Electrical systems actually must generate exactly the required amount of electricity at all times. One particularly interesting tidbit I learned in one of my engineering classes: hydro dams are like giant capacitors. If you can't "push" more energy into the power grid, then you also can't push more water through the turbines; it gets harder to turn the turbines due to electromagnetism. And so the dam starts to fill up, more or less automatically, because of the laws of physics.
In dynamo-generated power it's harder to turn the tubines when the electricity is "used," that is, when there is current flowing. The more current that is flowing the harder they are to turn.
Turning the wires of the coil in the magnetic field sets up an emf. If you allow that emf to drive a current, you now have current-carrying wires moving through a magnetic field feeling a force opposite their motion. This is felt as resistance by whatever is driving the turbine. So, when you "use" the electricity the turbines are harder to turn, not easier.
This also makes sense in terms of conservation of energy. You are taking the energy out of the water and using it to power the users connected to the electric grid. The more power you want to supply, the more you must take out of the water (or steam, if you are using steam to drive the turbines).
So the inability to push more water through the turbine is due to the turbine delivering max current (that is, it is at full load). Any less load means less current and less resistance.
In fact I understand (not an expert on their construction) that it is a problem when the load reduces at some types of plants because you are still driving the turbines with the same fuel but their resistance drops. You have to adjust the fuel to the turbines to keep everything balanced.
With Nukes this may be a slow processess, so that is perhaps why you hear of plants "dumping" energy-- using it in some (wasted) way just to keep the current in the coils high enough to provide resistance to whatever is driving the turbines.
I've heard of places that pump water into elevated storage so as to use it again when demand suddenly picks up- kind of a capacitor effect- but I don't know anything about it first-hand.
Thanks for reintroducing logic to my brain. The rest of my argument mostly holds up though, right?
Re-orient your thinking a little and it will make sense. We're talking AC systems (not a SWER) so it is a closed-loop as far as the electricity is concerned. The electricity doesn't go anywhere, it constantly oscillates. What you are actually generating in a meta sense is ability to do work. You are overcoming the electrical resistance from all those users' machines doing work.
Does that analogy help?
So.. electricity gets pumped into the grid from some power station, and then [a given unit of electricity] loops repeatedly around the loop until it's either consumed by an appliance or is converted to heat by the resistance inherent to the transmission lines themselves?
Also, at Bell you can buy 40GB extra for $5 if you pay in advance, which equals $0.125/GB.
http://www.bell.ca/shopping/popups/personal/internet/usage_e...
Most quotes I've heard recently put the cost of bandwidth at around 1-2c/GB. I assumed, quite possibly erroneously and I'll go hunting for the source in a moment to confirm/refute said assumption, that when TekSavvy's CEO said that a conservative estimate (read: "highest reasonable estimated price") of bandwidth could be 30c/GB, I was assuming that he was taking things like administrative costs and possibly infrastructure costs into consideration. I made this assumption because of the extreme disparity between the two cost points - 1c/GB from some versus 30c/GB from him.
I realize that I didn't make this clear in my original post - I didn't even hint at it, actually - so I apologize for that.
Also, when I just now checked my wording in my original post, I noticed that you edited your comment to include a back-of-the-napkin estimate of Bell's incremental cost of delivery to be around 13c/GB. In my opinion, this only strengthens my assumption, since I sincerely doubt Bell would provide additional bandwidth for less than cost.
Looking at Bell's website right now, I see that their "Performance" plan in Ontario lists 6mbps speed with a 25GB cap for $31.95 per month.
Using your calculation of cost of bandwidth being 13c/GB, this means that full utilization of the cap would cost them $3.25 per month, leaving $28.70 for administrative costs and upgrading.
In a completely unfair comparison, I'll mention that Rogers charges me $6.95 per month (last I checked, anyway) as a usage fee/basic fee/system access fee. Let's nearly double that up to $12.72 (to make the next calculation easier) and call that the cost of doing business (tech support, paper pushing, general administration). This brings the total up to $15.97 a month, which is HALF of what Bell charges for this plan.
Bell then gets to choose how much of their 100% profit they want to use to upgrade their network, and how much they want to use to give their CEO a raise.
As far as the overage is concerned, let's note that when I called bandwidth costs 13c, that was on the assumption that Bell made no profit from their "insurance" option. Anyone who doesn't pay insurance gets to pay $2/GB, of which at least $1.87 is pure profit. Their overage would be 1538% of cost - a nice markup.
Edit: I just noticed[1] that the true cost of this plan is actually $41.95, but they offer $5 off for the first twelve months and an additional $5/mo discount to be used towards their satellite service. I'm not going to bother doing more calculations, but I thought it was worth noting.
[1] http://www.bell.ca/shopping/jsp/pageblock_styles/Pricing/fle...
(I had to disable JavaScript to get that link. The JS-Enabled version requires you to drill down two links to get the full details to appear, so it was impossible to get a direct link to the JS-Enabled version of these details)
At the wholesale level, bandwidth is sold as a committed throughput rate. Essentially like an "open pipe" of various diameters (to use a semi-bad analogy).
Wholesale bandwidth is currently around $15/Mbit/mo. This means that for $15 you can get a 1Mbs connection to a wholesale Internet provider for 1 month (not really, because they generally want you to buy at least a 10Mbs handoff). If you leave this connection completely idle, you pay $15/mo. If you saturate it 24/7, you pay $15/mo.
Because of the bursty nature of Internet traffic, ISP's will typically resell multiples of their committed bandwidth. You might have a 100Mbps handoff, but could sell 500 or 88Mbps worth of traffic to your customers.
Bandwidth really doesn't get metered down to the MB or GB until you get to the consumer level. This is mostly done to "shape" demand somewhat, and allow for a billing method to charge users who are heavy users (eg: if your customers all tried to actually use that 800Mbps you were selling them and you really only had 100Mbps of available bandwidth, you're going to have to do something about that...). For higher-level datacenter type customers, it's typically a 95th percentile billing method, not a direct $/GB overage charge.
You can do some math and take your operating costs and divide by your bandwidth and figure out your cost/GB, but that's not going to be highly accurate. In many cases you can go from say a 40Mbps connection to an 80Mbps connection with all the exact same equipment, and just pay the upstream for the extra 40Mbps/mo. This would mean the same infrastructure cost is divided over a larger amount of bandwidth, resulting in lower cost/GB (until you need to add another card to your edge router, or add another router altogether...).
I think that part of the problem people are having with coming up with fair and realistic pricing for bandwidth is that unlike gas or water or electricity, you're not really delivering a physical commodity with costs that vary directly by quantity used.
I've always known that there was something inherently wrong with charging by traffic, but I could never put my finger on what it was. A clear explanation, such as you've just provided to me, on exactly how the internet works and is billed at higher levels makes things obvious.
Traffic is a problem, but only insofar as it contributes to saturation of the overall available bandwidth of the existing infrastructure. Calling the problem "traffic" is just a convenient scapegoat for other concerns that can't be directly billed as easily, even if traffic isn't the real problem.
Thanks. I'm always impressed by the quality of discourse on Hacker News.
This is how every single element of the economy works.
Plus, if the ISPs really want usage based pricing, then they should do that instead of this arbitrary download cap stuff that is nothing about usage based pricing.
Yes, 25GB is an absurdly low cap. Most providers have a basic plan that provides 60GB, with a usually quite inexpensive way to dramatically increase the cap. People keep quoting the $2.50/GB bit which is just shockingly ignorant -- if you know that you'll use more (which not everyone will. The casual user will not and Avg Facebook and Email User has no reason to subsidize the "everything through my internet connection" guy), it's more like $0.12/GB.
This whole discussion is a perfect example of how people can embrace positions wholeheartedly just because they selfishly want them to be true.
And even if people tried to stream Netflix 24 hours a day, it makes more sense to throttle the connection then to just toss around ridiculous overage charges as if the problem is the amount of data being downloaded in a month is the problem rather than the amount of bandwidth being consumed at peak times.
Sure, people can increase their cap by buying the "insurance", but with the low level caps, that basically becomes the true price of the connection for anyone who isn't using their internet for trivial uses (from a bandwidth) perspective. It's there in the hopes that you worry about going over your cap and being screwed in charges rather than as some benevolent gesture on the part of the ISPs. If they had any decency, they would just automatically upgrade your connection to the higher level if you go over the initial cap. Why is that such an unreasonable action to take?
I'm not pretending anything. Quit with the trollish tactics. The service adapts to the market.
Sure, people can increase their cap by buying the "insurance"
You realize that terminology was coined by Teksavvy right? That company has done a brilliant job playing the public like a fiddle, and you're playing just the note they want to hear.
I'm with Cogeco. I have a 60GB cap. I've gone over it once. For $6 more I have a 125GB cap and a higher throughput, so $0.09 per GB. For a few dollars more I would have a 150GB cap. And so on. The same is true at Bell, Rogers, etc.
It's there in the hopes that you worry about going over your cap
ABSOLUTELY! That is, without a doubt or question, exactly why they have caps. They don't want you to come anywhere close to your cap. They would rather Joe Average uses 5GB / month of their 60GB cap. On my cell phone I have a 5GB cap, and in an average month I use about 200MB (always around WiFi).
However every provider warns you as you approach your cap. One whiny complainer was posting a screenshot of how he went over his cap and there was big letters on it explaining exactly what happened and how, asking him to upgrade now.
They do exactly what you ask for. Turns out that some people are just irresponsible.
Indeed. There were a few lines which particularly stood out to me as having been written by someone who doesn't really get the situation:
> To encourage competition, major telecom operators that have spent heavily on infrastructure are required to lease bandwidth on their networks to small providers.
I get that the system is maintained by these large companies, and I'm sure they have spent a lot on the infrastructure, but (and correct me if I'm wrong) I believe the system was built largely with taxpayer dollars, and it continues to be heavily subsidized by same.
> Although critics say the CRTC ruling will lead to lower download limits and higher rates, major Internet service providers say usage-based billing based is fair because it means heavy users pay more than those who just surf the web and use email.
There are two problems with this. First, internet isn't like electricity. Electricity must be generated at a cost, and those who consume more cost the generating company more (perhaps not directly, since unused electricity is simply lost, but at least indirectly by requiring higher generation to accommodate peak usage). With the internet, however, the lines are already laid and it's merely a matter of sending the information down the wire. There is almost no additional cost associated with higher usage (see next point).
Second, the 'incremental cost" of bandwidth appears, from what I've read recently, to be somewhere around a penny per gigabyte, if any exists at all. Let's give the companies the benefit of the doubt and say that it's actually 2.5 cents per GB. Now let's give them a 100% markup to be nice. So now we're calling a "reasonable markup" on this service to have a price of 5 cents per GB. Bell wanted $2 per GB for folks who go over their limit. They wanted 40 times what I just gave as a reasonable number. Even if you want to use the most conservative estimate I found, which was the CEO of TekSavvy who said maybe it could get as high as 30 cents per GB, then Bell would still be asking us to pay 666% (<obvious joke here>) of the actual cost. That's one hell of a markup.
Also, and again using the 30c/GB estimate, Bell's offer in Ontario of a 25GB limit would equate to a $7.50 cost? Anyone want to take a guess at how much the service actually is actually priced at? I'll give you a hint: it's not $7.50.
Sorry for the rant, but I've been worried about this decision and I have a lot of pent up anger about it. It's nice that the government has actually stepped up to support the small, reasonable ISPs and the consumer.