I have run a small business myself and I can tell you this is 100% true. There are folks online like Leo Laporte of Twit.tv who obviously has no idea about how networking operates because I routinely have heard him say the bits are free. That’s absolutely false and I have talked about it as well in an earlier post. I am not advocating for Comcast and Verizon but they are the biggest accused offenders. There are very real costs to running any network.
Karl Denninger of the Market Ticker blog gives an in depth breakdown of the math fallacies of the Net Neutrality screamfest that has once again started. I provide the article here in full. The reason i often provide KD’s articles in full is that they expire after a couple of weeks and disappear. These works are not mine and I do not claim that they are.
Math-challenged people******me off, and Net Neutrality is one of the bigger ones — so here we go.
Let’s assume I’m an ISP.
We’ll use nice round numbers to make this easy.
Let us assume I have 1Gbps of transport available to me on my network. I sell service with “speeds” of 10 Mbps and put connections through a “traffic shaper” that delivers “up to” 10Mbps for each customer.
I sell 500 of these connections in your neighborhood. I do this because I know, with a good degree of certainty (because I modeled it over the period of several months or years) that your average use as a home user will be under 2Mbps all the time, with occasional higher bursts.
Since 2Mbps x 500 = 1Gbps, I can support this userbase. If you run a “Speed Test” you will usually get the full 10Mbps that you bought. In rare circumstances you may not. If I have my traffic shaper implement a “fairness algorithm” I can prevent anyone from being “starved” entirely — but it is simply impossible for me to deliver 5Gbps (that is, 10Mbps to every one of my 500 customers at once) as that’s 500% of what my network is capable of doing!
All is well for quite a while.
Then along comes some new and innovative service. The “new and innovative service” charges $10/month (my Internet service to you costs $50/month, so I am collecting a total of $25,000 a month in revenue.)
But, that “new and innovative service” requires that you pull down 5Mbps for the entire time you are using it, and it requires that there be no jitter at all to work (in other words the 5Mbps has to be delivered from the time you start using it until you finish without exceptions, or your user experience will be unacceptable.) In addition the rest of your household use will still be there, so that 5Mbps requirement is additive to the 2Mbps I already modeled on an average basis.
Now let us assume this “new and innovative service” becomes wildly popular and half the people on my network subscribe to it.
Suddenly my 2Mbps model is no longer any good. It is now, for 50% of my customers during the 6:00 PM to 11:00 PM hours, 7Mbps.
My former network build-out required that I be able to deliver 1Gbps reliably.
Suddenly I need to deliver (250 * 7) + (250 * 2) or 2Gbps — twice as much — or everyone screams and calls me a schmuck, swear that I run a terrible ISP and more.
The facts are — and I am speaking as a former ISP CEO and guy who has built networks for a living for roughly 30 years — that attempting to “over commit” a network by 100%, that is, demand twice what it is capable of delivering, doesn’t cause everyone to get half of what they want. Due to how TCP works and the retries that are generated when buffers overflow everyone (not just the people who want to watch streaming) will get very close to nothing at all. Some modern operating systems will attempt to “throttle back” their demanded bandwidth in an attempt to maintain operation but not all, and consumer devices such as cellphones, tablets, desktop and laptop computers, especially older ones, are some of the worst in this regard.
Let’s assume (for simplicity) the following breakdown of my expenses monthly (simplified but good enough to make the point):
- $10,000 is spent on bandwidth provision (directly proportional to that 1Gbps)
- $10,000 is not proportional to the bandwidth provision (building, staff, power, routers for the most part, etc.)
- $2,500 goes to promotion and marketing (attracting new customers, advertising, etc)
- $2,500 is my profit (10% of sales — not really all that good, but about right for a mature business.)
Your “demand” for that “new an innovative service” just doubled that first $10,000 line; it goes to $20,000. In order to prevent you from destroying my network’s performance for everyone I must spend the additional $10,000 yet “net neutrality” says I cannot charge those who caused this expense more money nor can I “shape” or block what amounts to an economic and network terrorist.
I am now losing $7,500 a month. I have been forced to spend the $10 large by an outside firm I have no contract with or control over because if I don’t my network has unacceptable performance for everyone. That outside firm solicited people to buy their service knowing that this would happen because they believe they can force me to EAT that additional $10,000 in cost.
Worse, the faster I grow my customer base or the more people adopt this “new service” the more money I lose because my loss is a percentage, not a dollar amount!
I thus have only the following options available to me if I wish to remain in business:
1. Charge the “new and innovative service” for the performance it demands from my network that is beyond what was reasonably engineered for. In other words they get charged a “tariff” to the extent they force network operation beyond engineered limits, and if they refuse shape their traffic so it conforms to both what my network was designed for and what nearly all other services on the Internet fit within. This is something said “new and innovative” service might be able to mitigate. For example Netflix could be “unlimited” to the customer only if you queued what you wanted to watch the night before, allow it to transfer the data to your computer or phone when everyone is sleeping on a rate-limited basis and thus there’s no load impact on me as an ISP. If you instead demand to watch now, and “now” is in the evening hours, you pay a buck an hour to Netflix (and Netflix pays the ISP that, less the handling costs) for your decision to impose the load at that particular time. Note that if I charge back the $10,000 then Netflix is forced to raise its prices to $50 from $10 since the additional $40 in hard cost they tried to shove off on me per-customer gets thrown back at them. How many customers does Netflix have at $50/month? NONE!
2. Charge the user directly for the “burst” traffic on a metered basis. In other words you have a 10Mbps link but if you consume a lot of data during “busy” periods you will get hit with a “demand” charge. This is how the electric utilities work for commercial customers; you have a base charge and then a “demand” charge that applies to your heaviest power demand during periods of heavy use. That charge is large because it is intended to recover the expense of being able to meet your extraordinary demand for electricity. The market has deemed #2 unacceptable, period; note that the government is able to force this billing paradigm in the commercial power delivery market (and in a few areas in the residential market too) because electrical service is a government-granted monopoly.
3. Charge everyone irrespective of their use of said new service — or not. In other words I now have to be able to deliver the 2Gbps as an ISP but I can’t charge on a differential basis for it based on who’s making me purchase the additional capacity so I am now forced to charge everyone 25% more whether they use the new and innovative service or not. In this case you pay for your neighbor’s decision to subscribe to Netflix. As a cable company I might get that from you in higher internet prices or higher cable TV prices but I have to get another $20/month from you somehow. If you’re wondering why $200 cable+internet monthly bills are now rather common and it’s damn hard to get both under $100 a month even with basic cable, maybe you will finally realize that you screwed yourself with all your insane screaming because this is why it happened.
Net Neutrality effectively forced the ISP to do #3 — Hastings got his $200 stock price because you have been robbed whether you are a Netflix subscriber or not.
Oh by the way, it’s not just Netflix. It’s also Amazon (with their Prime Video), Hulu (their subscription service), Youtube “Red”, Disney’s newly-announced service coming online soon, MLB’s “streaming” service and hundreds of new services yet to be developed and marketed.
Since nobody subscribes to all of these services yet all of them (plus the ones to come that are bandwidth hogs outside of the expected norm) will wind up in this position the odds are that even if you are a Netflix subscriber you’re going to get robbed in order to subsidize someone else’s subscription to something.
There are serious monopoly problems with “last mile” data provision, especially in America. They were present in the 1990s (in places where we had multiple DS-1 providers in the 1990s, for example, the price was usually 1/2 to 1/5th of that where there was only one “choice”!) and are worse today for broadband where for most consumers there is exactly one option of like kind and quality. But Net Neutrality does not address that problem because it can’t; it instead imposes a forced-subsidy model on those who don’t want a given service and makes the monopoly problem worse.
If you want an example in another “market” have a look at Health Care, where EMTALA (mandating “neutrality” of treatment of all emergency patients) is arguably one of the biggest causes of the last three decade’s worth of 500+% increases in the “cost” of health care and “insurance”, never mind the complete inability in many cases to get a price at all prior to services being rendered. Tell me again about how “affordable” said “insurance” actually is if you don’t qualify for some sort of subsidy….