impose the cap to everyone?
Comcast has already made enough money and this move has nothing but against social good.
I am an international student from China in US. If you ever go to China, you will find how affordable Internet (or Phone Data plans) is. And it is even affordable in sparsely-populated rural areas, unlike the US rural areas I've heard of.
These big private companies are just lagging US behind and the country cannot do anything with it.
If the hardware is made in china, the companies can get it for probably cheaper than anywhere in the world. Also the labor in china is dramatically cheaper. Is there a union of telecom workers in China?
Of course China internet is going to be cheaper. That doesn't mean there should be such a difference between USA and China prices, but not surprised there is some, nor would I expect it to be the same or cheaper in the US. US has higher impute costs. As a pct of income of avg person what is internet access in China vs US?
This of course leave aside the what constitutes the internet you get in China vs USA
Sounds like a bit of an unpopular opinion in this thread but I think pay-for-what-you-use plans with reasonable per GB charge and a slightly higher per GB charge above a certain number seems reasonable, irrespective of your opinion of Comcast.
Something like 30 dollars for a residential connection, followed by 2.5 cents per GB for the first 1024 GB, and then increased to say 4 cents per GB for another 1024 GB.
If you use 1 TB, you pay 30+0.025*1024 ~ 55 dollars.
Why should we subsidize those who download tens or hundreds of TBs?
No one thinks this is unreasonable for electricity, why is it unreasonable for data? Bandwidth issues affect data too.
This comment was posted two minutes ago and it's already showing up as grey to me. It's a shame, because what was said has merit, even though the conclusion might not be appealing to the average HN reader.
eg.
>Why should we subsidize those who download tens or hundreds of TBs?
Consumer internet is oversubscribed. By using 10x as much bandwidth as everyone else, you are effectively getting your internet connection subsidized by everyone else.
A reasonably high cap like this doesn’t seem completely unreasonable. But where the cap is set, who it applies to, when it’s updated and when it comes into effect are more important than the number of TBs.
For instance, setting it at 1.2 TB today seems fine, but in 10 years that won’t be nearly enough, but will Comcast update it? Or just push ‘heavy’ user to their ‘prosumer’ plans.
It’s similar in financial regulation. Transactions over $10,000 must be reported for KYC/AML purposes. Sounds fine when it was implemented, but now it seems very low.
It wouldn’t take much to future proof decisions like this, by default. Inflate by 10% per year, or some other metric. And the lack of such features hints at the fixed amount being the feature.
Higher prices per kWh after some limit is a regulatory tool to drive down energy use.
There's no regulatory reason to drive down data use. Given that, the 1st data allotment should include the cost of the last mile connection, and any subsequent data allotments should be significantly lower cost, closer to incremental transit costs (which aren't very much). Commercial internet connections are often billed at a monthly fee based on the interconnect type, and then a per Mbps fee for 95 percentile usage.
Most residential customers don't really want that though, they just want an unmetered connection that works, with a monthly price they can count on; and one where the more they pay per month, the faster it goes, and the more stuff they can do. A flat cap set across all of the tiers doesn't make sense. And how are consumers expected to control all of their devices to make sure they don't download hundreds of gigs of updates overnight?
> Why should we subsidize those who download tens or hundreds of TBs?
Why should those who download tens or hundreds of TBs subsidize the rest of the people?
End-user equipment, underground cables, and ISP's routers ports are the same regardless on the bandwidth. Construction work is expensive, you need American people to lay cables in American soil, can't outsource to India.
Compared to these costs, bandwidth is borderline free, even in the long run.
Your internet plan is already pay for what you use: you pay for a given rate of data transfer and your ISP maintains the infrastructure to provide you that rate. The cable you need for 50 Mb/s costs exactly the same whether you use it 2 hours per day or 20. Anyone downloading 100 TB in a month is already paying a hefty premium for the (at bare minimum) 370 Mb/s they need to achieve that download.
For electricity, in part you are paying for the infrastructure, but you are also paying for the fuel that's being consumed by the powerplant. If comcast charged per GB of data that it hosted, that would be fine (and hosting providers do exactly that without complaint), but Comcast's costs don't vary with the amount you download.
In the extreme, costs do scale to some degree with duty cycle - a line with lower capacity can service more people so long as they don't use it at the same time, and technically sending a packet through causes some infinitesimally tiny wear and tear. In practice though, a data cap doesn't stop multiple people from doing their 1.2 TB download all at the same time, and the reduced lifetime of components is a rounding error compared to the variability just from their manufacture.
i had a pay per use plan more than a decade ago. i liked it because the competition all had capped plans that cost the same as my pay-per use would if i reached that cap. and if i didn't, then i paid less. the service also allowed me to set my own cap so that i would not accidentally go over a chosen spending limit (and it allowed me to change the cap at any time)
So if you play any game streaming platform like Google Stadia which at high quality and framerate use an average of 20GB/h, you'll bust that cap in little under 62h, and that's if you do nothing else.
I'd feel better about this if they gave a discount (even a modest one) to all customers simultaneously. Then it would feel like the plan is revenue neutral and redistributive. They'd get some additional revenue from high-usage customers and give up some revenue from their other customers. Instead it feels like a heads-I-win-tails-nothing-changes situation.
44 comments
[ 1790 ms ] story [ 810 ms ] threadhttps://muninetworks.org/content/muni-fiber-models-fact-shee...
https://startyourownisp.com/
https://news.ycombinator.com/item?id=24424910
I’ve researched my options at every place I’ve lived and I’ve consistently been stuck with Comcast. It’s really a shame.
This doesn't make any sense. A Starlink satellite's bandwidth is shared with everyone under its radius, so any throttling/bandwidth caps is going to be even worse. A quick search says that each starlink satellite provides 20 Gbps each, which works out to 6.57PB per month, or around 5000 1.2TB/month users. This is the best case scenario, assuming that their usage is uniformly distributed per month. Most bandwidth use is bursty (think netflix in the evening), so in reality it's actually going to be much worse. Assuming a 4K stream at 25 Mb/s, one satellite would only service 800 users. Given comcasts' statistics (5% of users use more than 1.2TB/month), that would mean you need dozens of satellites to service all the heavy users in even a small town.
https://www.xfinity.com/learn/internet-service/data
That page is new, and it is particularly obnoxious and sickening that the only way to get an unnecessary, arbitrary cap removed - the cheapest option is to use their hardware that you are not permitted to control, that emits Wifi hotspots that you cannot control, and remotely-controlled passwords. It's pretty much a Meraki rental that isn't yours. It's bordering on AT&T home internet's equivalence of "if you don't want us to sell all of your detailed DPI'd browsing history, add $50/m to your plan cost".
Comcast charges $0.20/GB for coaxial. Verizon charges $10/GB for LTE. My actual cost for hosting side projects is below $0.50 per 1 TB. It is unreasonable if they are going to fuck with your traffic, tamper with your HTTP payloads and inject scripts into them, deprioritize certain data and still charge multiple orders of magnitude more per GB.
The obvious problem besides the added complexity for users is that I suspect there is relatively little "background" traffic usage. Almost all of it is people watching streaming video or interactive downloads. I doubt things like software updates that can be scheduled add up to much.
It would be interesting though. I wonder if it would lead to services promoting their download options more (example auto-download likely videos every morning).
What I find especially galling is that the first 1.2TB costs half as much ($10 per 100GB, assuming a $120 monthly bill) than the subsequent data, even ignoring the fixed cost of the connection. I think if they merely charged a fair decreasing rate for usage, rather than trying to price gouge heavy users, they wouldn't be seen as such villains.
Alas, muni fiber for the win. Gigabit symmetric and forget about it.
Don't DSL connections also come with caps?
I believe cable caps arise from worries about last mile distribution segment capacity, and have less to do with actual Internet transit.
[1]: https://portal.ct.gov/-/media/PURA/Cable/VideoProviderServic...
If only 5% uses 1.2TB each month, why would Comcast want to impose the cap to everyone?
Comcast has already made enough money and this move has nothing but against social good.
I am an international student from China in US. If you ever go to China, you will find how affordable Internet (or Phone Data plans) is. And it is even affordable in sparsely-populated rural areas, unlike the US rural areas I've heard of.
These big private companies are just lagging US behind and the country cannot do anything with it.
This of course leave aside the what constitutes the internet you get in China vs USA
Something like 30 dollars for a residential connection, followed by 2.5 cents per GB for the first 1024 GB, and then increased to say 4 cents per GB for another 1024 GB.
If you use 1 TB, you pay 30+0.025*1024 ~ 55 dollars.
Why should we subsidize those who download tens or hundreds of TBs?
No one thinks this is unreasonable for electricity, why is it unreasonable for data? Bandwidth issues affect data too.
https://news.ycombinator.com/item?id=17426607
eg.
>Why should we subsidize those who download tens or hundreds of TBs?
Consumer internet is oversubscribed. By using 10x as much bandwidth as everyone else, you are effectively getting your internet connection subsidized by everyone else.
For instance, setting it at 1.2 TB today seems fine, but in 10 years that won’t be nearly enough, but will Comcast update it? Or just push ‘heavy’ user to their ‘prosumer’ plans.
It’s similar in financial regulation. Transactions over $10,000 must be reported for KYC/AML purposes. Sounds fine when it was implemented, but now it seems very low.
It wouldn’t take much to future proof decisions like this, by default. Inflate by 10% per year, or some other metric. And the lack of such features hints at the fixed amount being the feature.
https://www.littler.com/publication-press/publication/mexico...
There's no regulatory reason to drive down data use. Given that, the 1st data allotment should include the cost of the last mile connection, and any subsequent data allotments should be significantly lower cost, closer to incremental transit costs (which aren't very much). Commercial internet connections are often billed at a monthly fee based on the interconnect type, and then a per Mbps fee for 95 percentile usage.
Most residential customers don't really want that though, they just want an unmetered connection that works, with a monthly price they can count on; and one where the more they pay per month, the faster it goes, and the more stuff they can do. A flat cap set across all of the tiers doesn't make sense. And how are consumers expected to control all of their devices to make sure they don't download hundreds of gigs of updates overnight?
Why should those who download tens or hundreds of TBs subsidize the rest of the people?
End-user equipment, underground cables, and ISP's routers ports are the same regardless on the bandwidth. Construction work is expensive, you need American people to lay cables in American soil, can't outsource to India.
Compared to these costs, bandwidth is borderline free, even in the long run.
For electricity, in part you are paying for the infrastructure, but you are also paying for the fuel that's being consumed by the powerplant. If comcast charged per GB of data that it hosted, that would be fine (and hosting providers do exactly that without complaint), but Comcast's costs don't vary with the amount you download.
In the extreme, costs do scale to some degree with duty cycle - a line with lower capacity can service more people so long as they don't use it at the same time, and technically sending a packet through causes some infinitesimally tiny wear and tear. In practice though, a data cap doesn't stop multiple people from doing their 1.2 TB download all at the same time, and the reduced lifetime of components is a rounding error compared to the variability just from their manufacture.