I did not say that. My point was that Comcast kept pace with innovations in silicon and networking. Like AWS and GCP or Azure keep upgrading servers from Pentium to i7. They had to keep pace with technology advances and demands. Sometimes their investments were to lower costs, more customers supported by same dollar investment at upgrade time.
Comcast can try to increase profits as much as they like, the regulators and people are not interested in making google and facebook pay because they have more money.
Berkshire Hethaway doesn't own NASDAQ money because they are making a killing. There are plenty of startups that fail, Comcast is not subsidizing those startups. Plenty of startups that make it big, Comcast doesn't get to add their tax on that success.
That's one way of looking at it. Note that I do not think Comcast would successfully turn off youtube for any customers without creating a major uproar.
On the other hand, suppose I invent a game that mirrors the contents of every user's hard drive to every other user constantly, utilizing 99% of available upstream and downstream bandwidth. Suppose this game goes viral and becomes the next Youtube.
How should ISPs deal with it? Should they be required by law to allow it?
I think it's reasonable for ISPs to create a definition of residential bandwidth that prevents against unexpected bandwidth utilization trends. I want unfettered bandwidth, but I realize that there is additional cost associated with provisioning it.
They allow or disallow it? The question itself doesn't make sense. If the users want to use capacity they bought and Comcast sold, is the user's right. Comcast didn't foresee the demand is error of their actuaries. A insurance company pays up even if their actuaries make an error. Comcast doesn't get to externalize their costs, they could try, but we as customers have no obligation to honor it. I would claim that video demand has actually made the equation very simple for them with higher build out but lower risk of it not being utilized. The demand is very predictable and not "unexpected". And, monthly quotas (1TB per month) on top of bandwidth limits (150 Mbps) the risk is quite reduced. If someone with 250 Mbps ran at 99% capacity, 1TB quota would be over in less than 11 hours.
It's not as if Comcast sends you a check every time there is a network congestion and you don't get top speed that you are paying for. There is no consequence of congestion for Comcast but they are using it as a reason to get ransom from content owners. They are setting up the conversation such that they have less and less reason to remove congestion, more the congestion, more the end user suffer, more the case for google and netflix to open up their wallet.
$70 a month is same as cable bill and for cable Comcast has to pay channel operators. On internet, Comcast doesn't even have to pay to youtube or netflix. I mean internet service is pure profit compared to cable which runs on same infrastructure. They charge to take traffic at exchanges, one 100 byte URL in browser results in 2GB of data transferred from Netflix. It is a request that Comcast customer is making, trying to use the bandwidth and quota that the customer purchased, but Comcast makes money at exchange because they take on that 2GB of traffic. So they are already compensated for taking this traffic on both ends, by customer and at the exchange. Now on top of that they want to pry open the communication channels and charge more to the two parties talking. That's triple Comcast tax.
Comcast runs their cable programming on their tcp/ip network as well, so they are provisioned enough for now. If net neutrality goes away what promises youtube/netflix has that Comcast won't prioritize their xfinity platform over netflix?
The customers are using the capacity that they wanted and bought, Comcast sold that capacity, they must deliver. They are having no problem delivering so far. Networking gear (merchant silicone switches and OpenCompute switches) are dirt cheap, gone are the days of buying expensive gear from Cisco and BigIP. Cheaper networking hardware is making it easy to operate and upgrade for Comcast. Same goes for cable boxes, DVRs. Seen the disk prices lately? A raspberry pi can be effective cable box. Their unit costs have gone down significantly. They are benefiting from Cloud platforms like everyone else. Netflix uses AWS, Amazon uses AWS, Comcast uses AWS.
Look at hosting bandwidth costs, look at CDN bandwidth costs, content providers are paying for their side of it fairly. Comcast has a large share of CDN and can get even bigger there.
I think content providers should pick a month and ban entire Comcast network for a month together. Users will switch to shitty stuff that phone companies are selling but at least Comcast will learn a lesson. What do you think would people chose to watch xfinity or would they switch to telco ISP?
ISPs compete on the basis of their ability to provide residential bandwidth at competitive prices:
Suppose one ISP offers 1 Mbps up and down for $99 per month and another ISP offers 10Mbps down and 0.25 Mbps up. Both of these plans may cost the same to provide. Similarly, a 0.5 Mbps dedicated bandwidth circuit may cost 2x or 3x the cost of the residential circuit to provide.
Individual ISPs determine what markup to apply and how to price the bandwidth (with or without contracts, etc.). You are arguing that all circuits should be dedicated bandwidth circuits. There is nothing wrong with that argument, but it reduces the value of every piece of copper or fiber in the system because it prevents it from being maximally utilized.
Also, keep in mind that if a law were passed requiring all circuits to be dedicated bandwidth circuits, ISPs would have no incentive to preemptively upgrade hardware. Notice that customers do not demand symmetric dedicated bandwidth circuits. For most residential customers, only a minimum amount of upstream bandwidth is needed (for ACKs and the occasional upload) because most of the data is flowing to the customer and being watched on a screen.
T-Mobile's "Binge-on" service is a brilliant way of applying QoS to consumer demand to extract more value from the network, yet it was criticized as non-neutral. It's amazing the level of misunderstanding that is attached to net neutrality!
When did I say I am asking for dedicated circuit?
I actually granted that congestion could and would happen. I pointed out what happens if someone decides to saturate the line... would last only few hours each month before quota being hit.
What you are granting Comcast doesn't make sense technically, financially, legally or ethically. Don't open my packets other than headers for destination. Don't look inside the data of my packets. Don't offer intermediate certificates posing as proxy for end site. All those things are fine within a corporation's privately owned network. Comcast is similar to post office, they are not allowed to see inside my mail to see if it goes to bank or to insurance company, just the address on the envelope can be used. It is none of Comcast's business whom I talk to and what I talk about and how much I talk. As long as I am under quota and under bandwidth limit, they must carry the traffic.
In the catastrophic event they can stop traffic to certain IP address segments to keep voice lines and emergency sites accessible. They can work that out with regulators like for 911 service. But they can't use that as a way to extract payments from content providers. Complete stopping is preferred over degraded service. Comcast would have to come up with justification why they stopped the traffic, degraded is just a play to get ransom payment.
A regular HTTP GET is meant to be cacheable and meant not to be considered private. If you want private, use SSL and do not encode information in URLs that can be abused by intermediaries.
I realize you consider the attempt to charge upstream providers a fee to ensure high quality throughput extortion, but the fact is, some services utilize 99% of bandwidth.
Take the typical residential ISP, 99% of the upstream bandwidth it purchases likely moves traffic from youtube, youporn, etc. In the remaining 1% are gmail, HN, VOIP, etc.
The natural move would be for the ISP to adjust QoS to favor minimal streaming quality that will prevent customers from defecting to another ISP.
But content providers compete on the basis of content variety, quality, etc., and so their incentive is to maximize quality. If you are 1% more likely to click on an ad after watching porn at 4K, then it's potentially worth it to youporn to serve only 4K quality video since bandwidth is cheaper at scale and highly predictable.
But an ISP is operating at smaller scale. For last mile circuits, bandwidth cost is bound by hardware investments, and so a disproportionate amount of the cost of Youporn's added 1% chance you'll click is borne by the ISP.
So how can the ISP align its own incentives with youporn? Simple, by threatening a QoS that will reduce network load in exchange for a fee that helps share the cost of the additional bits.
So it's not extortion any more than Uber surge pricing is extortion. It helps increase the efficiency of the market process going on by aligning incentives where previously they were less aligned. Strong market incentive exists for ISPs not to overplay their hand, and to date as far as I know this hasn't been a problem.
I agree that last mile monopolies are bad, but the solution isn't to entrench them with top-down regs that hide the problem.
http://www.nytimes.com/2003/03/17/business/betting-speed-pur...
Comcast made a big bet in 2003 that many other companies did not make. Demand for broadband has been strong, but it was far from a given at the time Comcast made the investment that it would result in strong returns well into 2017.
It's easy to look back with full knowledge of how things played out and determine that Comcast was unsophisticated in its financial and business planning.
Comcast has grown because it has provided a "neutral enough" internet for its customers. There is absolutely no reason to believe it would begin to act differently now. If anything, Comcast's strong profit margins increase the chances that other companies will enter into competition for the business.
I'd rather have Google invest money and brainpower into creating things like Webm than to pay lawyers and PR specialists to promote net neutrality.
The sorts of packet inspection techniques you describe can be thwarted by encryption, and are no different than what is being done within Google to monetize user behavior/preferences to the nth degree.
You argue that Comcast wants to charge content providers for preferred access. I'm sure they'd like to, but this is the case for financial, not technical reasons:
The following is the most important point: Once a service (like Youtube) becomes popular enough, the traffic pattern of packets going to that service begin to resemble the traffic pattern of the internet itself, making arbitrage impossible. So the economics that once worked for ISPs to arbitrage based on upstream vs downstream bandwidth no longer work because upstream circuits are largely priced based on broad data access patterns which are heavily influenced by major content providers.
So the profit margin on residential bandwidth starts to approach $0 as 100% of the content consumption matches the exact characteristics of the upstream provider's pricing model, since upstream providers also do arbitrage, but with much less aggressive hedges and less leverage.
Consider it this way, you can start a competitor to the post office that is cheaper for some small list of addresses. Maybe you could profitably deliver all first class mail for 25 cents within a single zip code. But if you try to offer a similar service to all zip codes, you will likely face nearly identical constraints to the post office and you will not be able to undercut prices.
For ISPs, the more predictable customer demand gets, the more likely the arbitrage has already been done upstream, and the closer the residential circuit becomes to a dedicated circuit.
These days, suppose 90% of internet users follow a sine wave data usage pattern day after day that is divided between Netflix, Youtube, and a handful of other hosts. Video is massively more bandwidth intensive than other stuff so video content providers dominate. This means a lot less profit from bandwidth price arbitrage, which is why ISPs are toying with the idea of charging content providers for access (it's also a bit silly because the moment youtube goes down people would call in saying their cable modem was broken).
I don't think it's a bad thing for Google to help pay for the ISP's upstream bandwidth costs, since whatever contracts the ISP buys are largely serving the purpose of piping content from Google to consumers. It may or may not be extortion for Comcast to try to do this, but that should be up to regulators to decide after the fact. If this happened, by the way, it would likely accelerate Google's slow-as-molasses last mile effort and result in much more competition for residential ISPs.
ISPs provide a very helpful non-google intermediary for the small bit of our internet behavior that is not being captured and stored by Google. I think it is better to have a variety of firms, each with its own core specialty, fighting over this market than for regulators to simply follow Google's advice and hamstring ISPs into lower profit margins.
In an ideal world all internet connections would be unconstrained by QoS considerations. But if you think back to dated last mile tech like T1 lines, the financial arrangement is leasing the piece of copper wire and paying for a 24/7 circuit that can saturate that copper's bandwidth in both directions. For this reason, a T1 cost 10x the price of a residential connection supporting the same average bandwidth. If you are running a data center, this contract is important, you can't afford to deal with congestion and reduced bandwidth that is beyond your control. But for residential customers it is a desirable and worthwhile tradeoff that does not benefit from regulation whatsoever.